Wenshan Yang Machine Learning based Wind Power Forecasting for Operational Decision Support Vaasa 2022 School of Technology and Innovations Industrial Management Licentiate of Technology 2 UNIVERSITY OF VAASA School of Technology and Innovations Author: Wenshan Yang Title of the thesis: Machine Learning based Wind Power Forecasting for Operational Decision Support Degree: Licentiate of Technology Discipline: Industrial Management Supervisor: Iivari Kunttu, Heidi Kuusniemi Year: 2022 Pages: 195 ABSTRACT : To utilize renewable energy efficiently to meet the needs of mankind's living demands becomes an extremely hot topic since global warming is the most serious global environmental problem that human beings are facing today. Burning of fossil fuels, such as coal and oil directly for gen- erating electricity leads to environment pollution and exacerbates global warning. However, large-scale development of hydropower increases greenhouse gas emissions and greenhouse effects. This research is related to knowledge of wind power forecasting (WPF) and machine learning (ML). This research is built around one central research question: How to improve the accuracy of WPF by using AI methods? A pilot conceptual system combining meteorological information and operations management has been formulated. The main contribution is visualized in a proposed new framework, named Meteorological Information Service Decision Support System, consisting of a meteorological information module, wind power prediction module and opera- tions management module. This conceptual framework has been verified by quantitative anal- ysis in empirical cases. This system utilizes meteorological information for decision-making based on condition-based maintenance in operations and management for the purpose of optimizing energy management. It aims to analyze and predict the variation of wind power for the next day or the following week to develop scheduling planning services for WPEs based on predicting wind speed for every six hours, which is short-term wind speed prediction, through training, validating, and testing dataset. Accurate prediction of wind speed is crucial for weather forecasting service and WPF. This study presents a carefully designed wind speed prediction model which combines fully- connected neural network (FCNN), long short-term memory (LSTM) algorithm with eXtreme Gradient Boosting (XGBoost) technique, to predict wind speed. The performance of each mod- el is tested by using reanalysis data from European Center for Medium-Range Weather Fore- casts (ECMWF) for Meteorological observatory located in Vaasa in Finland. The results show that XGBoost algorithm has similar improved prediction performance as LSTM algorithm, in terms of RMSE, MAE and R2 compared to the commonly used traditional FCNN model. On the other hand, the XGBoost algorithm has a significant advantage on training time while compar- ing to the other algorithms in this case study. Additionally, this sensitivity analysis indicates great potential of the optimized deep learning (DL) method, which is a subset of machine learning (ML), in improving local weather forecast on the coding platform of Python. The results indicate that, by using Meteorological Information Service Decision Support System, it is possible to support effective decision-making and create timely actions within the WPEs. Findings from this research contribute to WPF in WPEs. The main contribution of this research is achieving decision optimization on a decision support system by using ML. It was concluded that the proposed system is very promising for potential applications in wind (power) energy management. KEYWORDS: Decision-making, Deep learning, Energy management, Machine learning, Oper- ations management, Strategic management, Wind power forecasting, XGBoost. 3 Acknowledgement Firstly, I would like to thank Dr. Iivari Kunttu at Häme University of Applied Sciences as my first supervisor. He is very supportable and patient all the time. Thanks for Prof. Heidi Kuusniemi at University of Vaasa being as my second supervisor. She gave guidance from key perspectives. Dr. Ari Venäläinen from Ilmatieteen Laitos (Finnish Meteorological Institute) guided me. He is a rigorous and amiable expert, and he provided me with the guidance from the domain of meteorology. Prof. Yang Liu from the Department of Management and Engineering in Linköping Uni- versity gave me support and lit up a lamp for me when I got confused. I would like to express my sincere gratitude to Prof. Donald Huisingh at University of Tennessee and for the invaluable comments during the writing phases. The dean of Technology and Innovation Management Unit in University of Vaasa, Dr. Raine Hermans, led and contributed to my graduation process much. Mrs. Ulla Laakko- nen is an experienced administration staff member, and I got a lot of help from her for the past few years. Prof. Marja Naaranoja always has constructive opinions and she made me more confident. Thanks to Mrs. Juuli Honko for doing lots of paperwork and job related. Tobias Glocker made important comments and supported me most of the time. I extend this warm Thank You to all researchers and staff at University of Vaasa and University of Twente who made my learning experience in Finland and The Nether- lands memorable. Besides these, my colleagues are always very supportive and en- couraging. Finally, I would like to thank my parents, my husband, and my little lovely princess. My parents are the teachers of my life and they taught me how to become the best version of myself. Although my husband devotes all to his work and dream, then I need to de- 4 vote more to my family, job, and study. I also got too much, for example, sharing every exciting moment with my husband, joining every second of happiness with my small sweetheart, enjoying the warm and sweet family atmosphere. Thanks for their uncon- ditional support, forever love and companion to me all through the way, especially dur- ing the most difficult time in doing this research. Meanwhile, the most important thing is that I am on the way to get the doctoral de- gree at the same time. My dearest father would be extremely proud of this. The ending of this study may be a new start for the research journey. Expecting nothing and appreciating everything is my life attitude. Doing is everything while saying is nothing. I believe life is preparing a gift for everyone! Kiitos, tack, bedankt, danke, thank you, 谢谢(謝謝) , 감사합니다, ขอบคุณ! Linköping and Vaasa, May 2022 Wenshan Yang 5 Contents 1 Introduction 12 1.1 Background 12 1.2 Wind Power around the World 13 1.3 Meteorological Source 15 1.4 Decision Support 17 1.5 Research Questions 20 1.6 Opportunities and Challenges 21 1.7 Structure of the study 22 2 Literature Review 24 2.1 Focused Literatures 24 2.2 Description of Material Reviewed 25 2.2.1 Distribution across time-period 25 2.2.2 Distribution of research methodologies 26 2.2.3 Classification of research topics 26 2.3 Circumstances of Mainly Research 27 2.3.1 Assessment for wind resources 27 2.3.2 Wind power forecasting 29 2.3.3 Commonly used wind power forecasting algorithms 30 2.3.4 State-of-art of machine learning 35 2.3.5 Wind-solar complementary 36 2.3.6 Energy storage system 37 2.3.7 Wind turbine control system 39 2.3.8 Errors and risk management 40 2.4 Research Gap 41 3 Methodology 45 3.1 Research Strategy 45 3.1.1 Research philosophy 45 3.1.2 Research approach 46 3.1.3 Research strategy 47 6 3.2 Research Methodology 48 3.2.1 Main research methodology 48 3.2.2 Programming platform 49 3.3 Method Design 50 3.3.1 Artificial intelligence, machine learning and deep learning 50 3.3.2 Fully-connected neural network, long short time memory, XGBoost 52 3.3.3 Research design 54 3.3.4 Data collection 58 4 Result and Findings 59 4.1 Results of Algorithm 59 4.2 Findings 66 4.2.1 Findings of algorithm 66 4.2.2 Findings of managerial aspect 67 5 Summary of Publications 68 5.1 Overview of Papers 68 5.2 Logical Connection of Papers 69 5.3 Summary of Individual Papers 70 5.3.1 A review of innovation in wind power forecast 70 5.3.2 The impact of climate change on wind power enterprises 71 5.3.3 Meteorological information service decision support system in wind park application 72 5.3.4 A Study on renewable energy potential based on the global atmospheric data 73 5.3.5 Research in this study 75 6 Discussion and Conclusions 77 6.1 Contribution 77 6.2 Managerial Implications 78 6.3 Research Limitations 79 6.4 Future Research 81 7 References 84 Appendices 105 Appendix 1. Paper 1: A Review of Innovation in Wind Power Forecast 105 Appendix 2. Paper 2: The impact of climate change on wind power enterprises 129 Appendix 3. Paper 3: Meteorological information service support system in wind park application 150 Appendix 4. Paper 4: A study on renewable energy potential based on the global atmospheric data 171 8 Figures Figure 1. Wind power global capacity and annual additions, 2010-2020. adopted from REN21 (2021) 14 Figure 2. Wind power capacity, top 10 countries, 2020. adopted from REN21 (2021) 14 Figure 3. Research method for research questions. 21 Figure 4. Distribution of publications per year across the period studied. 25 Figure 5. Mind map of related research topics for this article based upon variation of research field. 27 Figure 6. Literature search results. adopted from screenshot from SCOPUS (Elsevier) database 43 Figure 7. Approach to theory development. 46 Figure 8. To build a dataset for machine learning. 47 Figure 9. Structure chart of artificial intelligence. 50 Figure 10. Machine learning and human thinking mode. 51 Figure 11. Deep learning performance. adopted from Anurag Bhardwaj & Wei Di, 2018 52 Figure 12. Fully-connected neural network. 53 Figure 13. Long short-term memory network. 54 Figure 14. System structure and process. 56 Figure 15. Tree structure of technologies related to Artificial Intelligence. 59 Figure 16. Wind speed prediction using fully-connected neural network. 60 Figure 17. Prediction performance of fully-connected network. 60 Figure 18. Wind speed prediction using LSTMs. 61 Figure 19. Prediction performance of LSTMs. 61 Figure 20. Wind speed prediction using XGBoost Regression. 62 Figure 21. Prediction performance of XGBoost Regression. 62 Figure 22. Average daily wind speed during 2015-2019. 65 Figure 23. Monthly wind speed in each month in 2019. 66 Figure 24. Logical connections among the papers. 70 9 Tables Table 1. Summary of searching methodologies used for this research. 42 Table 2. Typical failures related to weather conditions. 56 Table 3. Performance for each algorithm. 62 Table 4. Training time and inference time for each model. 63 Table 5. List of articles´ titles with key words. 68 Table 6. Overview of the articles’ aims, methods and the main results/contribution. 68 Abbreviations AI ANN CBM CDR DL EWEs KF LSTM MAE ML NWP RERs RMSE SDE SVM TBM WERs WPD WPEs Artificial intelligence Artificial neutral network Condition-based maintenance Climate data record Deep learning Extreme weather events Kalman filters Long short-term memory Mean absolute error Machine learning Numerical weather prediction Renewable energy resources Root mean square error Standard deviation of error Support vector machine Time-based maintenance Wind energy resources Wind power density Wind power enterprises 10 WPF WPP XGBoost Wind power forecasting Wind power potential Extreme gradient boost 11 Appendix 1. Yang, W., Liu, Y., and Yang, G. (2014): A review of innovation in wind power forecast. Proceedings of the 11th International conference on Innovation & Management. Appendix 2. Yang, W., Liu, Y., and Yang, G. (2014): The impact of climate change on wind power enterprises. Proceedings of the 11th International conference on Innova- tion & Management. Appendix 3. Liu, Y. and Yang, W. (2015): Meteorological information service support system in wind park application. International journal of production economics, no. 2, pp. 222-237. Appendix 4. Yang, W. and Liu, Y. (2015): A study on renewable energy potential based on the global atmospheric data. Global Cleaner Production & Sustainable Con- sumption Conference. 12 1 Introduction 1.1 Background With the deterioration of the global environment and the exhaustion of fossil energy, the mineral energy resources consumed in the future will gradually be replaced by re- newable energy resources. The development and utilization of ecological energy is very important for environmental protection and has become a global issue. Renewable energy does not pollute the environment at the point of energy generation, and gener- ally has a much lower pollution footprint than traditional energy from installing to de- commissioning and can diversify the power generation technology (He et al., 2021). Increasing population growth requires more sustainable development of energy. Wind energy stands out when compared with other energy because it is free, clean, inexhaustible, has the capacity to generate greater power, and has lower energy costs. Hence, wind power plays an important role as a source of energy supply (Adeyeye, Ijumba, & Colton, 2020; Bórawski, Bełdycka-Bórawska, Jankowski, Dubis, & Dunn, 2020). Wind energy resource is becoming a leader in the current energy transition pro- cess as the most significant characteristics of wind energy are, clean, ecological, and inexhaustible (Gil-García, García-Cascales, Fernández-Guillamón, & Molina-García, 2019; Saleh Asheghabadi, Sahafnia, Bahadori, & Bakhshayeshi, 2019). Excessive consumption of traditional fossil energy, hydrocarbon fuel for energy produc- tion has led to a severe global air pollution and climate change. However, wind energy is widely considered to be a qualified renewable as it can mitigate climate change im- pacts and achieve low-carbon transformation (Cui, Liu, Ali, Gao, & Chen, 2020; Saeed, Ahmed, & Zhang, 2020; N. Shen, Wang, Peng, & Hou, 2020). 13 1.2 Wind Power around the World According to Renewables 2021 Global Status Report from REN21, the estimated share of renewables in global electricity generation was more than 29% by the end of 2020. Figure 1 shows wind power global capacity and annual additions during 2020-2020. China and the United States accounted for a bit more than three quarters of the global electricity production rise in 2020. Wind power capacity and additions of top 10 coun- tries in 2020 can be seen in Figure 2. Demand of renewable energy resources (RERs) is growing as the global population grows continuously and on the other hand fulfilling the climate change mitigation aims agreed in UNFCC COP 21 Paris 2015 require that an even larger and larger share of energy production be based on renewable energy. Ac- cording to Renewables 2016 Global Status Report, developed and developing countries have had increased investment in solar power by 12% and wind power by 4% while biomass and waste to energy, ocean, biofuels, small hydro, geothermal power reduced respectively by 42%, 42%, 35%, 29%, 23% in 2015. Global Wind Energy Council claimed that global installed wind power capacity has increased by 63,467 MW in 2015, repre- senting annual market growth of 22%. Although world electricity generation produced by wind power is still low, it is growing rapidly. Wind power capacity is 743 megawatts and ranked secondly among renewable power capacity while the hydropower capacity is 1,170 megawatts by the end of 2020. As well, U. S. Energy Information Administra- tion data show that particularly some European countries had the largest portion of their electricity generation from wind: Denmark (48%), Portugal (25%), Spain (22%), Ireland (38%), Germany (27%). So far, the most important wind gross electricity pro- ducers in the EU are Germany and Spain. The highest increase of wind cumulative in- stalled capacity in 2022 will be in Croatia (Bórawski et al., 2020). 14 Figure 1. Wind power global capacity and annual additions, 2010-2020. adopted from REN21, 2021 Figure 2. Wind power capacity, top 10 countries, 2020. adopted from REN21, 2021 Changes in temperature, precipitation, sea level, and the frequency and severity of extreme events will likely affect how much wind power generation is produced, deliv- ered, and consumed. For example, various weather phenomena, such as rainstorm, hail, thunderstorm, and tornado, can generate damage more or less to wind turbines. 281,0 122,5 62,6 38,6 27,4 17,7 17,6 9,3 6,8 4,0 Wind power capacity, top 10 countries, 2020 China United states Germany India Spain Brazil France Turkey Netherlands Norway 15 Despite the challenges the production of wind power is the foreseen trend (Kandpal & Broman, 2014). The wind power construction In Finland began later than in many other European countries. However, from 2012 to 2013, wind power construction has gained momen- tum and national construction and production statistics have been broken year after year. In Finland, the wind turbine capacity in offshore parks will increase to be over 7MW instead of the largest turbines currently 5MW in the future. According to Finnish wind power statistics 2021 from Suomen Tuulivoimayhdistys, at the end of 2021, there were 962 installed wind turbine generators, with a combined capacity of 3257 MW. They generated 11,7% of Finland’s electricity consumption in 2021. 141 new wind farms were built in Finland in 2021, but annual wind power pro- duction increased by 26 % comparing with previous year. They generated 9,3 % of Fin- land’s electricity consumption in 2021. Wind power production for the whole year was 8,061 TWh, or 11,7 % of all electricity production in Finland. Wind power covered 9,3 % of Finland’s electricity consumption and the amount is expected to increase to 25 % by 2025 at the latest. Suomen Tuulivoimayhdistys also pointed out that Finland has the potential to increase wind power capacity considerably. The objective of the wind power industry is to achieve at least 30 TWh of annual wind power production in Finland in 2030, which corresponds to approximately 30% of Finland's electricity consumption at that time. This means there is great potential in Finland’s wind power development. 1.3 Meteorological Source Wind is a type of meteorological phenomena and wind energy is one kind of natural resource which is obtained from the wind. It is one of the oldest-exploited energy sources by humans and today is the most seasoned and efficient energy of all renewa- ble energies. Wind energy results from horizontal air pressure differences, which 16 means air movement, have regional differences, and are affected by surrounding ter- rain. Wind power generation is the most efficient technology to produce energy in a safe and environmentally sustainable manner. It is a process of converting the energy produced by the movement of wind turbine blades driven by the wind, namely as ki- netic energy of the air, into electric energy (Emeis, 2018). Wind generated when the pressure gradient force, the Coriolis force and friction are combined. The greater pressure gradient force, the greater the wind force. The higher the latitude, the greater the Coriolis force. The rougher the underlying surface, the greater the friction and the smaller the wind force. Among them, the Coriolis force influences wind direction, friction influences wind speed while the pressure gradient force influences both wind speed and wind direction. There are many factors that affect wind speed, such as, topography, meteorological factors. Topography includes geomorphology, surface obstacles, and so on. For mete- orological factors, temperature, humidity, pressure, etc. have a greater impact on the changes in wind speed, making the daily variation regularity of wind speed not good, and the prediction accuracy is not high. Wind farms, namely wind parks, where planted groups of wind turbines, are located on open land, on mountain ridges, or offshore in lakes or the ocean. Wind farms can be either onshore or offshore. According to the 8th edition of WMO Guide to Meteorological Instructions and Meth- ods of Observation from World Meteorology Organization (WMO) in 2014, the meas- urement of wind speed should be taken from a site that is well exposed to the wind, and not in the lee of obstructions such as buildings, trees, and hillocks. If it is possible, the measurement site should be a good distance from obstructions, namely at least 10 times of the obstructions’ height and upwind or sideways by at least twice of the ob- struction’s height. Direction should be estimated from a vane (or banner) mounted on 17 a pole that has pointers indicating the principal points of the compass. The vane is ob- served from below, and wind direction may be estimated to the nearest of the 16 points of the compass. There are advantages and disadvantages of developing wind power generation based on its own natural characteristics of wind energy. a. Environment friendly: Wind energy is a source of renewable energy. Wind turbines do not release emis- sions that can pollute the air or water. Wind turbines may also reduce the amount of electricity generation from fossil fuels (Tong, Cheng, & Tong, 2021). b. Inexhaustible: Wind power provides energy from air movement and has the capacity to generate greater power. This process will continue as long as there is weather on planet Earth, meaning that energy can be gained from air movement forever (Oñederra, Asensio, Saldaña, Martín, & Zamora, 2020). c. Unstable: As the wind power is proportional to the cubic wind speed, even small errors in estimation of wind speed can have large effects on the energy. d. Unpredictable Renewable energy sources, such as wind energy, solar energy, are innately unpre- dictable, owing to the uncertain nature of themselves and bring about more challenges in the distribution networks (Rezaeian-Marjani, Masoumzadehasl, Galvani, & Talavat, 2020). Even wind energy is variable but intermittent, but not completely random and unpredictable. 1.4 Decision Support To ensure the proper operation of renewable energy-based hybrid systems, and ensure demand and increase system performance, energy management techniques and a de- cision support element is needed for efficient management of energy. The strategic management process should be turned into a management tool with a decision sup- 18 port element in terms of sustainability. A robust energy management strategy allows the system to meet demand, to increase the lifetime of the components, increasing operating costs and, to ensure maximum use of renewable sources, to reduce energy costs output, to protect components from overload damage and enhance the reliability of the power system as a result, to optimize system performance (Ammari, Belatrache, Touhami, & Makhloufi, 2021; Çetin & Ziya Sogut, 2021). To establish an assessment model, find crucial solutions, support industrial decision- makers highlighting specific actions, some models with proposed energy management strategy were designed, the energy management strategy was optimized, proved effec- tive for intelligent energy systems (X. Huang, Zhang, & Zhang, 2021; Trianni, Cagno, Bertolotti, Thollander, & Andersson, 2019). The tool may serve as a point of reference for energy and environmental decision support aids in communities where important cultural resources, values, and traditions are potentially impacted by energy manage- ment decisions (Necefer, Wong-Parodi, Small, & Begay-Campbell, 2018). A powerful reliability management tool, to deal with the risk assessment, is indispen- sable in decision-making. Common requirements for an effective decision support plat- form include credibility, relevance, legitimacy, model accessibility, end-user satisfaction, timeliness, and costs for maintenance and computing. Among these, accurate identifi- cation of the risks and timely quality management of the risks play an important role in improving the quality, safety, and reducing loss costs. Strategic decision-making on long-term drought risk management can be supported by integrated assessment mod- els to explore uncertain future conditions and potential policy actions (Hamilton et al., 2019; P. Liu & Li, 2021; Mens, Minnema, Overmars, & van den Hurk, 2021). Deep learning (DL) algorithms train data longer than ML algorithms since training big data, so it is meaningless of decision-making in some cases of requiring results in a lim- ited time. Results are only valuable if they can be obtained within a specified time- period. By reviewing articles which related to relevant topics, it is found that lots of 19 articles related to decision support and ML are linked to the domain of clinics and med- icine, such as, surgical decisions, triage for patients, emerging decision support, and diagnostic Decision Support in Radiology. Compared to those topics, fewer articles are relevant to the topic of WPF. The main objective of this research is to support grid dispatchers and decision-makers in electricity transition towards climate friendly economies by giving them suggestions and options in planning and designing low carbon solutions. To do so, Meteorological Information Service Decision Support System integrates artificial intelligence (AI) algo- rithms with meteorological information decision support platform while it develops optimal operational planning via predicting wind speed to optimize energy manage- ment decision-making. It is an AI decision support system that considers the uncertain- ty of wind power output and proves that the machine learning (ML)-based system can optimize the prediction results while comparing with some other traditional algorithms for wind power enterprises (WPEs). This decision support system consists of a meteorological information module, wind power prediction module and operations management module. It utilizes meteorologi- cal information for decision-making based on condition-based maintenance in opera- tions and management for the purpose of optimizing energy management. This re- search attempts to make full use of distributed new energy and rationalize the energy management strategy of grid dispatching companies. In this research, decision maker refers to the level of person who is involved in the operational decision-making process, focuses on more strategic decisions, and makes the final decision organizationally to adopt the practice. The expected result is that, by using Meteorological Information Service Decision Sup- port System, it is possible to support effective decision-making and create timely ac- tions within the WPEs. Besides this, findings from this research contribute to WPF in 20 WPEs. The main contribution of this research is to achieve decision optimization on a decision support system by using ML algorithms. 1.5 Research Questions The main research objective of this research is to improve wind power prediction through increasing wind speed accuracy by using AI algorithm as the key research method. To achieve its objectives, decision support platform of Meteorological Infor- mation Service Decision Support System answers the following questions. Judging from the background and research objectives, the central research question (RQ) is as follows. Central RQ: How to improve the accuracy of wind power forecasting by using artificial intelligence methods? The five sub-questions can be formulated: Sub-question 1. What is the innovation in the development process of WPF among so much relevant research? Sub-question 2. How climate change influences WPEs and what factors affect wind power output? Sub-question 3. Can there be a general framework to help forecasting wind speed and wind power more effectively in decision-making? Sub-question 4. How to use the global atmospheric reanalysis data to analyze the po- tential of WERs in Finland? Sub-question 5. What is the sufficient ML algorithm to improve the accuracy of wind speed prediction? The five sub-questions are depicted as above and developed from the main research objective. The sub-question 1, 2, 3 and 4 are responded respectively by paper 1, 2, 3 and 4 while sub-question 5 is answered in chapter 3. These four related papers are at- tached in the appendix. The structure of the research method used is presented in fig- ure 3. 21 Figure 3. Research method for research questions. 1.6 Opportunities and Challenges RERs are one of the solutions to solve the challenges related to energy production and mitigation of climate change. It is possible to replace fossil fuels by developing a variety of renewable energy, including hydro, wind, solar, wave and biomass, geothermal and ocean. Making strategies for converting present energy systems into a 100% sustaina- ble renewable energy system is crucial (Dincer, 2000; Lund, 2007). As an example, since the early 1980s, Denmark has been one of the leading countries in the world in the field of wind energy utilization based on the management of development and diffu- sion of sustainable technologies (Christensen & Lund, 1998). One research predicted that global energy demand in 2040 will be approximately 30% higher than it was in 2010. Because the typical characteristics of wind energy are stochastic and intermittent, it is important to know and use appropriate renewable energy technologies in the whole process of producing wind power generation (Dashwood, 2012). Another re- 22 search in China showed developing green energy business in emerging economies, with the aim being long term sustainability of a healthy level of overall flexibility of the wind power industry chain, pay special attention to competition flexibility, technology flexibility, and intellectual property and talent flexibility (Z.-Y. Zhao, Zhu, & Zuo, 2014). In one word, the future development of wind power presents a significant opportunity in terms of providing low carbon energy. As it known to everyone, challenges always go with opportunities. Wind power is fluc- tuating, intermittent as wind has the characteristics of volatility, intermittent, and low energy density. These features do not have a significant impact on the grid when wind power accounts for a small proportion of the grid. However, as wind power develops rapidly in this decade, wind power production will face serious problems which are electrical system safety, operations stability. Meanwhile, there exist also environmental challenges and technical challenges. 1.7 Structure of the Study This study is published as articles based. The structure of this study is divided into six chapters as follows. Chapter one presents an introduction and background of this research. After describing the opportunities and challenges, depicting the function of strategic decision support in risk management, it also displays the central research question, five sub-questions, opportunities, and challenges. Chapter two firstly reviews, makes a statistic and analysis on the relevant articles, and presents circumstances of mainstream research towards relevant topics. Then it de- scribes the research gap of this research. Chapter three presents the research methodology. It includes research philosophy, re- search approach, research strategy, research methods, research design and data collec- tion. Research process is described also. 23 Chapter four depicts the results and findings of the case study. Chapter five provides the summary of the publications. This section interprets the logic connections and depicts the main content of each article. Chapter six makes conclusions for this study, provides contribution, managerial implica- tions, and research limitations, gives some final remarks, and proposes for the future research. The Appendix consists of four original articles (paper 1-4) and author’s role in the whole research. 24 2 Literature Review This chapter firstly makes a statistic of related articles, and then classifies them into different categories by time-period, research methodologies and research topics. Chap- ter 2 also summarizes and reviews state-of-art articles on wind energy resources (WERs), wind power prediction, commonly used wind power forecast (WPF) algorithms, energy storage system, wind turbine control system, errors, and risk management. Research gap is depicted after reviewing literatures. 2.1 Focused Literatures There exist large number of issues in aspects of wind energy that must be examined. For example, studies may focus on installed capacity, mathematical models, optimiza- tion of energy output, facility maintenance, or excessive energy storage. On the other hand, there also exist many literature reviews investigated in the same areas. For in- stance, review on forecasting wind speed, wind power density (WPD) and generated power, review wind energy resources (WERs) in the urban environment, review on wind power short-time prediction, specific wind power forecasting (WPF) models, local energy plans and policies. Besides reviewing papers, this study also reviews the state of the art of wind energy conversion systems and technologies, wind energy status in a specific year, global renewable electricity scenario, wind speed probability distributions in application, etc. This research gives a comprehensive review on the WERs, WPF, whole developing process, innovative technologies, and the related areas in Chapter 2. Based on more than 500 selected articles, the main objectives of this review work can be formulated as follows: (i) a summary of the previous studies, (ii) a construction framework of related research topics, and (iii) the identification of the future research. These articles were mainly chosen from SCOPUS (Elsevier), ScienceDirect (Elsevier), Web of Science (ISI Web of Knowledge) and Google Scholar, classified by Mendeley, and analyzed by Microsoft Excel statistical function. The earlier publications may not be 25 displayed on the Internet, and this may have a small influence on the reviewed litera- ture work. 2.2 Description of Material Reviewed 2.2.1 Distribution across time-period After searching by relevant keywords, the number of reviewed papers is 506. The tem- poral variation of reviewed publications during the period 1976-2021 is shown in Fig- ure 4. These are incomplete statistics since the papers produced in 2020 continue to be published. Thus, high numbers of publications are found for the time-period between 2011 and 2021. In general, the total number of papers increases steadily year by year during 2014-2020. The review of publications was based on articles from SCOPUS (Elsevier), ScienceDirect (Elsevier), Web of Science (ISI Web of Knowledge), Google Scholar, national and international renewable energy reports. Figure 4. Distribution of publications per year across the period studied. 66 14 16 23 39 41 42 45 48 52 63 57 0 10 20 30 40 50 60 70 Number of papers (N=506) 26 2.2.2 Distribution of research methodologies Five research methodologies were differentiated in this research: (1) theoretical and conceptual papers; (2) empirical papers and case studies; (3) surveys and review pa- pers; and (4) books. It shows the classification of publications according to the research methodologies. Among them, the number of empirical papers was the largest group is 421 while the number of theoretical papers is 33 and the number of review papers is 42, and the number of books is 10. 2.2.3 Classification of research topics While reviewing the research publications, it ended up with classifying the publications into seven categories. The seven main topics include: (1) Wind resource assessment; (2) Wind speed prediction, numerical weather forecast, climate changes; (3) Wind power prediction methods; (4) Wind-solar complementary; (5) Wind energy storage; (6) Wind turbine control and service; and (7) Wind power forecast errors and risk management. Figure 6 shows the framework and classification of the framework of the publications, which based on the inductive analysis approach, revealed that there is a cluster of in- novative technologies pertaining to wind power generation. 27 Figure 5. Mind map of related research topics for this article based upon variation of research field. adopted from Liu & Yang, 2015 (Paper 3) 2.3 Circumstances of Mainly Research 2.3.1 Assessment for wind resources Previous research has been done on assessing the potential WERs all around the world. Various geographic characteristics also cause a wide variety of temperature and climate differences. Among the publications identified in the related searching, some evaluated the installed wind capacity, some interpreted wind power assessment metrics, while some reviewed the global renewable resources. 28 The capacity of installed wind turbines is increasing in many wind farms all around the world. Through statistical methods, Staid and Guikema (Staid & Guikema, 2013) had investigated the factors that influence the installed wind capacity in each state of U. S. are the physical and geographic characteristics of the state. Flora, Marques, & Fuinhas (2014) studied the wind idle capacity during the year of 1998-2011 among 18 European countries to help policymakers when adjusting energy policy. In China, there existed a large discrepancy between installed capacity and wind power generation even with the dramatic increasing installed wind capacity (M. Yang, Patiño-Echeverri, & Yang, 2012). Turkey had an extremely low installed wind power capacity 0.22% of the total economy power capacity (Güler, 2009). Chadee and Clarke (Chadee & Clarke, 2014) assessed regional wind resources through comparing statistical wind power density (WPD). They used reanalysis wind data for the period 1979–2010. The results show that although the prevailing winds are from the east-north-east over the eastern Caribbean islands, their wind direction distribu- tions are bimodal. Moreover, other papers (Carta & Mentado, 2007; Hennessey, 1977; Lu, Yang, & Burnett, 2002; Sedefian, 1980; Shamshirband et al., 2016) also estimated the WPD distribution function in different districts. J. Zhang, Chowdhury, & Messac (2014) proposed to use a more comprehensive metric named Wind Power Potential (WPP). Compared to WPD, WPP is more credible because it not only accounts for wind speed information but also considers the joint distribution of wind speed and direction. The results from four sites of North Dakota, that WPD and WPP follow different trends, and that the realistic resource potential measure was not captured by WPD. Additionally, Ucar & Balo, 2010; W. Zhou, Yang, & Fang (2006) meas- ured the wind energy resource via WPP. 29 2.3.2 Wind power forecasting In practice, ultra-short-term, short-term, medium-term, and long-term time scales are used to predict wind power in WPEs (Soman, Zareipour, Malik, & Mandal, 2010). There are different timescales when classifying WPF species according to time periods and one example is as follows (Colak, Sagiroglu, & Yesilbudak, 2012; De Giorgi, Ficarella, & Tarantino, 2011). a. Long-term forecasting It predicts from one day to one week ahead and aims to optimize maintenance and repair of wind turbines. It is usually used for planning and designing wind farms. b. Medium-term forecasting It predicts from six hours to one day ahead and aims to optimize power system management and energy trading. It is usually used for dispatching the electricity grid rationally. c. Short-term forecasting It predicts from thirty minutes to six hours ahead and aims to optimize pre-load sharing. It is usually used for repairing and debugging. d. Ultra-short-term forecasting It predicts from a few seconds to thirty minutes ahead and aims to optimize turbine control and load tracking. It is usually used for controlling wind turbines and stabilizing electrical energy. The methods of wind power prediction are mainly divided into the following three groups (González-Mingueza & Muñoz-Gutiérrez, 2014; Jung & Broadwater, 2014; Lei, Shiyan, Chuanwen, Hongling, & Yan, 2009): • Physical forecasting approach: In contrast to statistic approach, the physical approach is based on the use of physical considerations. It needs detailed physical descriptions of the wind farm and their surroundings, including description of the wind farm (wind farm layout and wind turbine power curve, etc.) and description of the terrain (orography, roughness, obsta- 30 cles, etc.). This approach aims to get the optimized predicting wind speed and direction in different hub height of wind turbine generator system. • Statistic forecasting approach: The statistical approach is based on mathematical statistics analysis of the main variables associated with the relationship between energy generation and meteorolog- ical information. The meteorological data, obtained from historical data or output of Numerical Weather Forecast (NWP), mostly used as input. For example, wind speed, wind direction, temperature, and atmospheric pressure in the wind farm. • Combination approach: The hybrid method is a useful predicting way as it can improve the WPF accura- cy by offsetting random error with one method from each other. In some models, a combined approach is used to integrate advantages of both approaches. 2.3.3 Commonly used wind power forecasting algorithms Based on different input data, which means whether to use NWF, the WPF can be di- vided into NWF forecasting method and historical meteorological data forecasting method. From 1977 until now, many articles described different algorithms of WPF. Some of the representative models are reviewed in this section. a. Kalman filters Kalman filters (KF) is an optimal recursive data processing algorithm, and it has been firstly achieved by Stanley Schmidt in 1958. Some papers regarded the KF model as an algorithm which applied to wind speed numerical prediction to improve predic- tion accuracy. Cassola and Burlando (Cassola & Burlando, 2012) reported that meteoro- logical models are usually unable to provide reliable surface wind speed forecasts due to the shortcomings in horizontal resolution, physical parameterizations, initial and boundary conditions. Thus, they used KF wind speed data to forecast the wind energy output, and the percentage error between simulated and measured wind energy val- ues was still very low and showed a stable evolution. (Louka et al., 2008) developed a wind speed forecasts method, which includes two limited-area atmospheric models 31 based on KF, efficiently eliminating systematic errors, even in the lower resolution cas- es, and contributing further to the significant reduction of the required CPU time. To improve the performance of KF models, (Poncela, Poncela, & Perán, 2013) substituted the traditional way of setting the values of the model parameters by estimating them by quasi maximum likelihood methods for a certain forecast horizon. They showed that the improved model is close to an optimum for all the horizons and provides more ac- curate predictions, with up to 60% of improvement for the RMSE. b. Time series model Commonly used time series models include auto regressive (AR), moving aver- age (MA), auto regressive moving-average model (ARMA) and auto regressive integrat- ed moving average (ARIMA). Among these, one ARIMA model established by Box and Jenkins (Box & Jenkins, 1976) have been widely used for the purpose of time series forecasting. Meanwhile, this book is extremely comprehensive because it interprets each kind of time series model in detail and gives examples. Huang and Chalabi (Z. Huang & Chalabi, 1995) used AR model to forecast wind speed from one hour to a few hours ahead because it takes into account the non-stationary nature of wind speed. Based on multidimensional ARMA series, Soder (Kavasseri & Seetharaman, 2009) provided a method that can simulate possible outcomes of wind speeds based on available forecasts. This method was es- tablished based on the assumption that wind speed forecasts are available in several regions and that the forecast errors in different regions are correlated. Kavasseri and Seetharaman (Kavasseri & Seetharaman, 2009) forecasted wind speeds on the day- ahead (24 h) and two-days-ahead (48 h) by using a fractional-ARIMA model. The results showed that significant improvements in forecasting accuracy are obtained with the proposed models compared to the persistence method. (Kamal & Jafri, 1997) found the ARMA model is suitable for predicting intervals and probability forecasts. c. Artificial neural network 32 Various Artificial neural network (ANN) models are widely used, such as back propagation (BP) and radial basis function (RBF) neural networks. The ANN is an infor- mation-processing method, which works like a human brain processes to find an algo- rithmic solution to select the structure from the existing data (Kavasseri & Seetha- raman, 2009; Olaofe, 2014). Based on the original BP network, one new wind power prediction model which optimized the tabu search algorithm with memory function was developed by Han, Li, & Liu (2011). Guo, Wu, Lu, & Wang (2011) proposed a new hybrid wind speed forecasting method based on a BP neural network and the idea of eliminating seasonal effects from actual wind speed datasets using seasonal exponen- tial adjustment then get lower mean absolute errors. By investigating the use of weather ensemble predictions in the application of ANNs, Taylor and Buizza (Taylor & Buizza, 2002) found that the average of the load scenarios is a more accurate load fore- cast than that produced using traditional weather forecasts. Alexiadis, Dokopoulos, & Sahsamanoglou (1999) developed an ANN algorithm that significantly improved fore- casting accuracy compared to the persistence forecasting model. Salcedo-Sanz (Salcedo-Sanz, Ortiz-García, Portilla-Figueras, Prieto, & Paredes, 2009) presented the hybridization of the fifth-generation mesoscale model (MM5) with ANN to address a problem of short-term wind speed prediction. The adopted strategies were individual ANN and hybrid strategy based on the physical and the statistical methods. Peng, Liu, & Yang (2013) comprehensively compared the performance of two prediction methods and his calculated results showed that the individual ANN prediction method can quickly produce the prediction results. d. Support vector machine (SVM) Support Vector Machine (SVM) was firstly developed by Corinna Cortes and Vapnik in 1995. The most apparent difference between SVM and ANN is that the for- mer focuses on mathematical methods and optimization mechanisms even though they are similar. (Mohandes, Halawani, Rehman, & Hussain, 2004) introduced SVM, the latest neural network algorithm, to wind speed prediction. The result indicated that SVM compared 33 to multilayer perceptron (MLP) neural networks is closer to the actual wind speed. Ortiz-García et al. (Ortiz-García et al., 2011) proposed an improvement to an existing wind speed prediction system, using banks of regression Support Vector Machines (SVMr) to manage the diversity in input data arising from the use of different global forecasting models and several parameterizations of a mesoscale model. They showed that the system implementing SVMr banks outperforms the basic system without tak- ing diversity into account in the input data. (Q. Hu, Zhang, Xie, Mi, & Wan, 2014) devel- oped a technique of the uniform model of ν-support vector regression for the general noise model (N-SVR). The existing studies on using SVM for wind speed prediction are very limited as these studies usually only use one particular kernel function and a spe- cific combination of parameters. J. Zhou, Shi, & Li (2011) applied Least-squares Support Vector Machines (LS-SVM) with linear, Gaussian, and polynomial kernels to perform short-term wind speed forecasting. e. Fuzzy logic Initially, one approach draws definite conclusions from vague, ambiguous, or imprecise information, however it is not widely used because of the low accuracy as low ability of fuzzy logic prediction is low when studying (Klir & Folger, 1988). Metter- nicht developed a useful and practical technique for modelling complex phenomena that may not yet be fully understood owing to its ability to deal with imprecise, uncer- tain data, or ambiguous relationships among data sets (Metternicht, 2001). Fuzzy logic is a new and logical approach, which when applied in the field of engineering, a fuzzy logic model is useful for predicting wind speed, electrical power, or rotor's speed. There are few up-to-date literature reviews about this algorithm for predicting wind power. Therefore, this is a promising research gap. f. Fuzzy logic control The Fuzzy logic controller (FLC) approach, which tracks the generator speed with the wind velocity to extract the maximum power, takes into account the grid de- mands and power generation predictions, are used for efficiency optimization and per- 34 formance enhancement control in wind generation systems (Sarrias-Mena, Fernández- Ramírez, García-Vázquez, & Jurado, 2014; Simoes, Bose, & Spiegel, 1997a, 1997b). Mo- hamed, Eskander, & Ghali (2001) designed the tracking controller of the wind energy conversion system based upon fuzzy logic control (FLC) technique and this system has been tested to have robustness and effectiveness by simulation. Kamel, Chaouachi, & Nagasaka (2010) proposed and developed a new fuzzy logic pitch controller and an energy storage ultra-capacitor to smooth the output power of wind turbines and en- hance Micro-Grid (MG)'s performance in islanding mode, and the results proved that the proposed strategies are effective. g. Hybrid algorithm A combined, nonlinear hybrid KF-ANN model was found to be a better way of forecasting wind speed than KF and ANN separately, to solve the inaccuracy wind pow- er forecasting of linear ARIMA (Shukur & Lee, 2015). A hybrid wind speed forecasting model consists of ARIMA model and ANN model, predicted the wind velocities with a higher accuracy than the ARIMA and ANN model separately (Cadenas & Rivera, 2010). Two hybrid methods namely ARIMA-ANN model and ARIMA-Kalman model, which were based on single time series model, ANN model and KF model, had good forecast- ing accuracy and were suitable for the jumping wind samplings. One hybrid model named PMERNN and PAERNN, combine SVM with seasonal index adjustment (SIA) and Elman recurrent neural network (ERNN) methods, forecasted the daily wind velocities with a higher degree of accuracy over the prediction horizon (J. Wang, Qin, Zhou, & Jiang, 2015). Two hybrid models, namely, ARIMA–ANN and ARIMA–SVM, were selected to compare with the single ARIMA, ANN, and SVM forecasting models, showed that the hybrid methodology does not always outperform the individual forecasting models based on ARIMA, ANN, or SVM (Shi, Guo, & Zheng, 2012). 35 A novel hybrid modelling method named SVR–UKF was proposed for integrating un- scented Kalman filter (UKF) with SVR to precisely update the short-term estimation of wind speed sequence. With this method, the prediction errors were closer to zero with significantly smaller variations, whereas the prediction errors of the other methods were widely scattered (K. Chen & Yu, 2014). According to Guo et al. (2011), there is no single best forecasting algorithm that can be applied to any wind farm since wind speed patterns can be very different between wind farms and are usually influenced by many factors that are location-specific and difficult to control. Each of the physical models, statistical models, spatial correlation models and artificial intelligence models has advantages and disadvantages. For exam- ple, the time series model is one kind of statistical model, and it is popular because its computation is simple. But ANN and KF are more widely used for their good nonlinear performance. 2.3.4 State-of-art of machine learning DL has higher recognition accuracy on large sample data sets Comparing with tradi- tional machine learning (ML), such as SVM, convolutional neural network (CNN) has a better solution and effect on recognizing objects (P. Wang, Fan, & Wang, 2021). One new research proposed by Jiang et al develops a short-term wind speed forecast- ing method which combines statistical method, ANN, and DL. This system consists of four parts: optimal sub-model selection, point prediction based on a modified multi- objective optimization algorithm, interval forecasting based on distribution fitting, and forecasting system evaluation (Jiang, Liu, Niu, & Zhang, 2021). In a photovoltaic power generation system, Shin et al. demonstrate that RNN and LSTM are more suitable for the time series data structures compared with dynamic neural networks to achieve the best prediction results (D. Shin, Ha, Kim, & Kim, 2021). 36 Accurate wind speed forecasting. This study proposes a hybrid model named VMD-DE- ESN incorporating variational mode decomposition (VMD) and differential evolution (DE) and echo state network (ESN) for wind speed forecasting. This hybrid model was validated, mean absolute percentage errors (MAE) are 2.0161%, 3.4153%, 2.1544%, and 2.8478% respectively, which are much lower than several others (H. Hu, Wang, & Tao, 2021). By combining AI methods with statistical knowledge, Zhang et al. proposes a new in- terval prediction model based on the Fast Correlation Based Filter (FCBF) algorithm, the optimized Radial Basis Function (RBF) model and Fourier distribution for wind speed. The results show that the maximum and average value of the prediction error are only 0.8430 m/s, 0.1749 m/s, which are significantly better than several others (Y. Zhang, Pan, Zhao, Li, & Wang, 2020). 2.3.5 Wind-solar complementary The site selection plays an important role in wind power farms, photovoltaic power farms, and wind-solar hybrid power stations. Matlab, Simulink Software are commonly used to evaluate the performance of hybrid systems (Akyuz, Oktay, & Dincer, 2012; Dihrab & Sopian, 2010). In March 1995, Kimura, Onai, & Ushiyama (1996) documented complementary rela- tionships between solar energy and wind energy in a small-scale wind-solar hybrid power system. Ma, Yang, Lu, & Peng, 2014; H. Yang, Lu, & Zhou (2007) utilized the model of Loss of Power Supply Probability (LPSP) to minimize the cost of energy and help reduce the size of energy storage based on a techno-economic evaluation. Q. Huang, Shi, Wang, Lu, & Cui (2015) proposed an approach which used multiple small wind turbines instead of one bigger one. The results showed that at low wind speed, the former one has more power production. 37 (Chávez-Ramírez et al., 2013) focused on the integration of photovoltaic (PV) system, micro-wind turbine (WT), Polymeric Exchange Membrane Fuel Cell (PEM-FC) stack and PEM water electrolyzer (PEM-WE), for a sustained power generation system. Bhattacharjee and Acharya (Bhattacharjee & Acharya, 2015) performed a small-scale application of wind-solar hybrid simulation model in an educational building to allevi- ate grid dependence. Maouedj’s (Maouedj, Mammeri, Draou, & Benyoucef, 2014) hy- brid system consists of PV and wind subsystems, battery energy storage, load and a hybrid system, controller for battery charging and discharging condition. The experi- mental results showed that the photovoltaic panel group constituted the primary en- ergy supplier of the system; while the wind turbine was the secondary supplier since the contribution of the wind turbine was small compared to the share of the photovol- taic subsystem. To evaluate system efficiency, Xydis (Xydis, 2013) identified the overall Exergetic Capac- ity Factor (ExCF) for a wind-solar power generation complementary system. ExCF is a new parameter which can be used for better classification and evaluation of RESs. One research (Y. Shin, Koo, Kim, Jung, & Kim, 2015) presented one PV-wind-battery-diesel power generation system which optimizes power generation by sparse matrices and the linear programming algorithm. 2.3.6 Energy storage system Zahedi (Ahmad Zahedi, 2014) interpreted several benefits of integrating intermittent sources of energy such as solar and wind with energy storage has several benefits for the electricity grid. (Wu et al., 2014) identified the distribution of probabilistic methods to determine the optimal size of the Energy storage system ESS for a wind farm in elec- tricity markets. Maleki and Askarzadeh (Maleki & Askarzadeh, 2014) used a discrete version of harmony search (HS) to optimize the size. The decision variables (number of PV panels, wind turbines, and batteries) are optimized by use of HS for having a cost- effective system. 38 • Battery storage In 2010, Khalid and Savkin (Khalid & Savkin, 2010) designed a controller which was based on model predictive control (MPC) to smooth the wind power output. The pro- posed controller is capable of smoothing wind power by utilizing inputs from our pre- diction system, optimizes the maximum ramp rate requirement and the state of the charge of the battery under practical constraints. Four years later, they proposed a new semi-distributed battery energy storage system (BESS) scheme to minimize the capacity of BESS to ensure the lower cost of the system (Khalid & Savkin, 2014). Ge et al. (2013) used a dynamic mathematical model of Vanadium redox flow battery (VRB) in an energy storage system (ESS) to provide a stable and smooth power flow injected into the grid though the wind power fluctuated. X. Y. Wang, Mahinda Vi- lathgamuwa, & Choi (2008) illustrated using the proposed design method, a BESS in a buffer scheme, to attenuate the effects of unsteady input power from wind farms. Jan- nati, Hosseinian, Vahidi, & Li (2016) reduced the cost of BESS by using Smart Parking Lots (SPLs). Two researchers explored the benefit of optimally integrating wind power with pumped hydro storage (PHS) because the daily wind speed patterns do not match the average daily load pattern. The results of the survey revealed that PHS, in conjunction with the wind farm, can reduce the system’s total power output shortage and increase the expected daily revenue (Gao et al., 2014; Murage & Anderson, 2014). Kaldellis, Kapsali, & Kavadias (2010) used an integrated computational algorithm for sizing of PHS systems that exploit the excess wind energy produced by local wind farms, the contribution to the electrification of the remote islands becomes evident. • Wind-compressed air energy storage Wind-compressed air energy storage (Wind-CAES) is an inexpensive way to store massive amounts of energy for long periods of time. Satkin, Noorollahi, Ab- 39 baspour, & Yousefi (2014) developed a site selection method for wind-CAES power plants to identify the wind energy potential for wind-CAES sites. The case study from Fertigand and Apt (Fertig & Apt, 2011) in the U.S. showed CAES brought social benefits including avoiding construction of new generation capacity, improving air quality dur- ing peak times, and increasing economic surplus. In Germany, a stochastic electricity market model has been applied to estimate the effects of significant wind power gen- eration on system operation and on economic value of investments in CAES. This case showed that CAES can be economically beneficial in the case of large-scale wind power deployment (Swider, 2007). • Flywheels Flywheel based energy storage systems (FESSs) are designed to smooth the net power flow injected to the grid by a variable speed wind turbine. In a wind diesel pow- er system (Sebastián & Peña-Alzola, 2015), the main components of FESS include elec- trical machine, flywheel, grid converter and electrical machine converter, improving the power quality of the isolated micro-grid. According to the intermittency of the wind, researchers integrated and validated the energy storage systems. For instance, Zhao et al. developed one hybrid energy storage system, which was based on adiabatic com- pressed air energy storage and flywheel energy storage system, to deal with the wind power fluctuations (P. Zhao, Dai, & Wang, 2014; P. Zhao, Wang, Wang, & Dai, 2015). More RESs will be integrated into the electricity grid worldwide in future. Taking the limited storage unit to find a more effective solution to handle the reliability and stabil- ity for the hybrid energy storage system is important. 2.3.7 Wind turbine control system Eriksson and Bernhoff (Eriksson, Bernhoff, & Leijon, 2008) compared three different wind turbines through a case study. The vertical axis wind turbine appears to be advan- tageous to the horizontal axis wind turbine in several aspects. Uddin and Kumar found out that life cycle assessment (LCA) study varied from location to location due to indus- 40 trial performance, countries energy mix and related issues (Uddin & Kumar, 2014). Demir and Taşkın thought that environmental impacts are low for the turbines with high hub heights due to increase in electricity production of those turbines (Demir & Taşkın, 2013). Novak, Ekelund, Jovik, & Schmidtbauer (1995) proposed a model to de- sign and evaluate the number of linear and nonlinear control schemes for wind-turbine speed regulation. Normally, wind turbines will reach the end of their service lives after 20-40 years. Or- tegon, Nies, & Sutherland (2013) considered the management of end-of-service (-) life of wind turbines (EOSLWTs) should also be considered by the wind power industry. According to ISO 14040 standard, which allows us to make an LCA study quantifying the overall impact of a wind turbine and each of its components, Martínez et al. (Martínez, Sanz, Pellegrini, Jiménez, & Blanco, 2009) analyzed the wind turbine during all the phases of its life cycle, from cradle to grave, with regard to the manufacture of its key components (through the incorporation of cut-off criteria), transport to the wind farm, subsequent installation, start-up, maintenance and final dismantling and stripping down into waste materials and their treatment. Schleisner developed a model to assess the life cycle of the production and manufacture of materials in a wind farm in Den- mark (Schleisner, 2000). Bonou, Skelton, & Olsen (2016) proposed an eco-design framework which was based on LCA to drive sustainable innovations in components, product systems, technologies, and business models. 2.3.8 Errors and risk management The stochastic electricity market is influenced not only by the uncertainty of nature's wind resources but also wind power forecast errors, as forecasting plays a crucial role in the renewable wind energy market. Holttinen outlined the forecast errors of wind power producers in the electricity market, pointing out shorter times between bids and delivery of production is to handle the forecast error (Holttinen, 2006). Two research (Díaz-González, Hau, Sumper, & Gomis-Bellmunt, 2015; Taraft, Rekioua, & Aouzellag, 2013) found out that the measurements of the power output and reduction of the en- 41 semble wind power forecast error depends on the size of the region. Pinson and Ka- riniotakis introduced a new methodology for assessing the prediction risk of short-term wind power forecasts. Their purpose was to find a linear relation between the Meteo- risk Index (MRI) and the resulting prediction errors (Pinson & Kariniotakis, 2004). Considering the wind power fluctuations under extreme weather conditions, Lin et al. proposed a model in the frequency domain to assess the wind power reduction under extremely high wind speed conditions. This model was validated and demonstrated to be valuable for both power system planning and operation with high wind penetration under extreme wind conditions (Lin, Sun, Cheng, & Gao, 2012). Hosseini-Firouz used the conditional value-at-risk methodology based on stochastic programming to opti- mally solve the wind power problem faced by the uncertainty issues, derived from wind availability, market prices, and balancing energy needs (Hosseini-Firouz, 2013). Soukissian and Papadopoulos used the Error-In-Variables approach to find the effects of alternative wind data sources on the wind climate analysis by examining the off- shore WPD (Soukissian & Papadopoulos, 2015). González-Aparicio and A. Zucker used the stochastic scenario extensions of dispatch models to take the value of flexibility into account to combine with the nature of fore- cast uncertainties. It applied clustering techniques to reduce the range of uncertainty, and regressive techniques to forecast the probability density functions of the intra-day price. (González-Aparicio & Zucker, 2015). 2.4 Research Gap Even though many published articles refer to topics which are classified as above, the trend is more and more publications are mainly about using AI in the science of WPF in WPEs. Many recent studies show that AI technology can improve the accuracy of wind speed prediction. Table 1 shows the searching methodologies used for this research. 42 Table 1. Summary of searching methodologies used for this study. Literature search strings Search field Numbers of documents results Limit to (“wind” and “speed” and “prediction”) Article title, Abstract, Keywords 8911 Search field Documents, (“wind” and “speed” and “prediction”) AND (“artificial” and “intelligence”) Article title, Abstract, Keywords 144 Documents, English (“wind” and “speed” and “prediction”) AND (“machine” and “learning”) English 215 Documents, English (“wind” and “speed” and “prediction”) AND (“artificial” and “intelligence”) AND (“ma- chine” and “learning”) Article title, Abstract, Keywords 39 Documents, English (“wind” and “speed” and “prediction”) AND (“artificial” and “intelligence”) AND (“ma- chine” and “learning”) AND (“deep” and “learning”) English 4 Documents, English (“wind” and “speed” and “prediction”) AND (“artificial” and “intelligence”) AND (“ma- chine” and “learning”) AND (“deep” and “learning”) AND (“xgboost”) Article title, Abstract, Keywords 39 Documents, (“wind” and “speed” and “prediction”) AND (“machine” and “learning”) AND (“xgboost”) English 0 Documents, English (“wind” and “speed” and “prediction”) AND (“machine” and “learning”) AND (“deep” and “learning”) AND (“xgboost”) Article title, Abstract, Keywords 3 Documents, English (“wind” and “speed” and “prediction”) English 2 Documents, English (“wind” and “speed” and “prediction”) AND (“artificial” and “intelligence”) Article title, Abstract, Keywords 0 Documents, English From the angle of artificial intelligence (AI) algorithm, there are 144 journal papers searched within article title, abstract, keywords with “artificial intelligence” among 8911 papers with “wind speed prediction” in SCOPUS (Elsevier) database. After explor- ing with narrow down, only 2 document results display. In other words, there only ex- isted two published articles which show in figure 6 are specifically related to wind speed prediction, machine learning (ML), deep learning (DL) and eXtreme Gradient Boosting (XGBoost) technique. 43 Figure 6. Literature search results. adopted from screenshot of SCOPUS (Elsevier) database The research uses Uncertainties in Ensembles of Regional Reanalysis (UERRA) database which contains a big amount of data but not big data. Using these reanalysis data can solve the problem of lacking real measured data. From the aspect of AI technology, noise refers to unreal data or wrong data and not all of them can be seen from human beings. The noise may come from sensor failures or inaccurate measurement. Noise is a problem for analyzing data by AI method since AI algorithms cannot recognize which are real useful data, and which are noise. In this circumstance, there is a need to do data cleaning to get a good result. However, manually cleaning data is a heavy work- load and it cannot be done by manpower as this massive data brings burden to work- ing stations, even some computers. On the other hand, machine learning (ML) cannot recognize right or wrong data and treat all wrong data as right data, so if training real measured data from meteorological observatory directly then the result can have er- rors. Deep learning is more suitable for predicting wind speed in a very long term by training massive amounts of data without noise. 44 The main innovation of the research is to develop an effective machine learning (ML) algorithm which is based on LSTM algorithm and XGBoost and the final goal is to im- prove the accuracy and save the model running time of wind speed prediction base as the decision-making time is limited. These two newly published journal papers which is described in figure 7 has been listed above sets a benchmarking for researchers who explore the topic of optimizing algorithms in WPF area using AI technology through reanalyzing meteorological reanalysis data. This research develops a neural network algorithm, which is based on LSTM and XGBoost, and this algorithm is validated to show better performance when compared with traditional ones. XGBoost is the opti- mal choice as it needs no big data and operates quickly. Searching for these predic- tions, high accuracy requirement with limited data and the computational time of XGBoost must be reasonably low. More attention has been paid to utilize renewable energy to produce electricity, but the random input does not always match the demand. Therefore, a set of management policies with different levels of participation of the decision maker can optimize pro- cesses in energy management (Azcárate, Blanco, Mallor, Garde, & Aguado, 2012). Be- sides this, effective information could be provided to support decision-making toward appropriate energy models and systems for isolated areas with different scales and demands (Y. Liu, Yu, Zhu, Wang, & Liu, 2018). Effective wind energy potential analysis and accurate forecasting can reduce the operating cost of wind parks. A wind energy decision system which combines these two can not only provide an effective wind en- ergy assessment but can also satisfactorily approximate the actual wind speed forecast- ing rather than poor decisions (X. Zhao, Wang, Su, & Wang, 2019). 45 3 Methodology This chapter covers the research philosophy, approach, strategy, and method where methodological choice is described as following based on the theory of Research Onion developed by Sauders et al. firstly in 2007. This is interdisciplinary research that com- bines the knowledge of Industrial Management, Artificial Intelligence and Meteorology. The main method is quantitative analysis and Python is the programming tool. This chapter also depicts research strategy, research design, data collection and results, data analyzes and application and managerial implications. 3.1 Research Strategy 3.1.1 Research philosophy This research is a positivism one based on it is quantitative research which aims to pre- dict wind speed and explain the whole process. The results are verified and depicted in chapter 4. The science used in this research can be judged by logic rather than com- mon sense. Positivism adopts a clear quantitative approach to investigating phenomena, as op- posed to post-positivist approaches, which aim to describe and explore in-depth phe- nomena from a qualitative perspective (Crossan, 2003). The five main principles of positivism research philosophy can be summarized as the following (Dudovskiy, 2018): 1. There are no differences in the logic of inquiry across sciences. 2. The research should aim to explain and predict. 3. Research should be empirically observable via human senses. Inductive reasoning should be used to develop statements (hypotheses) to be tested during the research process. 46 4. Science is not the same as common sense. Common sense should not be allowed to bias the research findings. 5. Science must be value - free and it should be judged only by logic. 3.1.2 Research approach The logical sequence of deduction is from rule to case to result, and induction is from case to result to rule, whereas abduction follows another process – from rule to result to case (Taylor, Fisher & Dufresne 2002; Danermark 2001). This research uses both in- ductive and deductive approaches to develop theory. This research uses mainly deduc- tion to test and evaluate the data-driven model in empirical research by carrying out three different algorithms. In paper 4, it also uses some induction to build theory from case study research, for example the integration of individual models of meteorological information, wind power prediction module and operations management module to construct a holistic model, named Meteorological Information Service Decision Support System. Figure 7. Approach to theory development. Figure 7 depicts the logic relationship of research approach among four papers. Paper 1 made a literature review about state-of-art from related articles. This research uses an inductive approach to indicate and display the core content in paper 3 by finding and observing the universality or commonality both from paper 1 and paper 2, to summa- rize the theory and framework. Paper 4 used a deductive method to test the proposed framework and system from paper 3. 47 Figure 8 describes the process to build a dataset for machine learning. It mainly in- cludes training dataset and test dataset by using different algorithms. Data driven approaches include machine learning, deep learning, parameter tuning, training, validation, and test. Training: to train the models. Validation: to make sure the models are not overfitting. Test: to determine the accuracy of the models. Figure 8. To build a dataset for machine learning. 3.1.3 Research strategy This is empirical research which draws conclusions strictly from concretely empirical and verifiable evidence. The empirical evidence can be gathered using both quantita- tive market research and qualitative market research methods. This research chooses a 48 quantitative method to carry out research. It uses case study and archival research as research strategies. This research chooses Vaasa wind farm as research site (63.05641°N，21.55187 °E) since the fourth related paper has done data analysis there. Besides this, this re- search obtained the collected reanalysis data of 2015-2018 from open sources to pre- dict wind speed of 2019 then compared them with reanalysis data itself to get the comparison difference. Overall plan is to predict wind speed and wind power density in 2019 by analyzing me- teorological reanalysis data 2015-2018 in the way of traditional commonly used wind power forecasting methods, ML and DL mathematical modelling, combined with XGBoost technique in a real case study. After comparing the performance and accuracy from them, it can be shown which algorithm makes the strategy reliable. The whole procedures and action include start, establishing research topic, reviewing literature, and exploring, defining research questions, mathematical modeling, collect- ing data, analyzing data, results, validating and end. There are several steps taken to complete the study. 3.2 Research Methodology 3.2.1 Main research methodology There are two basic approaches to research, qualitative and quantitative approaches. The quantitative approach can be sub-classified into inferential, experimental and sim- ulation approaches to research (Kothari, 2004). This is applied, quantitative, and empir- ical research. The preferred main research methods are quantitative ones. To be specific, this research firstly gives deeper literature review by using systematic review, association analysis and cluster analysis method to find out the research gap. 49 After summarizing and classifying innovative topics, it explores the research content and describes the research questions. The case study in Vaasa region in Finland has been studied by analyzing reanalysis data. Meanwhile, the core part of this research depicts machine learning (ML), deep learning (DP) algorithms through mathematical modelling which needs quantitative analysis methods including time series analysis, regression analysis, decision tree, ML. In general, this is interdisciplinary research and uses an interdisciplinary approach. Attached are four papers and this study uses a variety of research approaches which are shown as below. Paper 1 uses qualitative analysis methods including literature review and descriptive research. Paper 2 uses qualitative analysis methods including literature review, descriptive re- search, and contingency approach. Paper 3 uses qualitative analysis methods including literature review, descriptive re- search, exploratory research, and interdisciplinary approach. Paper 4 is the starting of quantitative research and it is exactly a case study to get into the core part of this research. This paper uses mathematical modeling and quantitative analysis, such as, time series analysis, regression analysis, to improve algorithms. 3.2.2 Programming platform MATLAB is a high-level language and interactive environment that enables it to perform computationally intensive tasks faster than with traditional programming languages such as C, C++, and Fortran. It is designed for the way of analyzing data, developing algorithms, or creating models. Python is an interpreted, high-level, general-purpose programming language and aims to help programmers write clear, logical code for small and large-scale projects. Python is more productive when compared with other programming languages, such as, C++ and JAVA. This research adapts Python to exe- cute quantitative analysis since Python can be used to make decisions involving big 50 data while Matlab can be used to teach introductory mathematics such as calculus and statistics. 3.3 Method Design 3.3.1 Artificial intelligence, machine learning and deep learning Artificial intelligence (AI) refers to any technique that enables computers or other de- vices to mimic human behavior. Machine learning (ML), a subset of AI, aims to make predictions or decisions by building mathematical models to train datasets. As a branch of ML, Deep learning (DL) underlying features a great amount of data using deep neu- ral networks. Figure 9 shows the relationship among AI, ML and DL. In general, DL is a subset of ML while ML is a subset of AI. Figure 9. Structure chart of artificial intelligence. ML are commonly used in image and video recognition, face recognition, picture de- scription, image style conversion, automatic speech recognition and synthesis, text classification, machine translation, image, and poetic creation and so on in daily life over the past several years and now. ML is an interdisciplinary technique which tries Machine Learning Data Science Deep Learning Artificial Intelligence 51 training computers or other devices to forecast the unknown features by describing the behavior of the dataset, inputting models with features regards to the expected output, forecasting output with features regards to historical data by feature extraction (Alpay- din, 2019; Griffith, 1974). Figure 10 depicts the differences of the mode between ML and human thought. ML algorithms are one of the alternatives to forecast wind power based on wind speed data as it can increase productivity, quality, and profit levels by predicting effectively in academia as well as industry (Lee, Yoo, Kim, Lee, & Hong, 2019). Deep learning (DL) is a subset of ML and pushes ML technology to be one of the essen- tial enablers for the renewed AI success with a great process (Duan, Edwards, & Dwivedi, 2019). Figure 10. Machine learning and human thinking mode. The traditional ML approaches have unavoidable limitations while producing satisfacto- ry accuracy. Firstly, the traditional ML methods lack capability to analyze and derive full value from large volumes of data. Another limitation is that the performance of the traditional ML methods highly depends on how the undergoing trend of the data could be represented by the extracted characteristic features (S. Shen, Sadoughi, Chen, Hong, & Hu, 2019; Sheng Shen, Mohammadkazem Sadoughi, Xiangyi Chen, Mingyi Hong, 2019). In other words, DL is large neural networks due to DP dealing with big data compared with extracting features of ML. However, it is very hard to identify appropri- ate characteristic features when establishing ML models. 52 DL, as a kind of supervised learning technique, has acquired growing attention and it has become more popular in recent years. It is well known for its capacity for learning hidden patterns in data (LeCun, Bengio, & Hinton, 2015). Moreover, the performance of DL models tends to be better when the dataset size grows, and DL techniques have become useful tools in data analytics (Le Cun et al., 2015). However, when the data size is small, the performance of DL tends to be jeopardized (C. Chen, Liu, Kumar, Qin, & Ren, 2019). Figure 11 shows the performance level of DP, traditional ML and human thinking. Traditional ML methods perform stable and better with a minimum intake of data. However, after crossing the threshold point, DL methods performance increases with increasing the amount of data. (Sharma, Sharma, & Jindal, 2021). Figure 11. Deep learning performance. adopted from Bhardwaj & Di, 2018 3.3.2 Fully-connected neural network, long short time memory, XGBoost Fully-connected neural networks (FCNNs) are a classic type of artificial neural network architecture, in which all the nodes or neurons in one layer are connected to the neu- rons in the next layer. A fully connected layer offers learning features from all the com- binations of the features of the previous layer, but they are incredibly computationally 53 expensive. Usually, FCNNs are only used to combine the upper layer features (Fiesler, Caulfield, Choudry, & Ryan, 1990). Figure 12 shows the network structure of FCNNs. Figure 12. Fully-connected neural network. Long Short-Term Memory networks (LSTMs) were introduced by Hochreiter & Schmid- huber (1997), and were refined and popularized by many people in the following work. LSTMs inspired mostly by circuitry, not so much biology, try to combat the vanishing / exploding gradient problem by introducing gates and an explicitly defined memory cell. LSTMs are explicitly designed to avoid the long-term dependency problem. The biggest advantage of LSTMs is that remembering information for long periods of time is practi- cally their default behavior rather than struggling to learn. Figure 13 shows the net- work structure of LSTMs (Hochreiter & Schmidhuber, 1997). 54 Figure 13. Long short-term memory network. Extreme gradient boosting (XGBoost) is an efficient and scalable implementation of gradient boosting framework by Friedman in 2001. (Friedmanet, 2001). XGBoost pack- age includes an efficient linear model solver and tree learning algorithm. It supports various objective functions, including regression, classification, and ranking. XGBoost has emerged as a robust ML technique that has been applied in several areas (Lim & Chi, 2019; D. Zhang et al., 2018) XGBoost is a decision-tree-based ensemble ML algorithm that was developed for re- gression and classification problems, which produces a prediction model in the form of an ensemble of weak prediction models. When a decision tree is the weak learner, the resulting algorithm is called gradient boosted trees, which usually outperforms random forest (Friedman, 2001). 3.3.3 Research design WPF uses wind farm historical power, historical wind speed, topography and terrain, and wind turbine operating status to establish a wind farm output power prediction model. Wind speed, power, or numerical weather forecast data are used as input for the model. Equipment status and operating conditions to get the future output power 55 of the wind farm. The real-time operation of a WPF system needs NWF data, real ane- mometer tower data, real wind power output data, wind turbine generators, and wind farm running status. This is a prediction method which uses NWP rather than historical data. NWP cannot be used for predicting wind power directly, so the power of wind farms is calculated by NWP models. As it is depicted in Chapter 2 Literature Review part, WPF usually pro- ceeds by physical forecasting approach or statistical forecasting approach. This research chooses the former approach to predict wind speed, wind direction and air density in the selected wind farm. It is important to do horizontal extrapolation from measure- ment height to hub height, from meteorological observatory site to wind farm. This research uses reanalysis data of 2015-2018 from Public Datasets in European Cen- tre for Medium-Range Weather Forecasts to predict wind speed of 2019 then compare them with historical data. The reanalysis data in this research uses short term forecast- ing and picks up wind speed data every six hours. Some related information of the se- lected wind farm is listed as follows. Kunta (Municipality):Vaasa site Sijainti (Location):（63.05641°N，21.55187°E） Vuosi (Year):2012 Kokonaisteho (Total power):4 MW Turbiineja (Turbines):1 Omistaja (Owner):Wasa Wind Oy Laitevalmistaja (Equipment manufacturer): Mervento Located area: Kronvik Wind power station code: 6741 56 Figure 14. System structure and process. adopted from Liu & Yang, 2015 (Paper 3) Table 2. Typical failures related to weather conditions. Source: Meteorological information service support system in wind park application, 2015 (pa- per 3) Failure parts Possible reasons Weather condi- tions Actions Blade Blade drive not ready EWEs Emergency stop Rotor Result of imbalance, blades and hub corrosion etc., brake sensor failure Rain, snow and other hash me- teorological con- dition Normal stop 57 Gearbox Over temperature, gearbox oil pressure too low High tempera- ture Normal stop Generator Over speed, over tempera- ture, bearing faults, current too high/low, frequency sensor failure High tempera- ture and/or hu- midity Emergency stop Normal stop Yaw system Yaw brake set unintentionally Extreme changes in wind speed / direction Normal stop Tower Weather or other failure may cause excessive vibra- tion EWEs Emergency stop Maintenances include regular, active, and passive maintenance. Passive maintenance, which is sudden maintenance, accounts for a portion of operating maintenance ex- penses and revenue. Sudden maintenance includes all unplanned failures that require man-made repair, from manually resetting the wind turbine to replacing damaged gearbox. Accidental failure of critical components (including gearbox, generator, shaft, blade, hydraulic system, transformer, and converter) can significantly increase mainte- nance costs. Failure to replace these components in a timely manner can lead to signif- icant wind turbine downtime and loss of revenue. Maintenance should be scheduled to carry out in time of low wind power. Mean absolute error of the forecasting results 𝑀𝐴𝐸 = 1 𝑛 ∑ 𝑛 𝑖=1 | 𝑌𝑖 − 𝑌𝑖´ | Root Mean Square Error 𝑅𝑀𝑆𝐸 = √ 1 𝑛 ∑ 𝑛 𝑖=1 (𝑌𝑖 − 𝑌𝑖´ )2 Standard deviation of error 𝑆𝐷 = √ 1 𝑛 − 1 ∑ 𝑛 𝑖=1 (𝑌𝑖 − 𝑌𝑖´)2 Coefficient of determination 58 𝑅2 = √ 1 𝑛 − 1 ∑ 𝑛 𝑖=1 (𝑌𝑖 − 𝑌𝑖´) 2 Wind speed (m/s) 𝑉 = √𝑢2+𝑣2 Wind power density (W/m2) ρ: kg/m3 𝐷𝑤𝑝 = 1 2𝑛 ∑ 𝑛 𝑖=1 (𝜌 ∙ 𝑉𝑖 3) 3.3.4 Data collection The important quantitative process is to predict wind speed in 2019 through ML and DL algorithms by training reanalysis data during 2015-2018, which retrieved from Climate Data Store. This research analyzes reanalysis data, which named Uncertainties in Ensembles of Regional Reanalysis (UERRA), obtained from Public Datasets in European Centre for Medium-Range Weather Forecasts (ECMWF). ECMWF aims at advancing global numer- ical weather forecasting (NWP) through international collaboration. UERRA is a re- search project among 5 pre-operational Copernicus Projects in ECMWF during 2014- 2018. The objective of UERRA is to produce ensembles of European regional meteoro- logical reanalysis of Essential Climate Variables (ECVs) for several decades and to esti- mate the associated uncertainties in the data sets. It also includes recovery of historical (last century) data and creation of user-friendly data services. Data format is .netcdf and .grib and this research retrieved the former format. Python is used for creating mathematical models and analyzing data in this research. The reanalysis data of wind speed obtained four times in each day for 00-, 06-, 12-, and 24-h. 59 4 Result and Findings This chapter depicts results and findings of this research about three algorithms. The training performance and testing performance are also showed here. 4.1 Results of Algorithm This research uses AI technologies, such as, fully-connected neural network (FCNN), long short time memory (LSTM) and extreme gradient boosting (XGBoost). Figure 15 shows the hierarchy diagram about them. Figure 15. Tree structure of technologies related to Artificial Intelligence. • Fully-connected neural network The line chart of predicted wind speed by using a FCNN is shown in figure 16 and figure 17 depicts its prediction performance. 60 Figure 16. Wind speed prediction using fully-connected neural network. Figure 17. Prediction performance of fully-connected network. • Long Short-Term Memory network The line chart of predicted wind speed by using a fully-connected neural network is shown in figure 18 and figure 19 depicts its prediction performance. 61 Figure 18. Wind speed prediction using LSTMs. Figure 19. Prediction performance of LSTMs. • XGBoost Regression The line chart of predicted wind speed by using a fully-connected neural network is shown in figure 20 and figure 21 depicts its prediction performance. 62 Figure 20. Wind speed prediction using XGBoost Regression. Figure 21. Prediction performance of XGBoost Regression. Training performance and testing performance, which include the value of RMSE, MAE, R2, for each algorithm are displayed in table 3. Table 3. Performance for each algorithm. Algorithm Value Training Performance Testing Performance Fully-connected neural network RMSE 1.892591 2.006670 MAE 1.428197 1.526733 R2 0.667309 0.685465 LSTMs RMSE 1.81 1.90 MAE 1.37 1.46 R2 0.694925 0.717418 XGBoost Regression RMSE 1.394653 1.938518 MAE 1.084520 1.472376 R2 0.819341 0.706468 63 Training time and testing time for each algorithm are displayed in table 4. Table 4. Training time and inference time for each model. (second : s) Training Time Inference Time Fully-connected neural network 151.4038s for 1 time epochs=60 batch size=2 verbose=2 1.8487s for 10 times LSTMs 182.8195s for 1 time epochs=20 batch size=1 verbose=2 0.5338s for 10 times 0s 42.037s for 10 times 64 XGBoost is applied to quickly achieve comparable accuracy with LSTM but a lot less time to improve the prediction. As it can be seen from table 3, XGBoost performs bet- ter than FCNNs and can also reach nearly the same effect as LSTMs. From the aspect of statistics, the results of XGBoost and LSTMs are equivalent. XGBoost only has a little bit overfitting when using the model to train the data set and predict as it can be neglect- ed that the differences in a few digits after the decimal point. With the help of this sit- uation, the uncertainty in this decision-making process can be handled more effectively. Besides this, it is also employed to reduce computational time. XGBoost is not DL but a ML technique used for regression and classification problems, so its significant ad- vantage is running many times faster than DL algorithm. It is a ML algorithm rather than a neural network or DL, so it has no training time but learning time, which is the same as running time of model, so-called inference time which is shown in table 4. As it can be seen in table 3 and table 4, the results clearly demonstrate that the XGBoost algorithm shows an overall better performance as compared to the traditional FCNN and LSTMs method as it saves much model running time and has equivalent MAE, RMSE and R2 with DL algorithm. It is a regression based on a decision tree. Wind speed forecasts can be deep ML or not deep. It is commonly believed that DL has better per- formance. The algorithm used in this research compares both, XGBoost is one of the best algorithms for which is not DL since it is validated that it can reach the same level of accuracy but save much computational time. Figure 22 shows reanalysis data, training data and testing data of wind speed from 2015 to 2019. As it can be seen from the graph, the blue curve displays the trend of reanalysis data during 2015-2019, the orange curve represents the variation of training data on wind speed and green curve shows the predicted wind speed. Besides these, it presents the variation of wind speed for each year during 2015-2019. The highest wind speed occurred in March, February, April, March, and April separately 65 in the years of 2015, 2016, 2017, 2018 and 2019. Vice versa, the lowest wind speed occurred in August, February, July, August, and December separately in the years of 2015, 2016, 2017, 2018 and 2019. Figure 22. Average daily wind speed during 2015-2019. As it can be seen in figure 23, the average wind speed for every 6 hours in each month in the year of 2019 are drawn individually. The variation features of wind speed are: Judging the variation of wind speed in 2019, the highest wind speed occurred in April and the lowest wind speed occurred in December in the year 2019. 66 Figure 23. Monthly wind speed in each month in 2019. According to Finnish Meteorological Institute, Finland located in the zone of westerly air disturbances, there are great variations in air pressure and winds, especially in win- ter. In the whole country, the wind blows most commonly from the southwest and least commonly from the northeast. The average wind speed is between 2.5 and 4 m/s inland, slightly higher on the coast and 5 to 7 m/s in maritime regions. Wind speeds are typically highest in winter and lowest in summer. Moderate winds are typical of Finland; high winds are rare, particularly inland. Vaasa located in the southwest coast of Finland. The research results are consistent with features above. 4.2 Findings 4.2.1 Findings of algorithm ML and DL algorithms include an exciting prospect for many industries and businesses to drive self-service, increase agent productivity and make workflows more reliable. Based on the overview of ML and DL with illustrations and differences, each of them is focusing on respective characteristics and future trends. DL algorithms need to clean up big data and they do not suit every case. In some practical cases, it makes nonsense when making decisions in the management layer as the dataset training time for DL algorithms is too long. In the Research Design part of Chapter 3, it has been found that an ML approach has been applied to predict wind speed. Wind power prediction mod- els are achieved in an adaptive and effective way by effectively reducing training time of FCNN and LSTM respectively. The hybrid algorithm introduced in this research, which is based on FCNN, LSTM and XGBoost is completely novel in this disciplinary area, which uses hybrid DL algorithm to execute energy management, basically only two re- lated exist in the existing literature. Moreover, from the data analysis and results, the optimized algorithm which is based on LSTMs algorithm and XGBoost technique has better performance in the Vaasa meteorological observation site. 67 The WPF method is an effective way to reduce the impact of wind power intermittently to the power grid. Forecast results can be published and sent to the power grid dis- patch terminal and wind farm monitoring center in real time. Meanwhile, the grid scheduling center and wind farm monitoring center can make prediction requests to the forecast server at any time. 4.2.2 Findings of managerial aspect This research uses meteorological methods to provide a decision support tool for deci- sion-makers. This meteorological information service decision support system in wind park application is beneficial both to wind farms and power systems. One hand, it pro- vides support for reasonable maintenance plans, participates in market competition, and reduces the operating cost of wind farms. On the other hand, it provides support for power grid scheduling and proper scheduling, effectively reducing the adverse ef- fects of intermittent wind power on power systems. Its successful implementation will produce enormous economic and social benefits. The main contribution of this research is to achieve decision optimization on a decision support system by using AI technology. It was concluded that the proposed system is very promising for potential applications in wind (power) energy management. The findings of this research will provide strategic management for more enterprises in the field of wind power, which plan to implement systems with awareness of risk factors to avoid equipment failure, supply with regular, active, and passive maintenance, optimize energy management, give businesses an advantage over competitors and always be aware of the changing market. For further work, it is necessary to use many methods to reach the optimum results in wind speed prediction, EWEs risk management and optional service decisions, the comparison between many methods ensure best performance of the system and real- ize the objective aims of the research. 68 5 Summary of Publications 5.1 Overview of Papers Titles of published articles with keywords are listed in Table 5. An overview of the aims, methods and the main results/contribution are shown in Table 6. The original articles (paper 1-4) are attached in Appendix. Table 5. List of articles´ titles with key words. Paper Title Key words Paper 1 A review of Innovation in Wind Power Forecast Wind energy resource, WPF, wea- ther forecast, literature research, descriptive research. Paper 2 The Impact of Climate Change on Wind Power Enterprises WPF, WPEs, climate change, glo- bal wind energy resource distri- bution, climate data record. Paper 3 Meteorological information service support system in wind park application Operations management, Deci- sion support systems, Informa- tion management, Paper 4 A Study on Renewable Energy Potential based on the Global Atmospheric Data Renewable energy, climate change, reanalysis, ERA-20C data, resources potential. Table 6. Overview of the articles’ aims, methods and the main results/contribution. Aims Methods Main results/contribution Paper 1 To review the most recent articles in the topic of WER and WPF Systematic review, cluster analysis, asso- ciation analysis, litera- Explore the research gaps in this area and highlight the possible future 69 methods. To review the innovations. ture review, descrip- tive research. research points. Paper 2 To survey the impact of climate change on WPF and the in- fluences on WPEs. Literature review, des- criptive research, con- tingency approach. WPD, which represents wind resources, plays a critical role. Air tempera- ture, humidity, wind di- rection, wind speed, air pressure and rainfall di- rectly influence the wind power output. Paper 3 To propose a concep- tual framework and make it can be used for decision-making. Literature review, des- criptive research, ex- ploratory research, mathematical mode- ling, interdisciplinary approach. This structured frame- work, which involves three major modules and certain processes, pro- vides new insight for de- cision making in WPEs. Paper 4 To find a correlation between the meteoro- logical factors and the renewable energy po- tential and make the conceptual framework empirical. Literature review, case study, quantitative analysis, mathemati- cal modeling, time series analysis, re- gression analysis, in- terdisciplinary ap- proach. Use the global atmosphe- ric reanalysis data to ana- lyze the potential of renewable energy sources In Vaasa region in Finland. 5.2 Logical Connection of Papers This study is founded on a thorough literature review. Hence paper 1 is the foundation for the following research since it reviews the innovations in the domain of WPF. Paper 70 2 makes a survey on the specific branch. In fact, paper 4 tests the framework which was proposed from paper 3. Figure 24 shows the logical connections among four papers. Figure 24. Logical connections among the papers. 5.3 Summary of Individual Papers 5.3.1 A review of innovation in wind power forecast Sub-question 1. What is the innovation in the development process of WPF among so much relevant research? This paper is the theoretical foundation of the whole research. It roughly researches main topics in the field of WERs and WFP. In answering sub-question 1, Paper 1 uses different methods to review relevant articles to find the possible research prospect. In this paper, several methodologies, including systematic review, cluster analysis and association analysis, are used to depict and generalize different popular WPF methods. Paper 1 reviewed several different WPF methods, which were used in wind energy sys- tems, to summarize their own theory and characteristics. The corresponding innova- 71 tions are also reviewed in detail. This paper aims to find out the research gap and open a new view to provide a research path for researchers in this same field. Paper 1 describes some single algorithms, such as Kalman Filters (KF), Artificial Neural Network (ANN), Support Vector Machine (SVM), fuzzy logic and time series model. Time series model includes auto regression (AR), moving average (MA), autoregressive moving-average (ARMA) and autoregressive integrated moving average (ARIMA). Some hybrid algorithms, such as KF+SVM, ARIM+ANN, ARIMA+KF, ARIMA+SVM, also are compared in determined case studies. As it can be seen, there is no single best forecasting algorithm that can be applied to any wind farm. As a result of this study, it can be said that each algorithm or combined algorithm both has its advantages and disadvantages. Meanwhile, the characteristics of terrain in a variety of districts in different countries are different. Therefore, paper 1 shows there is still space to increase the prediction accuracies respectively in scales of long-term, medium-term, short-term, and very short-term WPF. Paper 1 suggests that the future study can be topics about increasing the prediction accuracy and system reliability. 5.3.2 The impact of climate change on wind power enterprises Sub-question 2. How climate change influences WPEs and what factors affect wind power output? This paper is a preliminary basis of connecting wind power with meteorology. It aims to study the impact of climate changes on WPEs and draw a framework to determine the relationship between wind power density and wind power. In answering sub-question 2, Paper 2 was developed based on paper 1. It reviewed some most recent relevant research which combines climate change with wind power predicting and summarized the art-of-state. Paper 2 draws a framework which assesses 72 wind resources and finds correlations between wind power and some meteorological elements. Six major meteorological elements, including air temperature, humidity, wind direction, wind speed, air pressure and rainfall, may have a relationship with wind power. Besides these, other meteorological elements may also be connected to wind power, such as relative humidity, rainfall, and snowfall. In some practical cases, wind farms can be lo- cated in places with complex terrain. However, different terrain, surface roughness, obstacles, and ground conditions such as undulating terrain, land-to-sea junctions, or uneven distribution of precipitation or cloud volumes also influence wind speed. The outcome of Paper 2 is proposing a rough framework which aims to help decision- makers of WPEs to make strategic decisions including site selection, management and maintenance of wind power station, and long-term wind power generation forecast etc. Paper 2 suggested that future research should explore deeply in climate data record (CDR) to help develop more effective wind speed and WPF methods by finding specific algorithms. 5.3.3 Meteorological information service decision support system in wind park ap- plication Sub-question 3. Can there be a general framework to help forecasting wind speed and wind power more effectively in decision-making? This paper is the hub of this interdisciplinary research as it is involved in domains of operation management, weather forecast and wind power generation. It mixes these subjects together and practices them in a conceptual framework. In answering sub-question 3, this paper designs and provides a meteorological infor- mation service system. The proposed system, which involves meteorological infor- mation module, wind power prediction module and operations management decision- 73 making module, can be seen in the picture of system structure and process in Figure 10. It provides benchmarking to support decision making directly and indirectly based on processing meteorological information and evaluating its impact on service operations. Additionally, it provides meteorological forecasting and decision support in case of EWEs. The Meteorological Information module includes meteorological data collection and NWP. This module also sends early warning messages in case of EWEs, so managers can take countermeasures in advance to avoid economic losses. Wind Power Prediction module utilizes meteorological data to predict wind power output based on real-time measuring, NWP, and WPF. Operations Management module uses predicted results from the previous module to evaluate failure probabilities in different parts of the wind turbines. It can help a lot in decision making to optimize maintenance schedules and maximize wind power output. Table 1 shows the corresponding actions to specific ma- chine failures. Paper 3 suggests the future research will be implemented with the proposed concep- tual model. To some extent, paper 3 is the theoretical foundation of paper 4 which demonstrates empirical research by analyzing CDR data. 5.3.4 A Study on renewable energy potential based on the global atmospheric data Sub-question 4. How to use the global atmospheric reanalysis data to analyze the potential of WERs in Finland? This paper analyzes ERA-20C global datasets from ECMWF and tries to identify the me- teorological factors (wind speed, solar radiation, rainfall, evaporation etc.) with their effects on the overall utilization potential of these RERs. What is the correlation be- tween the meteorological factors and the renewable energy potential? 74 Paper 4 is developed based on paper 3 and uses the proposed pilot conceptual mete- orological information service decision support system from paper 3. In answering sub-question 4, this paper starts by retrieving the dataset of ERA-20C to analyze how climate change reflects on WPF. This paper continues such analysis in me- teorological information service decision support systems, with deeper insights into WPD during the past 50 years, and to verify how this system helps decision-making in WPEs especially when EWEs come. This paper covers an important topic with the clear- ly presented purpose in terms of RQ 4, WPD of past fifty years 1961-2010 were studied from a point of view by analyzing global atmospheric reanalysis data, to find out the correlation between the meteorological factors and wind energy potential. This research analyzes the existing ERA-20C global datasets describing the state of the atmosphere as well as land-surface and ocean-wave conditions from 1900 to 2010 ob- tained from Public Datasets in European Centre for Medium-Range Weather Forecasts (ECMWF). ECMWF aims at advancing global NWP through international collaboration. The data format is .netcdf and .grib. MATLAB and Python are used for creating models and analyzing data in this research. Paper 4 demonstrates how to analyze the potential of WERs by using reanalysis data in a real case study. The specified location is Vaasa region in Finland and retrieved format is NetCDF. The highest resolution grid 0.125 degree * 0.125 degrees was chosen and a total data of 18262 days in 50 years were analyzed. Paper 4 uses MATLAB R2014a as the programming language to calculate, analyze and plot figures. It also plotted the variations of maximum WPD from every five years and every ten years. Analyzed re- sults present a trend of WPD and give alarm to decision-maker to take action to avoid machine failures and financial losses. This Meteorological Information Service Decision Support system which was proposed in paper 3 can effectively help decision-maker at macro level and paper 4 is exactly a case study to validate this framework. The contri- 75 bution of this paper is analyzing the potentials of WERs in terms of mastering the trend of WPD in Vaasa region in Finland. Paper 4 proposes that the future research could be focused on forecasting the global potentials of RERs in the next decades. 5.3.5 Research in this study Sub-question 5. What is the sufficient ML algorithm to improve the accuracy of wind speed prediction? The follow-up research after these four papers in this study answers sub-question 5. It is developed based on paper 4 and reanalysis data of Uncertainties in Ensembles of Regional Reanalysis (UERRA) while paper 4 uses ERA-20C global datasets also retrieved from European Centre for Medium-Range Weather Forecasts (ECMWF). The research design in this study is the core part of this entire research. It uses reanalysis data of 2015-2018 from Public Datasets in ECMWF to predict wind speed of 2019. The reanaly- sis data in this research uses short term forecasting and picks up wind speed at the selected site for every six hours. In answering sub-question 5, the research design in this study compares traditional algorithm FCNN, ML algorithm LSTM and DL algorithm XGBoost, by calculating RMSE, MAE, R2 in terms of training performance and test performance, to find an optimal wind speed predicting method. Besides this, it observes training time and inference time for each algorithm. The results come out that the XGBoost algorithm shows better performance as compared to the traditional FCNN and LSTMs method as it saves quite much model running time but also reaches the equivalent effect of MAE, RMSE and R2 as DL algorithm LSTM. Besides this, the research design included in this study aims at optimizing energy man- agement decision-making by optimal operational planning via predicting wind speed. It 76 provides a plan about when to turn off wind turbine group in order to repair and carry out maintenance during the low power generation period for wind power forecast to the next year, provide better wind energy assessment results and to optimize energy storage for the whole electrical grid, and aims at reducing carbon emissions by utilizing renewable energy maximally instead of fossil fuel. Wind power needs to be predicted accurately to make up for problems derived from burning fossil fuels. Central research question: How to improve the accuracy of wind power forecasting by using artificial intelligence methods? In answering the central research question, the research design in this study compares traditional algorithm and DL algorithms and determines a ML algorithm namely XGBoost to predict wind speed. And it is the core part of the whole research. It is pos- sible that Meteorological Information Service Decision Support System, which was pro- posed in paper 3, can support decision-making effectively and create timely actions within the WPEs. Findings from this research contribute to WPF in WPEs. The main contribution of this research is to achieve decision optimization on a decision support system by using AI. It was concluded that the proposed system is promising for poten- tial applications in wind (power) energy management. 77 6 Discussion and Conclusions 6.1 Contribution The motivation of this research comes from the “environmentally friendly society”, “sustainable development” and “clean energy”. As everyone knows, the natural re- sources which human beings depend on are not unexhausted. Therefore, how to de- velop and utilize RERs in an efficient way is a popular and forever topic. Among all the RERs, wind energy has advantages, such as, low cost of wind power generation, clean to environment resource renewable despite the characteristic of uncertainty. A lot of WPF techniques and methods used in wind energy systems have been re- viewed in Chapter 2 of this study to summarize their own theory and characteristics in a variety of methods. And their corresponding innovations are also reviewed in detail. However, the main contribution of Chapter 2 is to provide a path for researchers in this same field. The important points of view for this research were described and summarized from Chapter 3 to Chapter 5. The proposed structure and process of this conceptual infor- mation service system for improving productivity can help decision makers in WPEs while the electricity grid balance must be maintained between electricity consumption and generation at any moment. In the module of Meteorological Forecast, real-time meteorological data and weather forecasts are collected through meteorological sen- sors and equipment. Managers can get warning signals and take countermeasures quickly in advance when EWEs happen by predicting 50-year maximum wind speed uninterrupted. In the module of Power Prediction, it can provide WPF prediction by utilizing real-time wind measuring data and historical data as input. Models which combine different algorithms usually have higher accuracy and reliability than compar- ing using just a single algorithm. In the module of Operation Management, failure probabilities are evaluated to help decision makers to reduce maintenance cost and time and to improve the operational efficiency and reliability. Correspondent actions 78 can be taken regarding typical failures in different parts of wind turbines according to real case statistics. Condition-based maintenance needs to be taken while there is a direct connection between business performance and operational management based on condition-based maintenance in WPEs. 6.2 Managerial Implications This research develops a decision support tool for decision-maker from the domain of grid dispatching companies. The important quantitative process of it is predicting wind speed in 2019 through ML algorithm by training reanalysis data during 2015-2018, which retrieved from Climate Data Store. Then validate through analyzing the values of MAE, RMSE, R2 in the categories of training performance and testing performance. The practical impact of this research is that it examines a method which can help the whole wind power generation process in a systematic way, develop a practical evalua- tion tool for management level, improve wind power prediction accuracy and reduce economic losses by increasing wind speed prediction accuracy. In the long term, it can effectively alleviate air pollution, water pollution and global warming problems. Be- sides of these, this research supports the growing recognition that the timeliness of making decision is just as important to the effectiveness of weather warnings as infor- mation provided in risk management of EWEs and actions of machine failures, and this factor should be considered in future research in addition to the investments and at- tention given to improving detection and warning capabilities. The major managerial implications of this research are described as follows. a. Provides a plan about when to turn off the wind turbine group to repair and carry out maintenance during the low power generation period for wind power forecast to the next year. Therefore, it can serve as an effective tool for wind farm management and decision-making. The proposed structure and process of this conceptual information service system for improving productivity can help decision makers in WPEs while the electricity grid bal- 79 ance must be maintained between electricity consumption and generation at any mo- ment. It is a holistic wind energy decision support system based on condition-based maintenance (CBM), which includes meteorological information module, wind power prediction module and operations management decision-making module, for decision makers to cut down operation and maintenance costs and implement a successful CBM strategy to achieve higher level of cost effectiveness. b. Provides better wind energy assessment results and to optimize energy storage for the whole electrical grid. It can generate a lot of value, such as, making electricity price can falling faster than expected. The development of energy storage has a close relationship with transition to the smart grid. Energy storage plays a crucial role in offsetting the intermittency of renewable energy including wind energy predictable plan helps in dispatching effec- tively and helps users to save electricity cost. Energy storage can help balance the power generation and improve power quality. Optimal plan of energy management can reduce economic loss when making decision. c. Reduces the carbon emissions by utilizing renewable energy maximally instead of fossil fuel. Burning fossil fuel is the main cause of climate change. Among a variety of traditional and new energy, to find the best plan of energy allocation among them to minimize carbon emissions and to make balance between them is becoming a popular topic. In the long term, this information system can reasonably help balancing the fossil fuels and renewable energy in the purpose of protecting the environment for human beings. 6.3 Research Limitations From this research itself, reanalysis data are retrieved at the height of 30 meter as the same as a meteorological observatory. 80 This research originally plans to use reanalysis data of 2015-2018 from ERA5 data to predict wind speed of 2019 then compare them with historical data from Finnish Me- teorological Institute in 2019. In practice, it is not easy to compare the predicted wind speed with historical wind speed in the year 2019 since there exists distortion since many observation data are missing. Wind energy is one of the most dynamic renewable sources of energy with commercial potential, clean and green, low-cost, widespread, inexhaustible. Wind power can effec- tively mitigate air pollution, water pollution and global warming while providing a sta- ble power supply for economic growth. However, wind power is intermittent and fluc- tuating as intermittent is a nature characteristic of wind energy itself. Wind power in- terval prediction plays an increasingly important role in evaluations of the uncertainty of wind power and becomes necessary for managing and planning power systems. Be- sides of this, energy storage systems with new technology can also compensate for improving the reliability of the system pertaining to power availability (Abazari, Babaei, Muyeen, & Kamwa, 2020; Vijay M, Singh, & Bhuvaneswari, 2020; R. Wang, Li, Fu, & Tang, 2020). Hannele, Jari and Samuli in Ilmatieteen Laitos (Finnish Meteorological In- stitue) have done related research about WPF accuracy and uncertainty in Finland in the year of 2013. They pointed out that the aggregation of wind power production will not only decrease prediction errors, but also decrease the variation and uncertainty of prediction errors by analyzing density function and kernel densities in three sites (Holt- tinen, Miettinen, & Sillanpää, 2013). The ecological problem of wind power generation is interference to birds. Some types of wind turbine projects cause bird death, and these deaths may contribute to declines in the population of species also affected by other human-related impacts. The wind energy industry and the U.S. government are researching ways to reduce the effect of wind turbines on birds and bats. Modern wind turbines can be very large machines, and they may visually affect the landscape. Some people do not like the sound that wind turbine blades make as they turn in the wind. Producing the metals and other 81 materials used to make wind turbine components has impacts on the environment, and fossil fuels may have been used to produce the materials. The main solution is offshore wind power generation with higher cost from power gen- eration but also high efficiency. Onshore wind power generation influences fisheries trade and marine mammals. In some regions over the world where the population is denser, to find the location for installing wind turbines are more restricted sometimes, offshore wind farms will be vigorously developed. Meanwhile, wind power generation makes a lot of noise, building wind farms in some empty places can be a possible bet- ter choice. For the algorithms, there is no single best forecasting algorithm that can be applied to all renewable energy systems. As a result of this research, it can be shown that each algorithm or combined algorithm has its advantages and disadvantages. There is still space to increase the prediction accuracies respectively in scale of very short-term, short-term, medium-term, and long-term wind power predictions. In general, there is considerable room for WPF development as wind power generation technology is not fully mature as there existed objective constraints. The limitations of this study include the research being established based on the existing literature. Meanwhile, some aspects are potentially ignored as this research is initially based on the existing literature. Furthermore, portability needs to be improved when planting into other wind farms. 6.4 Future Research As wind power generation develops rapidly and the installed capacity is increasing fast in recent years, the innovative application of “environmentally friendly power supply” is getting much closer to our daily life. The dependence on electricity is increasing while the rapid development of society is changing day by day. The improvement of social production and the continuous development of people's standard of living have a strong demand for power resources. Human society urgently needs to improve the 82 efficiency of electricity consumption as the power resources are in shortage. (Rudenko, Ershov, & Evstafiev, 2017). Currently, mostly WPEs consider the results of weather forecasts as a factor in helping managers to make decisions which rely on historical and prevailing meteorological data. However, nearly no research considers the impact of climate change while climate change over 10 years or even longer. Further research may mainly aim at figuring out how climate change over decades can influence wind power through comparing and analyzing climate data records in complex terrain. More specifically, the objective is to explore the impact of climate change on weather conditions, especially EWEs which influence the wind power output in wind power enterprises. Besides this, as wind power has a cubic dependency on wind speed, this error from wind speed forecast can increase when predicting wind power. Therefore, finding solutions for getting higher accuracy of wind speed observation sites and developing better prediction models are continuously research tasks. In the future, more research is needed in the field of wind power prediction for the purpose of optimizing real time data, increasing the predic- tion accuracy and system reliability. ML is widely used in many domains. The concept of DL has been introduced by Geof- frey Hinton, Yoshua Bengio, Yann Lecun in 2006. Recently, DL techniques have started to be used in the WPF area. DL usually has better performance than traditional ML in certain conditions when the training set is big enough. Electric energy is one kind of secondary energy, it cannot be stored since electricity must be generated as much as needed. It is waste if it generates more than needed while power cuts will happen if less than needed. Therefore, the power generation must follow the load of the power to adjust which is called peak regulation. Thermal power plants have strong peak-regulating capacity while wind farms have poor peak- regulating capacity. For this reason, wind power and thermal power plants must be packaged into the grid, to achieve wind power peak. Aims to be integrated into the 83 power grid, the wind power grid connection has always been a problem. The future research could include increasing electricity generating stability in power companies based on better peak regulation in practice. 84 References Abazari, A., Babaei, M., Muyeen, S. M., & Kamwa, I. (2020). Learning adaptive fuzzy droop of PV contribution to frequency excursion of hybrid micro-grid during pa- rameters uncertainties. International Journal of Electrical Power and Energy Sys- tems, 123(December 2019), 106305. https://doi.org/10.1016/j.ijepes.2020.106305 Adeyeye, K., Ijumba, N., & Colton, J. (2020). Exploring the environmental and economic impacts of wind energy: a cost-benefit perspective. International Journal of Sus- tainable Development and World Ecology, 27(8), 718–731. https://doi.org/10.1080/13504509.2020.1768171 Akyuz, E., Oktay, Z., & Dincer, I. (2012). Performance investigation of hydrogen produc- tion from a hybrid wind-PV system. International Journal of Hydrogen Energy, 37(21), 16623–16630. https://doi.org/10.1016/j.ijhydene.2012.02.149 Albrecht, J., Laleman, R., & Vulsteke, E. (2015). Balancing demand-pull and supply-push measures to support renewable electricity in Europe. Renewable and Sustaina- ble Energy Reviews, 49, 267–277. https://doi.org/10.1016/j.rser.2015.04.078 Alexiadis, M. C., Dokopoulos, P. S., & Sahsamanoglou, H. S. (1999). Wind speed and power forecasting based on spatial correlation models. IEEE Transactions on En- ergy Conversion, 14(3), 836–842. https://doi.org/10.1109/60.790962 Alpaydin, E. (2019). Introduction to machine learning. Ammari, C., Belatrache, D., Touhami, B., & Makhloufi, S. (2021). Sizing, optimization, control and energy management of hybrid renewable energy system- a review. Energy and Built Environment. https://doi.org/10.1016/j.enbenv.2021.04.002 Anurag Bhardwaj, Wei Di, J. W. (2018). Deep Learning Essentials: Your hands-on guide to the fundamentals of deep ... - Anurag Bhardwaj, Wei Di, Jianing Wei - Google Böcker. Retrieved April 14, 2021, from https://books.google.se/books/about/Deep_Learning_Essentials.html?id=ISBKD wAAQBAJ&redir_esc=y 85 Archer, C. L., & Jacobson, M. Z. (2013). Geographical and seasonal variability of the global “practical” wind resources. Applied Geography, 45, 119–130. https://doi.org/10.1016/j.apgeog.2013.07.006 Azcárate, C., Blanco, R., Mallor, F., Garde, R., & Aguado, M. (2012). Peaking strategies for the management of wind-H 2 energy systems. Renewable Energy, 47, 103– 111. https://doi.org/10.1016/j.renene.2012.04.016 Bhattacharjee, S., & Acharya, S. (2015). PV–wind hybrid power option for a low wind topography. Energy Conversion and Management, 89, 942–954. https://doi.org/10.1016/j.enconman.2014.10.065 Bonou, A., Skelton, K., & Olsen, S. I. (2016). Ecodesign framework for developing wind turbines. Journal of Cleaner Production. https://doi.org/10.1016/j.jclepro.2016.02.093 Bórawski, P., Bełdycka-Bórawska, A., Jankowski, K. J., Dubis, B., & Dunn, J. W. (2020). Development of wind energy market in the European Union. Renewable Energy, 161, 691–700. https://doi.org/https://doi.org/10.1016/j.renene.2020.07.081 Box, G. E. P., & Jenkins, G. M. (1976). Time series analysis: forecasting and control, re- vised ed. Holden-Day. Breslow, P. B., & Sailor, D. J. (2002). Vulnerability of wind power resources to climate change in the continental United States. Renewable Energy, 27(4), 585–598. https://doi.org/10.1016/S0960-1481(01)00110-0 Cadenas, E., & Rivera, W. (2010). Wind speed forecasting in three different regions of Mexico, using a hybrid ARIMA–ANN model. Renewable Energy, 35(12), 2732– 2738. https://doi.org/10.1016/j.renene.2010.04.022 Çapik, M., Yılmaz, A. O., & Çavuşoğlu, İ. (2012). Present situation and potential role of renewable energy in Turkey. Renewable Energy, 46, 1–13. https://doi.org/10.1016/j.renene.2012.02.031 Carolin Mabel, M., & Fernandez, E. (2008). Analysis of wind power generation and pre- diction using ANN: A case study. Renewable Energy, 33(5), 986–992. https://doi.org/10.1016/j.renene.2007.06.013 86 Carta, J. A., & Mentado, D. (2007). A continuous bivariate model for wind power densi- ty and wind turbine energy output estimations. Energy Conversion and Man- agement, 48(2), 420–432. https://doi.org/10.1016/j.enconman.2006.06.019 Cassola, F., & Burlando, M. (2012). Wind speed and wind energy forecast through Kal- man filtering of Numerical Weather Prediction model output. Applied Energy, 99, 154–166. https://doi.org/10.1016/j.apenergy.2012.03.054 Çetin, O., & Ziya Sogut, M. (2021). A new strategic approach of energy management onboard ships supported by exergy and economic criteria: A case study of a car- go ship. Ocean Engineering, 219(March 2020), 108137. https://doi.org/10.1016/j.oceaneng.2020.108137 Chadee, X. T., & Clarke, R. M. (2014). Large-scale wind energy potential of the Caribbe- an region using near-surface reanalysis data. Renewable and Sustainable Energy Reviews, 30, 45–58. https://doi.org/10.1016/j.rser.2013.09.018 Chang, T.-J., Chen, C.-L., Tu, Y.-L., Yeh, H.-T., & Wu, Y.-T. (2015). Evaluation of the climate change impact on wind resources in Taiwan Strait. Energy Conversion and Man- agement, 95, 435–445. https://doi.org/10.1016/j.enconman.2015.02.033 Chávez-Ramírez, A. U., Vallejo-Becerra, V., Cruz, J. C., Ornelas, R., Orozco, G., Muñoz- Guerrero, R., & Arriaga, L. G. (2013). A hybrid power plant (Solar–Wind– Hydrogen) model based in artificial intelligence for a remote-housing applica- tion in Mexico. International Journal of Hydrogen Energy, 38(6), 2641–2655. https://doi.org/10.1016/j.ijhydene.2012.11.140 Chen, C., Liu, Y., Kumar, M., Qin, J., & Ren, Y. (2019). Energy consumption modelling using deep learning embedded semi-supervised learning. Computers & Indus- trial Engineering, 135, 757–765. https://doi.org/10.1016/J.CIE.2019.06.052 Chen, K., & Yu, J. (2014). Short-term wind speed prediction using an unscented Kalman filter based state-space support vector regression approach. Applied Energy, 113, 690–705. https://doi.org/10.1016/j.apenergy.2013.08.025 Christensen, P., & Lund, H. (1998). Conflicting views of sustainability: The case of wind power and nature conservation in Denmark. European Environment, 8(1), 1–6. 87 https://doi.org/10.1002/(SICI)1099-0976(199801/02)8:1<1::AID- EET139>3.0.CO;2-T Colak, I., Sagiroglu, S., & Yesilbudak, M. (2012). Data mining and wind power prediction: A literature review. Renewable Energy, 46, 241–247. https://doi.org/10.1016/j.renene.2012.02.015 Crossan, F. (2003). Research philosophy: towards an understanding. Nurse Researcher, 11(1), 46–55. https://doi.org/10.7748/nr2003.10.11.1.46.c5914 Cui, Q., Liu, Y., Ali, T., Gao, J., & Chen, H. (2020). Economic and climate impacts of re- ducing China’s renewable electricity curtailment: A comparison between CGE models with alternative nesting structures of electricity. Energy Economics, 91. https://doi.org/10.1016/j.eneco.2020.104892 Dashwood, J. (2012). Australia ’ s Energy Options : Renewables and efficiency. In Aus- tralia’s Energy Options: Renewables and efficiency (pp. 45–52). De Giorgi, M. G., Ficarella, A., & Tarantino, M. (2011). Assessment of the benefits of numerical weather predictions in wind power forecasting based on statistical methods. Energy, 36(7), 3968–3978. https://doi.org/10.1016/j.energy.2011.05.006 Demir, N., & Taşkın, A. (2013). Life cycle assessment of wind turbines in Pınarbaşı- Kayseri. Journal of Cleaner Production, 54, 253–263. https://doi.org/10.1016/j.jclepro.2013.04.016 Demolli, H., Dokuz, A. S., Ecemis, A., & Gokcek, M. (2019). Wind power forecasting based on daily wind speed data using machine learning algorithms. Energy Con- version and Management, 198, 111823. https://doi.org/10.1016/J.ENCONMAN.2019.111823 Díaz-González, F., Hau, M., Sumper, A., & Gomis-Bellmunt, O. (2015). Coordinated op- eration of wind turbines and flywheel storage for primary frequency control support. International Journal of Electrical Power & Energy Systems, 68, 313– 326. https://doi.org/10.1016/j.ijepes.2014.12.062 88 Dihrab, S. S., & Sopian, K. (2010). Electricity generation of hybrid PV/wind systems in Iraq. Renewable Energy, 35(6), 1303–1307. https://doi.org/10.1016/j.renene.2009.12.010 Dincer, I. (2000). Renewable energy and sustainable development: a crucial review. Renewable and Sustainable Energy Reviews, 4(2), 157–175. https://doi.org/10.1016/S1364-0321(99)00011-8 Duan, Y., Edwards, J. S., & Dwivedi, Y. K. (2019). Artificial intelligence for decision mak- ing in the era of Big Data – evolution, challenges and research agenda. Interna- tional Journal of Information Management, 48, 63–71. https://doi.org/10.1016/j.ijinfomgt.2019.01.021 Dudovskiy, J. (2018). An Ultimate Guide to Writing A Dissertation in Business Studies A Step-by-Step Assistance. Research Methodology, Multiple. Retrieved from https://research-methodology.net Emeis, S. (2018). Wind Energy Meteorology: Atmospheric Physics for Wind Power Gen- eration - Stefan Emeis - Google Books. Retrieved April 14, 2021, from https://books.google.fi/books?hl=en&lr=&id=6r1TDwAAQBAJ&oi=fnd&pg=PR5 &dq=wind+formation+wind+energy+kinetic+energy&ots=tL0AtcgUsX&sig=DVN SFnTmKsqc5vWRu-OYaI4Q7Yk&redir_esc=y#v=onepage&q=wind formation wind energy kinetic energy&f=false Eriksson, S., Bernhoff, H., & Leijon, M. (2008). Evaluation of different turbine concepts for wind power. Renewable and Sustainable Energy Reviews, 12(5), 1419–1434. https://doi.org/10.1016/j.rser.2006.05.017 Fant, C., Adam Schlosser, C., & Strzepek, K. (2015). The impact of climate change on wind and solar resources in southern Africa. Applied Energy. https://doi.org/10.1016/j.apenergy.2015.03.042 Fertig, E., & Apt, J. (2011). Economics of compressed air energy storage to integrate wind power: A case study in ERCOT. Energy Policy, 39(5), 2330–2342. https://doi.org/10.1016/j.enpol.2011.01.049 89 Fiesler, E., Caulfield, H. J., Choudry, A., & Ryan, J. P. (1990). Maximum capacity topolo- gies for fully connected layered neural networks with bidirectional connections. 827–831. https://doi.org/10.1109/ijcnn.1990.137672 Flora, R., Marques, A. C., & Fuinhas, J. A. (2014). Wind power idle capacity in a panel of European countries. Energy, 66, 823–830. https://doi.org/10.1016/j.energy.2013.12.061 Friedman, J. H. (2001). Greedy Function Approximation: A Gradient Boosting Machine. Annals of Statistics, 29(5), 1189–1232. Retrieved from http://marefateadyan.nashriyat.ir/node/150 Gao, D., Jiang, D., Liu, P., Li, Z., Hu, S., & Xu, H. (2014). An integrated energy storage system based on hydrogen storage: Process configuration and case studies with wind power. Energy, 66, 332–341. https://doi.org/10.1016/j.energy.2014.01.095 Ge, B., Wang, W., Bi, D., Rogers, C. B., Peng, F. Z., de Almeida, A. T., & Abu-Rub, H. (2013). Energy storage system-based power control for grid-connected wind power farm. International Journal of Electrical Power & Energy Systems, 44(1), 115–122. https://doi.org/10.1016/j.ijepes.2012.07.021 Gil-García, I. C., García-Cascales, M. S., Fernández-Guillamón, A., & Molina-García, A. (2019). Categorization and analysis of relevant factors for optimal locations in onshore and offshorewind power plants: A taxonomic review. Journal of Marine Science and Engineering, 7(11), 1–21. https://doi.org/10.3390/jmse7110391 González-Aparicio, I., & Zucker, A. (2015). Impact of wind power uncertainty forecasting on the market integration of wind energy in Spain. Applied Energy, 159, 334– 349. https://doi.org/10.1016/j.apenergy.2015.08.104 González-Mingueza, C., & Muñoz-Gutiérrez, F. (2014). Wind prediction using Weather Research Forecasting model (WRF): A case study in Peru. Energy Conversion and Management, 81, 363–373. https://doi.org/10.1016/j.enconman.2014.02.024 Griffith, A. K. (1974). A comparison and evaluation of three machine learning proce- dures as applied to the game of checkers. Artificial Intelligence, 5(2), 137–148. https://doi.org/10.1016/0004-3702(74)90027-7 90 Güler, Ö. (2009). Wind energy status in electrical energy production of Turkey. Renewa- ble and Sustainable Energy Reviews, 13(2), 473–478. https://doi.org/10.1016/j.rser.2007.03.015 Guo, Z., Wu, J., Lu, H., & Wang, J. (2011). A case study on a hybrid wind speed forecast- ing method using BP neural network. Knowledge-Based Systems, 24(7), 1048– 1056. https://doi.org/10.1016/j.knosys.2011.04.019 Hamilton, S. H., Fu, B., Guillaume, J. H. A., Badham, J., Elsawah, S., Gober, P., … Zare, F. (2019). A framework for characterising and evaluating the effectiveness of envi- ronmental modelling. Environmental Modelling and Software, 118, 83–98. https://doi.org/10.1016/j.envsoft.2019.04.008 Han, S., Li, J., & Liu, Y. (2011). Tabu Search Algorithm Optimized ANN Model for Wind Power Prediction with NWP. Energy Procedia, 12, 733–740. https://doi.org/10.1016/j.egypro.2011.10.099 He, Y., Li, M., Meng, Z., Chen, S., Huang, S., Hu, Y., & Zou, X. (2021). An overview of acoustic emission inspection and monitoring technology in the key components of renewable energy systems. Mechanical Systems and Signal Processing, 148. https://doi.org/10.1016/j.ymssp.2020.107146 Hennessey, J. P. (1977). Some Aspects of Wind Power Statistics. Journal of Applied Me- teorology, 16(2), 119–128. https://doi.org/10.1175/1520- 0450(1977)016<0119:SAOWPS>2.0.CO;2 Hochreiter, S., & Schmidhuber, J. (1997). Long Short-Term Memory. Neural Computa- tion, 9(8), 1735–1780. https://doi.org/10.1162/neco.1997.9.8.1735 Holttinen, H. (2006). Handling of wind power forecast errors in the Nordic power mar- ket. 2006 International Conference on Probabilistic Methods Applied to Power Systems, 1–6. https://doi.org/10.1109/PMAPS.2006.360288 Holttinen, H., Miettinen, J., & Sillanpää, S. (2013). Wind power forecasting accuracy and uncertainty in Finland. Retrieved from http://www.vtt.fi/publications/index.js Hosseini-Firouz, M. (2013). Optimal offering strategy considering the risk management for wind power producers in electricity market. International Journal of Electri- 91 cal Power & Energy Systems, 49, 359–368. https://doi.org/10.1016/j.ijepes.2013.01.015 Hu, H., Wang, L., & Tao, R. (2021). Wind speed forecasting based on variational mode decomposition and improved echo state network. Renewable Energy, 164, 729– 751. https://doi.org/10.1016/j.renene.2020.09.109 Hu, Q., Zhang, S., Xie, Z., Mi, J., & Wan, J. (2014). Noise model based ν-support vector regression with its application to short-term wind speed forecasting. Neural Networks : The Official Journal of the International Neural Network Society, 57C, 1–11. https://doi.org/10.1016/j.neunet.2014.05.003 Hua, Y., Oliphant, M., & Hu, E. J. (2016). Development of renewable energy in Australia and China: A comparison of policies and status. Renewable Energy, 85, 1044– 1051. https://doi.org/10.1016/j.renene.2015.07.060 Huang, Q., Shi, Y., Wang, Y., Lu, L., & Cui, Y. (2015). Multi-turbine wind-solar hybrid sys- tem. Renewable Energy, 76, 401–407. https://doi.org/10.1016/j.renene.2014.11.060 Huang, X., Zhang, D., & Zhang, X. (2021). Energy management of intelligent building based on deep reinforced learning. Alexandria Engineering Journal, 60(1), 1509–1517. https://doi.org/10.1016/j.aej.2020.11.005 Huang, Z., & Chalabi, Z. S. (1995). Use of time-series analysis to model and forecast wind speed. Journal of Wind Engineering and Industrial Aerodynamics, 56(2–3), 311–322. https://doi.org/10.1016/0167-6105(94)00093-S Jacobsson, S., & Karltorp, K. (2012). Formation of competences to realize the potential of offshore wind power in the European Union. Energy Policy, 44, 374–384. https://doi.org/10.1016/j.enpol.2012.01.069 Jannati, M., Hosseinian, S. H., Vahidi, B., & Li, G. (2016). A significant reduction in the costs of battery energy storage systems by use of smart parking lots in the pow- er fluctuation smoothing process of the wind farms. Renewable Energy, 87, 1– 14. https://doi.org/10.1016/j.renene.2015.09.054 Jiang, P., Liu, Z., Niu, X., & Zhang, L. (2021). A combined forecasting system based on statistical method, artificial neural networks, and deep learning methods for 92 short-term wind speed forecasting. Energy, 217. https://doi.org/10.1016/j.energy.2020.119361 Jung, J., & Broadwater, R. P. (2014). Current status and future advances for wind speed and power forecasting. Renewable and Sustainable Energy Reviews, 31, 762– 777. https://doi.org/10.1016/j.rser.2013.12.054 Kaldellis, J. K., Kapsali, M., & Kavadias, K. A. (2010). Energy balance analysis of wind- based pumped hydro storage systems in remote island electrical networks. Ap- plied Energy, 87(8), 2427–2437. https://doi.org/10.1016/j.apenergy.2010.02.016 Kamal, L., & Jafri, Y. Z. (1997). Time series models to simulate and forecast hourly aver- aged wind speed in Quetta, Pakistan. Solar Energy, 61(1), 23–32. https://doi.org/10.1016/S0038-092X(97)00037-6 Kamel, R. M., Chaouachi, A., & Nagasaka, K. (2010). Wind power smoothing using fuzzy logic pitch controller and energy capacitor system for improvement Micro-Grid performance in islanding mode. Energy, 35(5), 2119–2129. https://doi.org/10.1016/j.energy.2010.01.030 Kandpal, T. C., & Broman, L. (2014). Renewable energy education: A global status re- view. Renewable and Sustainable Energy Reviews, 34, 300–324. https://doi.org/10.1016/j.rser.2014.02.039 Kariniotakis, G. N., Stavrakakis, G. S., & Nogaret, E. F. (1996). Wind power forecasting using advanced neural networks models. IEEE Transactions on Energy Conver- sion, 11(4), 762–767. https://doi.org/10.1109/60.556376 Kavasseri, R. G., & Seetharaman, K. (2009). Day-ahead wind speed forecasting using f- ARIMA models. Renewable Energy, 34(5), 1388–1393. https://doi.org/10.1016/j.renene.2008.09.006 Kenabatho, P. K., Parida, B. P., & Moalafhi, D. B. (2012). The value of large-scale climate variables in climate change assessment: The case of Botswana’s rainfall. Physics and Chemistry of the Earth, Parts A/B/C, 50–52, 64–71. https://doi.org/10.1016/j.pce.2012.08.006 93 Khalid, M., & Savkin, A. V. (2010). A model predictive control approach to the problem of wind power smoothing with controlled battery storage. Renewable Energy, 35(7), 1520–1526. https://doi.org/10.1016/j.renene.2009.11.030 Khalid, M., & Savkin, A. V. (2014). Minimization and control of battery energy storage for wind power smoothing: Aggregated, distributed and semi-distributed stor- age. Renewable Energy, 64, 105–112. https://doi.org/10.1016/j.renene.2013.09.043 Kimura, Y., Onai, Y., & Ushiyama, I. (1996). A demonstrative study for the wind and solar hybrid power system. Renewable Energy, 9(1–4), 895–898. https://doi.org/10.1016/0960-1481(96)88424-2 Klir, G. J., & Folger, T. A. (1988). Fuzzy Sets, Uncertainty and Information [Hardcover]. Retrieved from http://www.amazon.com/Fuzzy-Sets-Uncertainty-Information- George/dp/0133459845/ref=sr_1_1?ie=UTF8&qid=1403952802&sr=8- 1&keywords=Fuzzy+sets,+uncertainty,+and+information Kothari, C. R. (2004). Research Methodology: Methods and Techniques. Retrieved from https://books.google.fi/books?id=8c6gkbKi-F4C Laberge, Y. (2011). Advising on Research Methods: A Consultant’s Companion. Journal of Applied Statistics, 38(12), 2991–2991. https://doi.org/10.1080/02664763.2011.559375 LeCun, Y., Bengio, Y., & Hinton, G. (2015). Deep learning. Nature, 521(7553), 436–444. https://doi.org/10.1038/nature14539 Lee, K. M., Yoo, J., Kim, S.-W., Lee, J.-H., & Hong, J. (2019). Autonomic machine learning platform. International Journal of Information Management, 49, 491–501. https://doi.org/10.1016/J.IJINFOMGT.2019.07.003 Lei, M., Shiyan, L., Chuanwen, J., Hongling, L., & Yan, Z. (2009). A review on the fore- casting of wind speed and generated power. Renewable and Sustainable Energy Reviews, 13(4), 915–920. https://doi.org/10.1016/j.rser.2008.02.002 Lim, S., & Chi, S. (2019). Xgboost application on bridge management systems for proac- tive damage estimation. Advanced Engineering Informatics, 41(October 2018), 100922. https://doi.org/10.1016/j.aei.2019.100922 94 Lin, J., Sun, Y., Cheng, L., & Gao, W. (2012). Assessment of the power reduction of wind farms under extreme wind condition by a high resolution simulation model. Ap- plied Energy, 96, 21–32. https://doi.org/10.1016/j.apenergy.2011.10.028 Liu, H. of two new A.-A. and A.-K. hybrid methods for wind speed prediction, Tian, H., & Li, Y. (2012). Comparison of two new ARIMA-ANN and ARIMA-Kalman hybrid methods for wind speed prediction. Applied Energy, 98, 415–424. https://doi.org/10.1016/j.apenergy.2012.04.001 Liu, P., & Li, Y. (2021). An improved failure mode and effect analysis method for multi- criteria group decision-making in green logistics risk assessment. Reliability En- gineering & System Safety, 215(May), 107826. https://doi.org/10.1016/j.ress.2021.107826 Liu, Y., Yu, S., Zhu, Y., Wang, D., & Liu, J. (2018). Modeling, planning, application and management of energy systems for isolated areas: A review. Renewable and Sustainable Energy Reviews, 82(June 2017), 460–470. https://doi.org/10.1016/j.rser.2017.09.063 Louka, P., Galanis, G., Siebert, N., Kariniotakis, G., Katsafados, P., Pytharoulis, I., & Kallos, G. (2008). Improvements in wind speed forecasts for wind power prediction purposes using Kalman filtering. Journal of Wind Engineering and Industrial Aerodynamics, 96(12), 2348–2362. https://doi.org/10.1016/j.jweia.2008.03.013 Lu, L., Yang, H., & Burnett, J. (2002). Investigation on wind power potential on Hong Kong islands—an analysis of wind power and wind turbine characteristics. Re- newable Energy, 27(1), 1–12. https://doi.org/10.1016/S0960-1481(01)00164-1 Lund, H. (2007). Renewable energy strategies for sustainable development. Energy, 32(6), 912–919. https://doi.org/10.1016/j.energy.2006.10.017 Ma, T., Yang, H., Lu, L., & Peng, J. (2014). Optimal design of an autonomous solar–wind- pumped storage power supply system. Applied Energy. https://doi.org/10.1016/j.apenergy.2014.11.026 Maleki, A., & Askarzadeh, A. (2014). Optimal sizing of a PV/wind/diesel system with battery storage for electrification to an off-grid remote region: A case study of 95 Rafsanjan, Iran. Sustainable Energy Technologies and Assessments, 7, 147–153. https://doi.org/10.1016/j.seta.2014.04.005 Mannig, B., Müller, M., Starke, E., Merkenschlager, C., Mao, W., Zhi, X., … Paeth, H. (2013). Dynamical downscaling of climate change in Central Asia. Global and Planetary Change, 110, 26–39. https://doi.org/10.1016/j.gloplacha.2013.05.008 Maouedj, R., Mammeri, A., Draou, M. D., & Benyoucef, B. (2014). Performance Evalua- tion of Hybrid Photovoltaic-wind Power Systems. Energy Procedia, 50, 797–807. https://doi.org/10.1016/j.egypro.2014.06.098 Martínez, E., Sanz, F., Pellegrini, S., Jiménez, E., & Blanco, J. (2009). Life cycle assess- ment of a multi-megawatt wind turbine. Renewable Energy, 34(3), 667–673. https://doi.org/10.1016/j.renene.2008.05.020 McVicar, T. R., Roderick, M. L., Donohue, R. J., Li, L. T., Van Niel, T. G., Thomas, A., … Dinpashoh, Y. (2012). Global review and synthesis of trends in observed terres- trial near-surface wind speeds: Implications for evaporation. Journal of Hydrol- ogy, 416–417, 182–205. https://doi.org/10.1016/j.jhydrol.2011.10.024 Mens, M., Minnema, B., Overmars, K., & van den Hurk, B. (2021). Dilemmas in develop- ing models for long-term drought risk management: The case of the National Water Model of the Netherlands. Environmental Modelling and Software, 143, 105100. https://doi.org/10.1016/j.envsoft.2021.105100 Mesinger, F., DiMego, G., Kalnay, E., Mitchell, K., Shafran, P. C., Ebisuzaki, W., … Shi, W. (2010). North American Regional Reanalysis. Retrieved from http://journals.ametsoc.org/doi/abs/10.1175/BAMS-87-3-343 Metternicht, G. (2001). Assessing temporal and spatial changes of salinity using fuzzy logic, remote sensing and GIS. Foundations of an expert system. Ecological Modelling, 144(2–3), 163–179. https://doi.org/10.1016/S0304-3800(01)00371- 4 Mohamed, A. Z., Eskander, M. N., & Ghali, F. A. (2001). Fuzzy logic control based maxi- mum power tracking of a wind energy system. Renewable Energy, 23(2), 235– 245. https://doi.org/10.1016/S0960-1481(00)00099-9 96 Mohandes, M. A., Halawani, T. O., Rehman, S., & Hussain, A. A. (2004). Support vector machines for wind speed prediction. Renewable Energy, 29(6), 939–947. https://doi.org/10.1016/j.renene.2003.11.009 Murage, M. W., & Anderson, C. L. (2014). Contribution of pumped hydro storage to integration of wind power in Kenya: An optimal control approach. Renewable Energy, 63, 698–707. https://doi.org/10.1016/j.renene.2013.10.026 Necefer, L., Wong-Parodi, G., Small, M. J., & Begay-Campbell, S. (2018). Integrating technical, economic and cultural impacts in a decision support tool for energy resource management in the Navajo Nation. Energy Strategy Reviews, 22(May 2017), 136–146. https://doi.org/10.1016/j.esr.2018.06.004 Novak, P., Ekelund, T., Jovik, I., & Schmidtbauer, B. (1995). Modeling and control of var- iable-speed wind-turbine drive-system dynamics. IEEE Control Systems Magazine, 15(4), 28–38. https://doi.org/10.1109/37.408463 Nygaard, I., Rasmussen, K., Badger, J., Nielsen, T. T., Hansen, L. B., Stisen, S., … Togola, I. (2010). Using modeling, satellite images and existing global datasets for rapid preliminary assessments of renewable energy resources: The case of Mali. Re- newable and Sustainable Energy Reviews, 14(8), 2359–2371. https://doi.org/10.1016/j.rser.2010.04.001 Olaofe, Z. O. (2014). A 5-day wind speed & power forecasts using a layer recurrent neu- ral network (LRNN). Sustainable Energy Technologies and Assessments, 6, 1–24. https://doi.org/10.1016/j.seta.2013.12.001 On the Vertical Extrapolation of Mean Wind Power DensitySedefian, L. (1980). On the Vertical Extrapolation of Mean Wind Power Density. Journal of Applied Meteor- ology, 19(4), 488–493. https://doi.org/10.1175/1520- 0450(1980)019<0488:OTVEOM>2.0.CO;2 Oñederra, O., Asensio, F. J., Saldaña, G., Martín, J. I. S., & Zamora, I. (2020). Wind ener- gy harnessing in a railway infrastructure: Converter topology and control pro- posal. Electronics (Switzerland), 9(11), 1–14. https://doi.org/10.3390/electronics9111943 97 Ortegon, K., Nies, L. F., & Sutherland, J. W. (2013). Preparing for end of service life of wind turbines. Journal of Cleaner Production, 39, 191–199. https://doi.org/10.1016/j.jclepro.2012.08.022 Ortiz-García, E. G., Salcedo-Sanz, S., Pérez-Bellido, Á. M., Gascón-Moreno, J., Portilla- Figueras, J. A., & Prieto, L. (2011). Short-term wind speed prediction in wind farms based on banks of support vector machines. Wind Energy, 14(2), 193–207. https://doi.org/10.1002/we.411 Ozcan, M. (2014). Assessment of renewable energy incentive system from investors’ perspective. Renewable Energy, 71, 425–432. https://doi.org/10.1016/j.renene.2014.05.053 Peng, H., Liu, F., & Yang, X. (2013). A hybrid strategy of short term wind power predic- tion. Renewable Energy, 50, 590–595. https://doi.org/10.1016/j.renene.2012.07.022 Pereira, E. B., Martins, F. R., Pes, M. P., da Cruz Segundo, E. I., & Lyra, A. de A. (2013). The impacts of global climate changes on the wind power density in Brazil. Re- newable Energy, 49, 107–110. https://doi.org/10.1016/j.renene.2012.01.053 Persson, A. (1991). Kalman filtering a new approach to adaptive statistical interpreta- tion of numerical meteorological forecasts. WMO Technical Document, (421). Pinson, P., & Kariniotakis, G. (2004). On-line assessment of prediction risk for wind power production forecasts. Wind Energy, 7(2), 119–132. https://doi.org/10.1002/we.114 Poncela, M., Poncela, P., & Perán, J. R. (2013). Automatic tuning of Kalman filters by maximum likelihood methods for wind energy forecasting. Applied Energy, 108, 349–362. https://doi.org/10.1016/j.apenergy.2013.03.041 Pryor, S. C., & Barthelmie, R. J. (2010). Climate change impacts on wind energy: A re- view. Renewable and Sustainable Energy Reviews, 14(1), 430–437. https://doi.org/10.1016/j.rser.2009.07.028 Rezaeian-Marjani, S., Masoumzadehasl, S., Galvani, S., & Talavat, V. (2020). Probabilistic assessment of D-STATCOM operation considering correlated uncertain variables. 98 International Transactions on Electrical Energy Systems, 30(10). https://doi.org/10.1002/2050-7038.12522 Rienecker, M. M., Suarez, M. J., Gelaro, R., Todling, R., Bacmeister, J., Liu, E., … Woollen, J. (2011). MERRA: NASA’s Modern-Era Retrospective Analysis for Research and Applications. Journal of Climate, 24(14), 3624–3648. https://doi.org/10.1175/JCLI-D-11-00015.1 Robert Vautard, Françoise Thais, Isabelle Tobin, François-Marie Bréon, Jean-Guy Devezeaux de Lavergne, Augustin Colette, P. Y. & P. M. R. (2014). Time series analysis : forecasting and control: George E. P. Jenkins, Gwilym M., Box: Ama- zon.com: Books. Rose, S., & Apt, J. (2015). What can reanalysis data tell us about wind power? Renewa- ble Energy, 83, 963–969. https://doi.org/10.1016/j.renene.2015.05.027 Rudenko, N. V., Ershov, V. V., & Evstafiev, V. V. (2017). The Environment Friendly Power Source for Power Supply of Mobile Communication Base Stations. IOP Confer- ence Series: Earth and Environmental Science, 66(1). https://doi.org/10.1088/1755-1315/66/1/012024 Saeed, M. A., Ahmed, Z., & Zhang, W. (2020). Wind energy potential and economic analysis with a comparison of different methods for determining the optimal distribution parameters. Renewable Energy, 161, 1092–1109. https://doi.org/https://doi.org/10.1016/j.renene.2020.07.064 Saha, S., Moorthi, S., Pan, H.-L., Wu, X., Wang, J., Nadiga, S., … Goldberg, M. (2010). The NCEP Climate Forecast System Reanalysis. Bulletin of the American Meteorolog- ical Society, 91(8), 1015–1057. https://doi.org/10.1175/2010BAMS3001.1 Sailor, D. J., Smith, M., & Hart, M. (2008). Climate change implications for wind power resources in the Northwest United States. Renewable Energy, 33(11), 2393– 2406. https://doi.org/10.1016/j.renene.2008.01.007 Salcedo-Sanz, S., Ortiz-García, E. G., Portilla-Figueras, A., Prieto, L., & Paredes, D. (2009). Hybridizing the fifth generation mesoscale model with artificial neural networks for short-term wind speed prediction. Renewable Energy, 34(6), 1451–1457. https://doi.org/10.1016/j.renene.2008.10.017 99 Saleh Asheghabadi, M., Sahafnia, M., Bahadori, A., & Bakhshayeshi, N. (2019). Seismic behavior of suction caisson for offshore wind turbine to generate more renew- able energy. International Journal of Environmental Science and Technology, 16(7), 2961–2972. https://doi.org/10.1007/s13762-018-2150-8 Sample, J. E., Duncan, N., Ferguson, M., & Cooksley, S. (2015). Scotland ׳s hydropower: Current capacity, future potential and the possible impacts of climate change. Renewable and Sustainable Energy Reviews, 52, 111–122. https://doi.org/10.1016/j.rser.2015.07.071 Sarrias-Mena, R., Fernández-Ramírez, L. M., García-Vázquez, C. A., & Jurado, F. (2014). Fuzzy logic based power management strategy of a multi-MW doubly-fed induc- tion generator wind turbine with battery and ultracapacitor. Energy, 70, 561– 576. https://doi.org/10.1016/j.energy.2014.04.049 Satkin, M., Noorollahi, Y., Abbaspour, M., & Yousefi, H. (2014). Multi criteria site selec- tion model for wind-compressed air energy storage power plants in Iran. Re- newable and Sustainable Energy Reviews, 32, 579–590. https://doi.org/10.1016/j.rser.2014.01.054 Schleisner, L. (2000). Life cycle assessment of a wind farm and related externalities. Renewable Energy, 20(3), 279–288. https://doi.org/10.1016/S0960- 1481(99)00123-8 Sebastián, R., & Peña-Alzola, R. (2015). Control and simulation of a flywheel energy storage for a wind diesel power system. International Journal of Electrical Power & Energy Systems, 64, 1049–1056. https://doi.org/10.1016/j.ijepes.2014.08.017 Seljom, P., Rosenberg, E., Fidje, A., Haugen, J. E., Meir, M., Rekstad, J., & Jarlset, T. (2011). Modelling the effects of climate change on the energy system—A case study of Norway. Energy Policy, 39(11), 7310–7321. https://doi.org/10.1016/j.enpol.2011.08.054 Shamshirband, S., Keivani, A., Mohammadi, K., Lee, M., Hamid, S. H. A., & Petkovic, D. (2016). Assessing the proficiency of adaptive neuro-fuzzy system to estimate wind power density: Case study of Aligoodarz, Iran. Renewable and Sustainable Energy Reviews, 59, 429–435. https://doi.org/10.1016/j.rser.2015.12.269 100 Sharma, N., Sharma, R., & Jindal, N. (2021). Machine Learning and Deep Learning Ap- plications-A Vision. Global Transitions Proceedings, 0–8. https://doi.org/10.1016/j.gltp.2021.01.004 Shen, N., Wang, Y., Peng, H., & Hou, Z. (2020). Renewable energy green innovation, fossil energy consumption, and air pollution-spatial empirical analysis based on China. Sustainability (Switzerland), 12(16). https://doi.org/10.3390/SU12166397 Shen, S., Sadoughi, M., Chen, X., Hong, M., & Hu, C. (2019). A deep learning method for online capacity estimation of lithium-ion batteries. Journal of Energy Storage, 25, 100817. https://doi.org/10.1016/J.EST.2019.100817 Sheng Shen, Mohammadkazem Sadoughi, Xiangyi Chen, Mingyi Hong, C. H. (2019). IEEE Xplore: IEEE Transactions on Pattern Analysis and Machine Intelligence. Re- trieved November 1, 2019, from Journal of Energy Storage website: https://ieeexplore.ieee.org/xpl/RecentIssue.jsp?punumber=34 Shi, J., Guo, J., & Zheng, S. (2012). Evaluation of hybrid forecasting approaches for wind speed and power generation time series. Renewable and Sustainable Energy Reviews, 16(5), 3471–3480. https://doi.org/10.1016/j.rser.2012.02.044 Shin, D., Ha, E., Kim, T., & Kim, C. (2021). Short-term photovoltaic power generation predicting by input/output structure of weather forecast using deep learning. Soft Computing, 25(1), 771–783. https://doi.org/10.1007/s00500-020-05199-7 Shin, Y., Koo, W. Y., Kim, T. H., Jung, S., & Kim, H. (2015). Capacity design and operation planning of a hybrid PV–wind–battery–diesel power generation system in the case of Deokjeok Island. Applied Thermal Engineering, 89, 514–525. https://doi.org/10.1016/j.applthermaleng.2015.06.043 Shukur, O. B., & Lee, M. H. (2015). Daily wind speed forecasting through hybrid KF-ANN model based on ARIMA. Renewable Energy, 76, 637–647. https://doi.org/10.1016/j.renene.2014.11.084 Simoes, M. G., Bose, B. K., & Spiegel, R. J. (1997a). Design and performance evaluation of a fuzzy-logic-based variable-speed wind generation system. IEEE Transactions on Industry Applications, 33(4), 956–965. https://doi.org/10.1109/28.605737 101 Simoes, M. G., Bose, B. K., & Spiegel, R. J. (1997b). Fuzzy logic based intelligent control of a variable speed cage machine wind generation system. IEEE Transactions on Power Electronics, 12(1), 87–95. https://doi.org/10.1109/63.554173 Soman, S. S., Zareipour, H., Malik, O., & Mandal, P. (2010). A review of wind power and wind speed forecasting methods with different time horizons. North American Power Symposium 2010, 1–8. https://doi.org/10.1109/NAPS.2010.5619586 Soukissian, T. H., & Papadopoulos, A. (2015). Effects of different wind data sources in offshore wind power assessment. Renewable Energy, 77, 101–114. https://doi.org/10.1016/j.renene.2014.12.009 Staid, A., & Guikema, S. D. (2013). Statistical analysis of installed wind capacity in the United States. Energy Policy, 60, 378–385. https://doi.org/10.1016/j.enpol.2013.05.076 Swider, D. J. (2007). Compressed Air Energy Storage in an Electricity System With Signif- icant Wind Power Generation. IEEE Transactions on Energy Conversion, 22(1), 95–102. https://doi.org/10.1109/TEC.2006.889547 Taraft, S., Rekioua, D., & Aouzellag, D. (2013). Wind Power Control System Associated to the Flywheel Energy Storage System Connected to the Grid. Energy Procedia, 36, 1147–1157. https://doi.org/10.1016/j.egypro.2013.07.130 Taylor, J. W., & Buizza, R. (2002). Neural network load forecasting with weather ensem- ble predictions. IEEE Transactions on Power Systems, 17(3), 626–632. https://doi.org/10.1109/TPWRS.2002.800906 Tong, Z., Cheng, Z., & Tong, S. (2021). A review on the development of compressed air energy storage in China: Technical and economic challenges to commercializa- tion. Renewable and Sustainable Energy Reviews, 135(September 2019), 110178. https://doi.org/10.1016/j.rser.2020.110178 Trenberth Kevin E. (1990). Recent Observed Interdecadal Climate Changes in the Northern Hemisphere. Bulletin of the American Meteorological Society, 71(7), 988–993. Trianni, A., Cagno, E., Bertolotti, M., Thollander, P., & Andersson, E. (2019). Energy management : A practice-based assessment model. Applied Energy, 102 235(December 2018), 1614–1636. https://doi.org/10.1016/j.apenergy.2018.11.032 Ucar, A., & Balo, F. (2010). Assessment of wind power potential for turbine installation in coastal areas of Turkey. Renewable and Sustainable Energy Reviews, 14(7), 1901–1912. https://doi.org/10.1016/j.rser.2010.03.021 Uddin, M. S., & Kumar, S. (2014). Energy, emissions and environmental impact analysis of wind turbine using life cycle assessment technique. Journal of Cleaner Pro- duction, 69, 153–164. https://doi.org/10.1016/j.jclepro.2014.01.073 Uppala, S. M., KÅllberg, P. W., Simmons, A. J., Andrae, U., Bechtold, V. D. C., Fiorino, M., … Woollen, J. (2005). The ERA-40 re-analysis. Quarterly Journal of the Royal Meteorological Society, 131(612), 2961–3012. https://doi.org/10.1256/qj.04.176 Vijay M, D., Singh, B., & Bhuvaneswari, G. (2020). A High-Performance Microgrid with a Mechanical Sensorless SynRG Operated Wind Energy Generating System. IEEE Transactions on Industrial Informatics, 16(12), 7349–7359. https://doi.org/10.1109/TII.2020.2976906 Wang, B., Ke, R.-Y., Yuan, X.-C., & Wei, Y.-M. (2014). China ׳s regional assessment of re- newable energy vulnerability to climate change. Renewable and Sustainable En- ergy Reviews, 40, 185–195. https://doi.org/10.1016/j.rser.2014.07.154 Wang, J., Qin, S., Zhou, Q., & Jiang, H. (2015). Medium-term wind speeds forecasting utilizing hybrid models for three different sites in Xinjiang, China. Renewable Energy, 76, 91–101. https://doi.org/10.1016/j.renene.2014.11.011 Wang, P., Fan, E., & Wang, P. (2021). Comparative analysis of image classification algo- rithms based on traditional machine learning and deep learning. Pattern Recog- nition Letters, 141, 61–67. https://doi.org/10.1016/j.patrec.2020.07.042 Wang, R., Li, C., Fu, W., & Tang, G. (2020). Deep Learning Method Based on Gated Re- current Unit and Variational Mode Decomposition for Short-Term Wind Power Interval Prediction. IEEE Transactions on Neural Networks and Learning Systems, 31(10), 3814–3827. https://doi.org/10.1109/TNNLS.2019.2946414 103 Wang, X. Y., Mahinda Vilathgamuwa, D., & Choi, S. S. (2008). Determination of Battery Storage Capacity in Energy Buffer for Wind Farm. IEEE Transactions on Energy Conversion, 23(3), 868–878. https://doi.org/10.1109/TEC.2008.921556 Wu, J., Zhang, B., Li, H., Li, Z., Chen, Y., & Miao, X. (2014). Statistical distribution for wind power forecast error and its application to determine optimal size of ener- gy storage system. International Journal of Electrical Power & Energy Systems, 55, 100–107. https://doi.org/10.1016/j.ijepes.2013.09.003 Xydis, G. (2013). On the exergetic capacity factor of a wind – solar power generation system. Journal of Cleaner Production, 47, 437–445. https://doi.org/10.1016/j.jclepro.2012.07.014 Yang, H., Lu, L., & Zhou, W. (2007). A novel optimization sizing model for hybrid solar- wind power generation system. Solar Energy, 81(1), 76–84. https://doi.org/10.1016/j.solener.2006.06.010 Yang, M., Patiño-Echeverri, D., & Yang, F. (2012). Wind power generation in China: Un- derstanding the mismatch between capacity and generation. Renewable Energy, 41, 145–151. https://doi.org/10.1016/j.renene.2011.10.013 Zahedi, A. (2010). Australian renewable energy progress. Renewable and Sustainable Energy Reviews, 14(8), 2208–2213. https://doi.org/10.1016/j.rser.2010.03.026 Zahedi, Ahmad. (2014). Sustainable Power Supply Using Solar Energy and Wind Power Combined with Energy Storage. Energy Procedia, 52, 642–650. https://doi.org/10.1016/j.egypro.2014.07.120 Zhang, D., Qian, L., Mao, B., Huang, C., Huang, B., & Si, Y. (2018). A Data-Driven Design for Fault Detection of Wind Turbines Using Random Forests and XGboost. IEEE Access, 6, 21020–21031. https://doi.org/10.1109/ACCESS.2018.2818678 Zhang, J., Chowdhury, S., & Messac, A. (2014). A comprehensive measure of the energy resource: Wind power potential (WPP). Energy Conversion and Management, 86, 388–398. https://doi.org/10.1016/j.enconman.2014.04.083 Zhang, Y., Pan, G., Zhao, Y., Li, Q., & Wang, F. (2020). Short-term wind speed interval prediction based on artificial intelligence methods and error probability distri- 104 bution. Energy Conversion and Management, 224(June), 113346. https://doi.org/10.1016/j.enconman.2020.113346 Zhao, P., Dai, Y., & Wang, J. (2014). Design and thermodynamic analysis of a hybrid en- ergy storage system based on A-CAES (adiabatic compressed air energy storage) and FESS (flywheel energy storage system) for wind power application. Energy, 70, 674–684. https://doi.org/10.1016/j.energy.2014.04.055 Zhao, P., Wang, M., Wang, J., & Dai, Y. (2015). A preliminary dynamic behaviors analysis of a hybrid energy storage system based on adiabatic compressed air energy storage and flywheel energy storage system for wind power application. Energy, 84, 825–839. https://doi.org/10.1016/j.energy.2015.03.067 Zhao, X., Wang, C., Su, J., & Wang, J. (2019). Research and application based on the swarm intelligence algorithm and artificial intelligence for wind farm decision system. Renewable Energy, 134, 681–697. https://doi.org/10.1016/j.renene.2018.11.061 Zhao, Z.-Y., Zhu, J., & Zuo, J. (2014). Sustainable development of the wind power indus- try in a complex environment: a flexibility study. Energy Policy, 75, 392–397. https://doi.org/10.1016/j.enpol.2014.10.017 Zhou, J., Shi, J., & Li, G. (2011). Fine tuning support vector machines for short-term wind speed forecasting. Energy Conversion and Management, 52(4), 1990–1998. https://doi.org/10.1016/j.enconman.2010.11.007 Zhou, W., Yang, H., & Fang, Z. (2006). Wind power potential and characteristic analysis of the Pearl River Delta region, China. Renewable Energy, 31(6), 739–753. https://doi.org/10.1016/j.renene.2005.05.006 105 Appendices Appendix 1. Paper 1: A Review of Innovation in Wind Power Forecast Yang, W., Liu, Y., and Yang, G. (2014): A review of innovation in wind power forecast. Proceedings of the 11th International conference on Innovation & Management. It has been published in Proceedings of the 11th International conference on Innovation & Management. Wenshan Yang: Defining of the research design, methodology and research methods, verification and analysis of findings, writing the body text of the original article, editing the article at different stages. Yang Liu: Editing the article at different stages. Gengsheng Yang: Acquisition of research resources. 106 A Review of Innovation in Wind Power Forecast Wenshan YANG1,2, Yang LIU*1, Gengsheng YANG2 1 Department of Production, University of Vaasa, Vaasa, Finland, 65200 2 Hubei Meteorological Bureau, Wuhan, P. R. China, 430074 (E-mail: yangw@student.uwasa.fi, yli@uva.fi, yanggengsheng11@163.com) Abstract It can be noticed that most existing literature related to the topic of wind energy resource focus on specif- ic areas, such as, specific forecasting model, local energy plan and policy, power engineering, etc. The purpose of this paper is to contribute to the topic of wind resource exploitation and wind power prediction for relevant researchers and professionals. A review of existing literature in the areas of wind energy re- source (WER) and wind power forecast (WPF) methods are presented in this paper, and the innovations in these relevant areas are also reviewed. We try to explore the research gaps in these areas to highlight the possible future research topics for the society. Keywords wind energy resource, wind power forecast, weather forecast, literature review 1 Introduction In the last few decades, numerous researchers have put effort on exploiting, utilizing and optimizing en- ergy. However, in recent years a large number of countries are moving to the exploitation of renewable and clean energy and this will be a long-term trend. Meanwhile, renewable energy education is becoming more and more popular and plays an important role in the improving of the quality of life (Kandpal & Broman, 2014). In contrary to fossil fuel based and nuclear energy sources, renewable energy sources (RES) can effec- tively utilize natural resources, alleviate the pressure of energy crisis, and minimize the negative envi- ronmental impacts (Ozcan, 2014). A new research predicts that global energy demand in 2040 will be approximately 30% higher than it was in 2010. It is urgent that. Because the typical characteristics of wind energy are stochastic and intermittent, it is important to know and use appropriate renewable energy technologies in the whole process of producing wind power generation (Dashwood, 2012). Changes in temperature, precipitation, sea level, and the frequency and severity of extreme events will likely affect how much wind power generation is produced, delivered, and consumed. One hand, weather forecast data as input of WPF system influence a lot. On the other hand, there exist various weather phe- nomena, such as, rainstorm, hail, thunderstorm and tornado, have high probability generate more or less damage to wind turbines. The structure of the rest of the paper is as follows: Section 2 describes the methodology of writing a liter- ature review upon which this paper is based and introduces some other methods. Section 3 reviews the major contribution and innovation of all the existing classic models in the wind power forecast and their links between in order to find research gaps. Section 4 discusses the research gaps and proposes for the future research. Section 5 concludes the paper and some final remarks. 2 Research methodology 107 The main methodology of this review paper is systematic review. The purpose of using this method is to find what kinds of advanced outcome did researchers have done before and specify the research gap, then determine what research content remain to study. Meanwhile, it can also provide a blueprint of state-of-art. The systematic review is a quite common way of collecting data, published in the literature, assessing methodological quality for high quality research questions. Systematic review are quite common in sci- ences where data are collected, published in the literature, and an assessment of methodological quality for a precisely defined subject would be helpful (Laberge, 2011). In this paper, contribution of each litera- ture were associate and analyze together, it aims at finding out advanced things filling out the research gap. Besides of this, cluster analysis helped to classify wind power forecast (WPF) methods. Furthermore, the methods of association analysis has also been used to do a synthesized analysis. Through a review of evidence from both qualitative and quantitative studies, disparate data are synthesized in order to better understand the topic of WER and WPF. 3 Descriptive analysis 3.1 Wind power forecasting by time scale People usually use ultra-short term, short-term, medium-term and long-term time scales to predict wind power prediction in practice (Soman et al., 2010). There are different time scales when classifying wind power forecasting species according to time periods and one example is as follows (Peng et al., 2013; Soman et al., 2010). a. Long term forecasting Long-term wind power predictions are utilized for maintenance and repair of the wind turbine and include the predictions from 1 day to 1 week. b. Medium term forecasting Medium-term wind power predictions are utilized for power system management and energy trading and include the predictions for 6 h to 1 day ahead. c. Short term forecasting Short-term wind power predictions are utilized for pre-load sharing and include the predictions from 30 min to 6 h. d. Ultra-short term forecasting Very short-term wind power predictions are used for turbine control and load tracking and in- clude the predictions for few seconds to 30 min ahead. 3.2 Wind power forecasting by predicting model The methods of wind power prediction are usually divided into three groups (González-Mingueza & Muñoz-Gutiérrez, 2014). ● Statistic model Statistical models are based on mathematical statistics analysis of the main variables associated with energy generation, such as wind speed and temperature at some points of measurement as well as the measures of wind generation at different point in the network. ● Physical model 108 Physical models are based on the use of numerical models. At first, get the results of meteorological data, such as, wind direction, wind speed, atmospheric pressure and air density. Then use some physical char- acteristics surrounding wind turbine to get the optimized predicting wind speed and direction in different hub height of wind turbine generator system. Finally, the wind power output can be predicted based on established physical model. ● Hybrid method Hybrid method is a useful predicting way as it can improve the WPF accuracy by offsetting random error with one method from each other. However, this technique is not very mature even it is commonly used nowadays. 3.3 Commonly used wind power forecasting methods In light of different input data which means whether use Numerical weather forecast (NWF), the wind power forecasting can be divided into numerical weather forecast forecasting method and historical mete- orological data forecasting method. ● Kalman filters Kalman filters (KF) is an optimal recursive data processing algorithm and it has been firstly achieved by Stanley Schmidt in 1958. The preliminary application of KF method for numerical weather forecasting has been reported in a few papers (Persson, 1991). In one case, a modified KF algorithms was applied to wind speed numerical predictions so as to improve the WPF accuracy. This literature indicates high per- formance in eliminating of any type of systematic errors and reducing the requirements in CPU time. In the end, this paper also mentions that this technique can not only be used in the traditional meteorological use but also engineering sector, such as, wind power integration (Louka et al., 2008). ● Time series model Commonly used time series models include auto regressive (AR), moving average (MA), autoregressive moving-average model (ARMA) and auto regressive integrated moving average (ARIMA). Among these, one ARIMA model established by Box and Jenkins have been widely used for the purpose of time series forecasting (Box & Jenkins, 1976). Meanwhile, this book is extremely overall because it interpretes each kind of time models in detail and also gives samples of modelling. ● Artificial neural network (ANN) Various artificial neural network (ANN) models are widely used, such as back propagation (BP) and radial basis function (RBF). An ANN is an information processing method which works like a human brain processes to find an algorithmic solution algorithmic solution in order to pick out the structure from the existing data (Carolin Mabel & Fernandez, 2008; Kariniotakis, Stavrakakis, & Nogaret, 1996). Based on normal BP network, one new wind power prediction model which optimized tabu search algorithm with memory function was developed (Han et al., 2011). Existing methods for this purpose tend to yield results with poor accuracy because they cannot properly account for seasonal effects over the long term. However, one updated method improve the accuracy of daily average wind speed forecasting. This study aims to forecast the daily average wind speed over a 109 long period of time, such as one year ahead. This method can forecast the daily average wind speed one year ahead with lower mean absolute errors compared to figures obtained without adjustment (Guo et al., 2011). ● Support vector machine (SVM) This method was firstly developed by Corinna Cortes and Vapnik in 1995. It is similar to ANN but the most apparent difference is SVM focus on mathematic method and optimization mechanism. One typical research applied this SVM method to wind speed prediction has been done in 2004. The paper introduces SVM, the latest neural network algorithm, to wind speed prediction and compares their performance with the multilayer perceptron (MLP) neural networks. The result indicates that SVM comparing to MLP is closer to the actual wind speed (Mohandes et al., 2004). The existing studies on using SVM for wind forecasting are very limited in that usually only one particu- lar kernel function and a specific combination of parameters are picked and used in these studies. A sys- tematic investigation focuses on kernel function encourage people to apply this method for wind energy applications. One research fill the research gap, it briefly introduce the principle of LS-SVM and analyze procedure for tuning LS-SVM parameters for optimal performance (J. Zhou et al., 2011). Least-squares support vector machines (LS-SVM) is a powerful technique which aims at getting higher accurate fore- casting of wind speed. And it is widely used for forecasting short-term wind speed forecasting. ● Fuzzy logic It is a useful and practical technique for modelling complex phenomena that may not yet be fully under- stood owing to its ability to deal with imprecise, uncertain data, or ambiguous relationships among data sets (Metternicht, 2001). This approach provides a simple method to draw definite conclusions from vague, ambiguous, or imprecise information, however it is not widely used because of the low accuracy as low ability of fuzzy logic prediction is low when studying (Klir & Folger, 1988). There are few up to date literature researching in this area, providing a possible research gap basing on its promising nature. ● Hybrid algorithm There is no single best forecasting algorithm that can be applied to any wind farm due to the fact that wind speed patterns can be very different between wind farms and are usually influenced by many factors that are location-specific and difficult to control (Guo et al., 2011). In a very recent literature, a novel hybrid modelling method which named SVR–UKF is proposed, inte- grating unscented Kalman filter (UKF) with support vector regression (SVR) in order to precisely update the short-term estimation of wind speed sequence (K. Chen & Yu, 2014). Using this method, the predic- tion errors are closer to zero with significantly smaller variations, whereas the prediction errors of the other methods are scattered more widely. Each one of physical models, statistical models, spatial correlation models and artificial intelligence models has its advantages and disadvantages, for example, the time series model is one kind of statistical models and it is popular in use because its computation is simple, ANN and KF are popular due to their 110 good nonlinear performance. Thus, another research which introduces two hybrid algorithms and com- pare show both of them have good performance. In this literature, the authors establish two hybrid meth- ods namely ARIMA-ANN model and ARIMA model based on single time series model, ANN model and KF model. The results show that: (1) Both of them have good forecasting accuracy; and (2) they are suit- able for the jumping wind samplings, which can be applied to real-time wind power systems (H. of two new A.-A. and A.-K. hybrid methods for wind speed prediction Liu, Tian, & Li, 2012). Other similar hybrid algorithms also exist, e.g. one research systematically and comprehensively investi- gated the applicability of this methodology based on two case studies on wind speed and wind power generation, respectively. Two hybrid models, namely, ARIMA–ANN and ARIMA–SVM, are selected to compare with the single ARIMA, ANN, and SVM forecasting models. The results show that the hybrid methodology does not always outperform the individual forecasting models based on ARIMA, ANN, or SVM. As such, the argument in some literature that the hybrid methodology is always superior to single models cannot hold for wind speed or power generation forecasting (Shi et al., 2012). Table 1 Innovations in different forecasting methods Methods Literature Innovation Kalman filters (Louka et al., 2008) This literature has introduced two limited-area atmospheric models for wind speed forecasts, and particularly utilise Kalman filter to these data to eliminate any possible system- atic errors, even in the lower resolution cases, contributing further to the significant reduction of the required CPU time. In particular, the paper contributes in the case of wind power prediction, which showed a remarkable improvement in the model forecasting skill. The major innovation is to counteract the drawback of Nu- merical Weather Prediction (NWP) models exhibiting sys- tematic errors in the forecasts of certain meteorological pa- rameters. Instead of increasing the model resolution that may provide considerable improvement of smaller scale flow characteristics, which remains as an open question to whether the use of higher resolution improves the forecast skill con- siderably, the methodology introduced in this paper showed high performance to the elimination of any type of systematic errors and most importantly it reduced the requirements in CPU time since its application to lower resolution data led to similar or even more accurate results compared to the costly high-resolution direct model outputs. 111 ANN (Han et al., 2011) It is urgent to improve the accuracy of short-term wind power forecast by NWF. BP network, as one of ANN, has been widely used in wind speed and wind power prediction. How- ever, the BP algorithm has its apparent disadvantages, easily getting into local minima and the convergence rate is slow. The major innovation in this literature is that the authors use Tabu Search (TS), another algorithm which can achieve the global optimizations, to train BP network. It can be shown that the new method namely MTS-ANN model can solve the inherent shortcoming of BP network by improving the con- vergence probability and precision of BP network apparently. (Guo et al., 2011) Another BP model also used to predict wind speed in the same year of 2011. In this literature, the authors integrate BP network with the idea of eliminating seasonal effects from actual wind speed datasets using seasonal exponential ad- justment. This study aims to forecast the daily average wind speed over a long period of time, such as one year ahead. Existing methods for this purpose tend to yield results with poor accuracy because they cannot properly account for seasonal effects over the long term. To improve the accuracy of daily average wind speed forecasting, SVM (Mohandes et al., 2004) This literature introduces support vector machine method for wind speed prediction, and compares it with multilayer per- ceptron (MLP). For these two algorithms, some results are shown after validating data from one case study which named Saudi Arabia which located in Madina city. One of the most important contributions is that parameters for both algorithms were optimized. Another finding is the lowest MSE of SVM is better that MLP in this case study. LS-SVM (J. Zhou et al., 2011) This literature, for the first time, presents a systematic study on fine tuning of LS-SVM model parameters for one-step ahead wind speed forecasting. The authors impleted three SVM kernels which including linear, Gaussian, and polyno- mial kernels. The results show that the performance of linear kernel is worse than the other two kernels when the training sample size or SVM order is small. For Gaussian and poly- nomial kernels, both types of parameters should be consid- ered jointly rather than independently for both kernels. LS-SVMs are compared against the persistence approach, and it is found that they can outperform the persistence model in the majority of cases. 112 Kalman filters+ SVM (K. Chen & Yu, 2014) Accuracy of wind speed forecasting is very important for improving and optimizing renewable wind power generation. However, one of the most apparent characteristics of the wind is strong stochastic nature and dynamic uncertainty. In this literature, a proposed approach named SVR–UKF, inte- grated unscented Kalman filter (UKF) with support vector regression (SVR), is developed to update the short-term estimation of wind speed sequence. The results indicate that the proposed method has much better performance in wind speed predictions than the other approaches across all the locations. ARIMA+ANN, ARIMA+Kalman (H. of two new A.-A. and A.-K. hybrid methods for wind speed prediction Liu et al., 2012) This literature introduces two new hybrid models namely ARIMA-ANN model and an ARIMA-Kalman. After respec- tively comparing the multi-step ahead prediction results by an ARIMA-ANN model, an ARIMA-Kalman model and a pure ARIMA model, it can be proved that the performance of the two hybrid models is better than that of the pure ARIMA model, and the performance of the ARIMA-Kalman model is better than that of the ARIMAANN model. The major innovation is they improved the performance of the pure ARIMA model and utilize in bigger number of fore- casting steps in order to lower the accuracy. ARIMA+ANN, ARIMA+SVM (Shi et al., 2012) This literature compares two typical hybrid models, namely ARIMA–ANN and ARIMA–SVM, with three separately single models through two case studies about wind speed and wind power generation. The results show two hybrid models are viable when predicting wind speed and wind power gen- eration. However, the most important contribution is that they found hybrid models are not superior to single methods in performance for all the forecasting time horizons investigat- ed. 4 Discussions and future research The present review provides a useful overview of the research on the use of identifying topic and key terms, identifying database and access software, conducting searches, identify sources as primary or sec- ondary, summarizing and analyzing primary sources, organizing and writing reviews. The purpose of this paper is to contribute to the topic of WES exploitation and WPF for relevant researchers and amateurs. Furthermore, this paper also gives an overall roadmap of each knowledge and descriptive analysis. Limitations of this review article are due to the scope and methods used. Only writing methods, classifi- cation of wind power forecast has been included. Surely, research derived from with other areas, for ex- ample, specific forecasting model, local energy plan and policy, power engineering. Meanwhile, the num- 113 ber of review papers is merely more than 20. Better statistic results are expected when there are more review samples. Currently, mostly wind power enterprises consider the results of weather forecast as a factor in helping managers to make decisions which rely on historical and prevailing meteorological data. However, nearly no research considers the impact of climate change while climate change over 10 years or even longer. Further research may mainly aim at figuring out how climate change over decades can influence the wind power through comparing and analyzing climate data record in complex terrain. More specifically, the objective is to explore the impact of climate change on weather condition, especially extreme weather events which influence the wind power output in wind power enterprises. 5 Conclusions In this paper, several methodologies, which including systematic review, cluster analysis and association analysis, are used to depict and generalize different popular WPF methods. Many different wind power forecasting methods used in wind energy system have been reviewed in this paper to summarize their own theory and characteristics in different methods. And their corresponding innovations are also reviewed in detail. The important points were descripted in Section 3. So the main contribution is to provide a path for researchers in this same field. There is no single best forecasting algorithm that can be applied to any wind farm and each algorithm. As a result of this study, it can be said that each algorithm or combined algorithm both has its advantages and disadvantages. There is still space to increase the prediction accuracies in very short-term, short-term, medium-term and long-term wind power predictions, respectively. In the future, more research will still to be tried in wind power prediction for the purpose of increasing the prediction accuracy and system reliability. References Abazari, A., Babaei, M., Muyeen, S. M., & Kamwa, I. (2020). Learning adaptive fuzzy droop of PV con- tribution to frequency excursion of hybrid micro-grid during parameters uncertainties. International Journal of Electrical Power and Energy Systems, 123(December 2019), 106305. https://doi.org/10.1016/j.ijepes.2020.106305 Adeyeye, K., Ijumba, N., & Colton, J. (2020). Exploring the environmental and economic impacts of wind energy: a cost-benefit perspective. International Journal of Sustainable Development and World Ecology, 27(8), 718–731. https://doi.org/10.1080/13504509.2020.1768171 Akyuz, E., Oktay, Z., & Dincer, I. (2012). Performance investigation of hydrogen production from a hy- brid wind-PV system. International Journal of Hydrogen Energy, 37(21), 16623–16630. https://doi.org/10.1016/j.ijhydene.2012.02.149 Albrecht, J., Laleman, R., & Vulsteke, E. (2015). Balancing demand-pull and supply-push measures to support renewable electricity in Europe. Renewable and Sustainable Energy Reviews, 49, 267–277. https://doi.org/10.1016/j.rser.2015.04.078 114 Alexiadis, M. C., Dokopoulos, P. S., & Sahsamanoglou, H. S. (1999). Wind speed and power forecasting based on spatial correlation models. IEEE Transactions on Energy Conversion, 14(3), 836–842. https://doi.org/10.1109/60.790962 Alpaydin, E. (2019). Introduction to machine learning. Ammari, C., Belatrache, D., Touhami, B., & Makhloufi, S. (2021). Sizing, optimization, control and ener- gy management of hybrid renewable energy system- a review. Energy and Built Environment. https://doi.org/10.1016/j.enbenv.2021.04.002 Anurag Bhardwaj, Wei Di, J. W. (2018). Deep Learning Essentials: Your hands-on guide to the fundamen- tals of deep ... - Anurag Bhardwaj, Wei Di, Jianing Wei - Google Böcker. Retrieved April 14, 2021, from https://books.google.se/books/about/Deep_Learning_Essentials.html?id=ISBKDwAAQBAJ&redir_ esc=y Archer, C. L., & Jacobson, M. Z. (2013). Geographical and seasonal variability of the global “practical” wind resources. Applied Geography, 45, 119–130. https://doi.org/10.1016/j.apgeog.2013.07.006 Azcárate, C., Blanco, R., Mallor, F., Garde, R., & Aguado, M. (2012). Peaking strategies for the manage- ment of wind-H 2 energy systems. Renewable Energy, 47, 103–111. https://doi.org/10.1016/j.renene.2012.04.016 Bhattacharjee, S., & Acharya, S. (2015). PV–wind hybrid power option for a low wind topography. Ener- gy Conversion and Management, 89, 942–954. https://doi.org/10.1016/j.enconman.2014.10.065 Bonou, A., Skelton, K., & Olsen, S. I. (2016). Ecodesign framework for developing wind turbines. Jour- nal of Cleaner Production. https://doi.org/10.1016/j.jclepro.2016.02.093 Bórawski, P., Bełdycka-Bórawska, A., Jankowski, K. J., Dubis, B., & Dunn, J. W. (2020). Development of wind energy market in the European Union. Renewable Energy, 161, 691–700. https://doi.org/https://doi.org/10.1016/j.renene.2020.07.081 Box, G. E. P., & Jenkins, G. M. (1976). Time series analysis: forecasting and control, revised ed. Holden- Day. Breslow, P. B., & Sailor, D. J. (2002). Vulnerability of wind power resources to climate change in the continental United States. Renewable Energy, 27(4), 585–598. https://doi.org/10.1016/S0960- 1481(01)00110-0 Cadenas, E., & Rivera, W. (2010). Wind speed forecasting in three different regions of Mexico, using a hybrid ARIMA–ANN model. Renewable Energy, 35(12), 2732–2738. https://doi.org/10.1016/j.renene.2010.04.022 Çapik, M., Yılmaz, A. O., & Çavuşoğlu, İ. (2012). Present situation and potential role of renewable ener- gy in Turkey. Renewable Energy, 46, 1–13. https://doi.org/10.1016/j.renene.2012.02.031 Carolin Mabel, M., & Fernandez, E. (2008). Analysis of wind power generation and prediction using 115 ANN: A case study. Renewable Energy, 33(5), 986–992. https://doi.org/10.1016/j.renene.2007.06.013 Carta, J. A., & Mentado, D. (2007). A continuous bivariate model for wind power density and wind tur- bine energy output estimations. Energy Conversion and Management, 48(2), 420–432. https://doi.org/10.1016/j.enconman.2006.06.019 Cassola, F., & Burlando, M. (2012). Wind speed and wind energy forecast through Kalman filtering of Numerical Weather Prediction model output. Applied Energy, 99, 154–166. https://doi.org/10.1016/j.apenergy.2012.03.054 Çetin, O., & Ziya Sogut, M. (2021). A new strategic approach of energy management onboard ships sup- ported by exergy and economic criteria: A case study of a cargo ship. Ocean Engineering, 219(March 2020), 108137. https://doi.org/10.1016/j.oceaneng.2020.108137 Chadee, X. T., & Clarke, R. M. (2014). Large-scale wind energy potential of the Caribbean region using near-surface reanalysis data. Renewable and Sustainable Energy Reviews, 30, 45–58. https://doi.org/10.1016/j.rser.2013.09.018 Chang, T.-J., Chen, C.-L., Tu, Y.-L., Yeh, H.-T., & Wu, Y.-T. (2015). Evaluation of the climate change impact on wind resources in Taiwan Strait. Energy Conversion and Management, 95, 435–445. https://doi.org/10.1016/j.enconman.2015.02.033 Chávez-Ramírez, A. U., Vallejo-Becerra, V., Cruz, J. C., Ornelas, R., Orozco, G., Muñoz-Guerrero, R., & Arriaga, L. G. (2013). A hybrid power plant (Solar–Wind–Hydrogen) model based in artificial intel- ligence for a remote-housing application in Mexico. International Journal of Hydrogen Energy, 38(6), 2641–2655. https://doi.org/10.1016/j.ijhydene.2012.11.140 Chen, C., Liu, Y., Kumar, M., Qin, J., & Ren, Y. (2019). Energy consumption modelling using deep learn- ing embedded semi-supervised learning. Computers & Industrial Engineering, 135, 757–765. https://doi.org/10.1016/J.CIE.2019.06.052 Chen, K., & Yu, J. (2014). Short-term wind speed prediction using an unscented Kalman filter based state-space support vector regression approach. Applied Energy, 113, 690–705. https://doi.org/10.1016/j.apenergy.2013.08.025 Christensen, P., & Lund, H. (1998). Conflicting views of sustainability: The case of wind power and na- ture conservation in Denmark. European Environment, 8(1), 1–6. https://doi.org/10.1002/(SICI)1099-0976(199801/02)8:1<1::AID-EET139>3.0.CO;2-T Colak, I., Sagiroglu, S., & Yesilbudak, M. (2012). Data mining and wind power prediction: A literature review. Renewable Energy, 46, 241–247. https://doi.org/10.1016/j.renene.2012.02.015 Crossan, F. (2003). Research philosophy: towards an understanding. Nurse Researcher, 11(1), 46–55. https://doi.org/10.7748/nr2003.10.11.1.46.c5914 Cui, Q., Liu, Y., Ali, T., Gao, J., & Chen, H. (2020). Economic and climate impacts of reducing China’s 116 renewable electricity curtailment: A comparison between CGE models with alternative nesting structures of electricity. Energy Economics, 91. https://doi.org/10.1016/j.eneco.2020.104892 Dashwood, J. (2012). Australia ’ s Energy Options : Renewables and efficiency. In Australia’s Energy Options: Renewables and efficiency (pp. 45–52). De Giorgi, M. G., Ficarella, A., & Tarantino, M. (2011). Assessment of the benefits of numerical weather predictions in wind power forecasting based on statistical methods. Energy, 36(7), 3968–3978. https://doi.org/10.1016/j.energy.2011.05.006 Demir, N., & Taşkın, A. (2013). Life cycle assessment of wind turbines in Pınarbaşı-Kayseri. Journal of Cleaner Production, 54, 253–263. https://doi.org/10.1016/j.jclepro.2013.04.016 Demolli, H., Dokuz, A. S., Ecemis, A., & Gokcek, M. (2019). Wind power forecasting based on daily wind speed data using machine learning algorithms. Energy Conversion and Management, 198, 111823. https://doi.org/10.1016/J.ENCONMAN.2019.111823 Díaz-González, F., Hau, M., Sumper, A., & Gomis-Bellmunt, O. (2015). Coordinated operation of wind turbines and flywheel storage for primary frequency control support. International Journal of Elec- trical Power & Energy Systems, 68, 313–326. https://doi.org/10.1016/j.ijepes.2014.12.062 Dihrab, S. S., & Sopian, K. (2010). Electricity generation of hybrid PV/wind systems in Iraq. Renewable Energy, 35(6), 1303–1307. https://doi.org/10.1016/j.renene.2009.12.010 Dincer, I. (2000). Renewable energy and sustainable development: a crucial review. Renewable and Sus- tainable Energy Reviews, 4(2), 157–175. https://doi.org/10.1016/S1364-0321(99)00011-8 Duan, Y., Edwards, J. S., & Dwivedi, Y. K. (2019). Artificial intelligence for decision making in the era of Big Data – evolution, challenges and research agenda. International Journal of Information Man- agement, 48, 63–71. https://doi.org/10.1016/j.ijinfomgt.2019.01.021 Dudovskiy, J. (2018). An Ultimate Guide to Writing A Dissertation in Business Studies A Step-by-Step Assistance. Research Methodology, Multiple. Retrieved from https://research-methodology.net Emeis, S. (2018). Wind Energy Meteorology: Atmospheric Physics for Wind Power Generation - Stefan Emeis - Google Books. Retrieved April 14, 2021, from https://books.google.fi/books?hl=en&lr=&id=6r1TDwAAQBAJ&oi=fnd&pg=PR5&dq=wind+for mation+wind+energy+kinetic+energy&ots=tL0AtcgUsX&sig=DVNSFnTmKsqc5vWRu- OYaI4Q7Yk&redir_esc=y#v=onepage&q=wind formation wind energy kinetic energy&f=false Eriksson, S., Bernhoff, H., & Leijon, M. (2008). Evaluation of different turbine concepts for wind power. Renewable and Sustainable Energy Reviews, 12(5), 1419–1434. https://doi.org/10.1016/j.rser.2006.05.017 Fant, C., Adam Schlosser, C., & Strzepek, K. (2015). The impact of climate change on wind and solar resources in southern Africa. Applied Energy. https://doi.org/10.1016/j.apenergy.2015.03.042 Fertig, E., & Apt, J. (2011). Economics of compressed air energy storage to integrate wind power: A case 117 study in ERCOT. Energy Policy, 39(5), 2330–2342. https://doi.org/10.1016/j.enpol.2011.01.049 Fiesler, E., Caulfield, H. J., Choudry, A., & Ryan, J. P. (1990). Maximum capacity topologies for fully connected layered neural networks with bidirectional connections. 827–831. https://doi.org/10.1109/ijcnn.1990.137672 Flora, R., Marques, A. C., & Fuinhas, J. A. (2014). Wind power idle capacity in a panel of European countries. Energy, 66, 823–830. https://doi.org/10.1016/j.energy.2013.12.061 Friedman, J. H. (2001). Greedy Function Approximation: A Gradient Boosting Machine. Annals of Statis- tics, 29(5), 1189–1232. Retrieved from http://marefateadyan.nashriyat.ir/node/150 Gao, D., Jiang, D., Liu, P., Li, Z., Hu, S., & Xu, H. (2014). An integrated energy storage system based on hydrogen storage: Process configuration and case studies with wind power. Energy, 66, 332–341. https://doi.org/10.1016/j.energy.2014.01.095 Ge, B., Wang, W., Bi, D., Rogers, C. B., Peng, F. Z., de Almeida, A. T., & Abu-Rub, H. (2013). Energy storage system-based power control for grid-connected wind power farm. International Journal of Electrical Power & Energy Systems, 44(1), 115–122. https://doi.org/10.1016/j.ijepes.2012.07.021 Gil-García, I. C., García-Cascales, M. S., Fernández-Guillamón, A., & Molina-García, A. (2019). Catego- rization and analysis of relevant factors for optimal locations in onshore and offshorewind power plants: A taxonomic review. Journal of Marine Science and Engineering, 7(11), 1–21. https://doi.org/10.3390/jmse7110391 González-Aparicio, I., & Zucker, A. (2015). Impact of wind power uncertainty forecasting on the market integration of wind energy in Spain. Applied Energy, 159, 334–349. https://doi.org/10.1016/j.apenergy.2015.08.104 González-Mingueza, C., & Muñoz-Gutiérrez, F. (2014). Wind prediction using Weather Research Fore- casting model (WRF): A case study in Peru. Energy Conversion and Management, 81, 363–373. https://doi.org/10.1016/j.enconman.2014.02.024 Griffith, A. K. (1974). A comparison and evaluation of three machine learning procedures as applied to the game of checkers. Artificial Intelligence, 5(2), 137–148. https://doi.org/10.1016/0004- 3702(74)90027-7 Güler, Ö. (2009). Wind energy status in electrical energy production of Turkey. Renewable and Sustaina- ble Energy Reviews, 13(2), 473–478. https://doi.org/10.1016/j.rser.2007.03.015 Guo, Z., Wu, J., Lu, H., & Wang, J. (2011). A case study on a hybrid wind speed forecasting method using BP neural network. Knowledge-Based Systems, 24(7), 1048–1056. https://doi.org/10.1016/j.knosys.2011.04.019 Hamilton, S. H., Fu, B., Guillaume, J. H. A., Badham, J., Elsawah, S., Gober, P., … Zare, F. (2019). A framework for characterising and evaluating the effectiveness of environmental modelling. Envi- ronmental Modelling and Software, 118, 83–98. https://doi.org/10.1016/j.envsoft.2019.04.008 118 Han, S., Li, J., & Liu, Y. (2011). Tabu Search Algorithm Optimized ANN Model for Wind Power Predic- tion with NWP. Energy Procedia, 12, 733–740. https://doi.org/10.1016/j.egypro.2011.10.099 He, Y., Li, M., Meng, Z., Chen, S., Huang, S., Hu, Y., & Zou, X. (2021). An overview of acoustic emis- sion inspection and monitoring technology in the key components of renewable energy systems. Mechanical Systems and Signal Processing, 148. https://doi.org/10.1016/j.ymssp.2020.107146 Hennessey, J. P. (1977). Some Aspects of Wind Power Statistics. Journal of Applied Meteorology, 16(2), 119–128. https://doi.org/10.1175/1520-0450(1977)016<0119:SAOWPS>2.0.CO;2 Hochreiter, S., & Schmidhuber, J. (1997). Long Short-Term Memory. Neural Computation, 9(8), 1735– 1780. https://doi.org/10.1162/neco.1997.9.8.1735 Holttinen, H. (2006). Handling of wind power forecast errors in the Nordic power market. 2006 Interna- tional Conference on Probabilistic Methods Applied to Power Systems, 1–6. https://doi.org/10.1109/PMAPS.2006.360288 Holttinen, H., Miettinen, J., & Sillanpää, S. (2013). Wind power forecasting accuracy and uncertainty in Finland. Retrieved from http://www.vtt.fi/publications/index.js Hosseini-Firouz, M. (2013). Optimal offering strategy considering the risk management for wind power producers in electricity market. International Journal of Electrical Power & Energy Systems, 49, 359–368. https://doi.org/10.1016/j.ijepes.2013.01.015 Hu, H., Wang, L., & Tao, R. (2021). Wind speed forecasting based on variational mode decomposition and improved echo state network. Renewable Energy, 164, 729–751. https://doi.org/10.1016/j.renene.2020.09.109 Hu, Q., Zhang, S., Xie, Z., Mi, J., & Wan, J. (2014). Noise model based ν-support vector regression with its application to short-term wind speed forecasting. Neural Networks : The Official Journal of the International Neural Network Society, 57C, 1–11. https://doi.org/10.1016/j.neunet.2014.05.003 Hua, Y., Oliphant, M., & Hu, E. J. (2016). Development of renewable energy in Australia and China: A comparison of policies and status. Renewable Energy, 85, 1044–1051. https://doi.org/10.1016/j.renene.2015.07.060 Huang, Q., Shi, Y., Wang, Y., Lu, L., & Cui, Y. (2015). Multi-turbine wind-solar hybrid system. Renewa- ble Energy, 76, 401–407. https://doi.org/10.1016/j.renene.2014.11.060 Huang, X., Zhang, D., & Zhang, X. (2021). Energy management of intelligent building based on deep reinforced learning. Alexandria Engineering Journal, 60(1), 1509–1517. https://doi.org/10.1016/j.aej.2020.11.005 Huang, Z., & Chalabi, Z. S. (1995). Use of time-series analysis to model and forecast wind speed. Journal of Wind Engineering and Industrial Aerodynamics, 56(2–3), 311–322. https://doi.org/10.1016/0167- 6105(94)00093-S Jacobsson, S., & Karltorp, K. (2012). Formation of competences to realize the potential of offshore wind 119 power in the European Union. Energy Policy, 44, 374–384. https://doi.org/10.1016/j.enpol.2012.01.069 Jannati, M., Hosseinian, S. H., Vahidi, B., & Li, G. (2016). A significant reduction in the costs of battery energy storage systems by use of smart parking lots in the power fluctuation smoothing process of the wind farms. Renewable Energy, 87, 1–14. https://doi.org/10.1016/j.renene.2015.09.054 Jiang, P., Liu, Z., Niu, X., & Zhang, L. (2021). A combined forecasting system based on statistical method, artificial neural networks, and deep learning methods for short-term wind speed forecasting. Energy, 217. https://doi.org/10.1016/j.energy.2020.119361 Jung, J., & Broadwater, R. P. (2014). Current status and future advances for wind speed and power fore- casting. Renewable and Sustainable Energy Reviews, 31, 762–777. https://doi.org/10.1016/j.rser.2013.12.054 Kaldellis, J. K., Kapsali, M., & Kavadias, K. A. (2010). Energy balance analysis of wind-based pumped hydro storage systems in remote island electrical networks. Applied Energy, 87(8), 2427–2437. https://doi.org/10.1016/j.apenergy.2010.02.016 Kamal, L., & Jafri, Y. Z. (1997). Time series models to simulate and forecast hourly averaged wind speed in Quetta, Pakistan. Solar Energy, 61(1), 23–32. https://doi.org/10.1016/S0038-092X(97)00037-6 Kamel, R. M., Chaouachi, A., & Nagasaka, K. (2010). Wind power smoothing using fuzzy logic pitch controller and energy capacitor system for improvement Micro-Grid performance in islanding mode. Energy, 35(5), 2119–2129. https://doi.org/10.1016/j.energy.2010.01.030 Kandpal, T. C., & Broman, L. (2014). Renewable energy education: A global status review. Renewable and Sustainable Energy Reviews, 34, 300–324. https://doi.org/10.1016/j.rser.2014.02.039 Kariniotakis, G. N., Stavrakakis, G. S., & Nogaret, E. F. (1996). Wind power forecasting using advanced neural networks models. IEEE Transactions on Energy Conversion, 11(4), 762–767. https://doi.org/10.1109/60.556376 Kavasseri, R. G., & Seetharaman, K. (2009). Day-ahead wind speed forecasting using f-ARIMA models. Renewable Energy, 34(5), 1388–1393. https://doi.org/10.1016/j.renene.2008.09.006 Kenabatho, P. K., Parida, B. P., & Moalafhi, D. B. (2012). The value of large-scale climate variables in climate change assessment: The case of Botswana’s rainfall. Physics and Chemistry of the Earth, Parts A/B/C, 50–52, 64–71. https://doi.org/10.1016/j.pce.2012.08.006 Khalid, M., & Savkin, A. V. (2010). A model predictive control approach to the problem of wind power smoothing with controlled battery storage. Renewable Energy, 35(7), 1520–1526. https://doi.org/10.1016/j.renene.2009.11.030 Khalid, M., & Savkin, A. V. (2014). Minimization and control of battery energy storage for wind power smoothing: Aggregated, distributed and semi-distributed storage. Renewable Energy, 64, 105–112. https://doi.org/10.1016/j.renene.2013.09.043 120 Kimura, Y., Onai, Y., & Ushiyama, I. (1996). A demonstrative study for the wind and solar hybrid power system. Renewable Energy, 9(1–4), 895–898. https://doi.org/10.1016/0960-1481(96)88424-2 Klir, G. J., & Folger, T. A. (1988). Fuzzy Sets, Uncertainty and Information [Hardcover]. Retrieved from http://www.amazon.com/Fuzzy-Sets-Uncertainty-Information- George/dp/0133459845/ref=sr_1_1?ie=UTF8&qid=1403952802&sr=8- 1&keywords=Fuzzy+sets,+uncertainty,+and+information Kothari, C. R. (2004). Research Methodology: Methods and Techniques. Retrieved from https://books.google.fi/books?id=8c6gkbKi-F4C Laberge, Y. (2011). Advising on Research Methods: A Consultant’s Companion. Journal of Applied Sta- tistics, 38(12), 2991–2991. https://doi.org/10.1080/02664763.2011.559375 LeCun, Y., Bengio, Y., & Hinton, G. (2015). Deep learning. Nature, 521(7553), 436–444. https://doi.org/10.1038/nature14539 Lee, K. M., Yoo, J., Kim, S.-W., Lee, J.-H., & Hong, J. (2019). Autonomic machine learning platform. International Journal of Information Management, 49, 491–501. https://doi.org/10.1016/J.IJINFOMGT.2019.07.003 Lei, M., Shiyan, L., Chuanwen, J., Hongling, L., & Yan, Z. (2009). A review on the forecasting of wind speed and generated power. Renewable and Sustainable Energy Reviews, 13(4), 915–920. https://doi.org/10.1016/j.rser.2008.02.002 Lim, S., & Chi, S. (2019). Xgboost application on bridge management systems for proactive damage estimation. Advanced Engineering Informatics, 41(October 2018), 100922. https://doi.org/10.1016/j.aei.2019.100922 Lin, J., Sun, Y., Cheng, L., & Gao, W. (2012). Assessment of the power reduction of wind farms under extreme wind condition by a high resolution simulation model. Applied Energy, 96, 21–32. https://doi.org/10.1016/j.apenergy.2011.10.028 Liu, H. of two new A.-A. and A.-K. hybrid methods for wind speed prediction, Tian, H., & Li, Y. (2012). Comparison of two new ARIMA-ANN and ARIMA-Kalman hybrid methods for wind speed pre- diction. Applied Energy, 98, 415–424. https://doi.org/10.1016/j.apenergy.2012.04.001 Liu, P., & Li, Y. (2021). An improved failure mode and effect analysis method for multi-criteria group decision-making in green logistics risk assessment. Reliability Engineering & System Safety, 215(May), 107826. https://doi.org/10.1016/j.ress.2021.107826 Liu, Y., Yu, S., Zhu, Y., Wang, D., & Liu, J. (2018). Modeling, planning, application and management of energy systems for isolated areas: A review. Renewable and Sustainable Energy Reviews, 82(June 2017), 460–470. https://doi.org/10.1016/j.rser.2017.09.063 Louka, P., Galanis, G., Siebert, N., Kariniotakis, G., Katsafados, P., Pytharoulis, I., & Kallos, G. (2008). Improvements in wind speed forecasts for wind power prediction purposes using Kalman filtering. 121 Journal of Wind Engineering and Industrial Aerodynamics, 96(12), 2348–2362. https://doi.org/10.1016/j.jweia.2008.03.013 Lu, L., Yang, H., & Burnett, J. (2002). Investigation on wind power potential on Hong Kong islands—an analysis of wind power and wind turbine characteristics. Renewable Energy, 27(1), 1–12. https://doi.org/10.1016/S0960-1481(01)00164-1 Lund, H. (2007). Renewable energy strategies for sustainable development. Energy, 32(6), 912–919. https://doi.org/10.1016/j.energy.2006.10.017 Ma, T., Yang, H., Lu, L., & Peng, J. (2014). Optimal design of an autonomous solar–wind-pumped stor- age power supply system. Applied Energy. https://doi.org/10.1016/j.apenergy.2014.11.026 Maleki, A., & Askarzadeh, A. (2014). Optimal sizing of a PV/wind/diesel system with battery storage for electrification to an off-grid remote region: A case study of Rafsanjan, Iran. Sustainable Energy Technologies and Assessments, 7, 147–153. https://doi.org/10.1016/j.seta.2014.04.005 Mannig, B., Müller, M., Starke, E., Merkenschlager, C., Mao, W., Zhi, X., … Paeth, H. (2013). Dynam- ical downscaling of climate change in Central Asia. Global and Planetary Change, 110, 26–39. https://doi.org/10.1016/j.gloplacha.2013.05.008 Maouedj, R., Mammeri, A., Draou, M. D., & Benyoucef, B. (2014). Performance Evaluation of Hybrid Photovoltaic-wind Power Systems. Energy Procedia, 50, 797–807. https://doi.org/10.1016/j.egypro.2014.06.098 Martínez, E., Sanz, F., Pellegrini, S., Jiménez, E., & Blanco, J. (2009). Life cycle assessment of a multi- megawatt wind turbine. Renewable Energy, 34(3), 667–673. https://doi.org/10.1016/j.renene.2008.05.020 McVicar, T. R., Roderick, M. L., Donohue, R. J., Li, L. T., Van Niel, T. G., Thomas, A., … Dinpashoh, Y. (2012). Global review and synthesis of trends in observed terrestrial near-surface wind speeds: Im- plications for evaporation. Journal of Hydrology, 416–417, 182–205. https://doi.org/10.1016/j.jhydrol.2011.10.024 Mens, M., Minnema, B., Overmars, K., & van den Hurk, B. (2021). Dilemmas in developing models for long-term drought risk management: The case of the National Water Model of the Netherlands. En- vironmental Modelling and Software, 143, 105100. https://doi.org/10.1016/j.envsoft.2021.105100 Mesinger, F., DiMego, G., Kalnay, E., Mitchell, K., Shafran, P. C., Ebisuzaki, W., … Shi, W. (2010). North American Regional Reanalysis. Retrieved from http://journals.ametsoc.org/doi/abs/10.1175/BAMS-87-3-343 Metternicht, G. (2001). Assessing temporal and spatial changes of salinity using fuzzy logic, remote sens- ing and GIS. Foundations of an expert system. Ecological Modelling, 144(2–3), 163–179. https://doi.org/10.1016/S0304-3800(01)00371-4 Mohamed, A. Z., Eskander, M. N., & Ghali, F. A. (2001). Fuzzy logic control based maximum power 122 tracking of a wind energy system. Renewable Energy, 23(2), 235–245. https://doi.org/10.1016/S0960-1481(00)00099-9 Mohandes, M. A., Halawani, T. O., Rehman, S., & Hussain, A. A. (2004). Support vector machines for wind speed prediction. Renewable Energy, 29(6), 939–947. https://doi.org/10.1016/j.renene.2003.11.009 Murage, M. W., & Anderson, C. L. (2014). Contribution of pumped hydro storage to integration of wind power in Kenya: An optimal control approach. Renewable Energy, 63, 698–707. https://doi.org/10.1016/j.renene.2013.10.026 Necefer, L., Wong-Parodi, G., Small, M. J., & Begay-Campbell, S. (2018). Integrating technical, econom- ic and cultural impacts in a decision support tool for energy resource management in the Navajo Nation. Energy Strategy Reviews, 22(May 2017), 136–146. https://doi.org/10.1016/j.esr.2018.06.004 Novak, P., Ekelund, T., Jovik, I., & Schmidtbauer, B. (1995). Modeling and control of variable-speed wind-turbine drive-system dynamics. IEEE Control Systems Magazine, 15(4), 28–38. https://doi.org/10.1109/37.408463 Nygaard, I., Rasmussen, K., Badger, J., Nielsen, T. T., Hansen, L. B., Stisen, S., … Togola, I. (2010). Using modeling, satellite images and existing global datasets for rapid preliminary assessments of renewable energy resources: The case of Mali. Renewable and Sustainable Energy Reviews, 14(8), 2359–2371. https://doi.org/10.1016/j.rser.2010.04.001 Olaofe, Z. O. (2014). A 5-day wind speed & power forecasts using a layer recurrent neural network (LRNN). Sustainable Energy Technologies and Assessments, 6, 1–24. https://doi.org/10.1016/j.seta.2013.12.001 On the Vertical Extrapolation of Mean Wind Power DensitySedefian, L. (1980). On the Vertical Extrapo- lation of Mean Wind Power Density. Journal of Applied Meteorology, 19(4), 488–493. https://doi.org/10.1175/1520-0450(1980)019<0488:OTVEOM>2.0.CO;2 Oñederra, O., Asensio, F. J., Saldaña, G., Martín, J. I. S., & Zamora, I. (2020). Wind energy harnessing in a railway infrastructure: Converter topology and control proposal. Electronics (Switzerland), 9(11), 1–14. https://doi.org/10.3390/electronics9111943 Ortegon, K., Nies, L. F., & Sutherland, J. W. (2013). Preparing for end of service life of wind turbines. Journal of Cleaner Production, 39, 191–199. https://doi.org/10.1016/j.jclepro.2012.08.022 Ortiz-García, E. G., Salcedo-Sanz, S., Pérez-Bellido, Á. M., Gascón-Moreno, J., Portilla-Figueras, J. A., & Prieto, L. (2011). Short-term wind speed prediction in wind farms based on banks of support vec- tor machines. Wind Energy, 14(2), 193–207. https://doi.org/10.1002/we.411 Ozcan, M. (2014). Assessment of renewable energy incentive system from investors’ perspective. Renew- able Energy, 71, 425–432. https://doi.org/10.1016/j.renene.2014.05.053 123 Peng, H., Liu, F., & Yang, X. (2013). A hybrid strategy of short term wind power prediction. Renewable Energy, 50, 590–595. https://doi.org/10.1016/j.renene.2012.07.022 Pereira, E. B., Martins, F. R., Pes, M. P., da Cruz Segundo, E. I., & Lyra, A. de A. (2013). The impacts of global climate changes on the wind power density in Brazil. Renewable Energy, 49, 107–110. https://doi.org/10.1016/j.renene.2012.01.053 Persson, A. (1991). Kalman filtering a new approach to adaptive statistical interpretation of numerical meteorological forecasts. WMO Technical Document, (421). Pinson, P., & Kariniotakis, G. (2004). On-line assessment of prediction risk for wind power production forecasts. Wind Energy, 7(2), 119–132. https://doi.org/10.1002/we.114 Poncela, M., Poncela, P., & Perán, J. R. (2013). Automatic tuning of Kalman filters by maximum likeli- hood methods for wind energy forecasting. Applied Energy, 108, 349–362. https://doi.org/10.1016/j.apenergy.2013.03.041 Pryor, S. C., & Barthelmie, R. J. (2010). Climate change impacts on wind energy: A review. Renewable and Sustainable Energy Reviews, 14(1), 430–437. https://doi.org/10.1016/j.rser.2009.07.028 Rezaeian-Marjani, S., Masoumzadehasl, S., Galvani, S., & Talavat, V. (2020). Probabilistic assessment of D-STATCOM operation considering correlated uncertain variables. International Transactions on Electrical Energy Systems, 30(10). https://doi.org/10.1002/2050-7038.12522 Rienecker, M. M., Suarez, M. J., Gelaro, R., Todling, R., Bacmeister, J., Liu, E., … Woollen, J. (2011). MERRA: NASA’s Modern-Era Retrospective Analysis for Research and Applications. Journal of Climate, 24(14), 3624–3648. https://doi.org/10.1175/JCLI-D-11-00015.1 Robert Vautard, Françoise Thais, Isabelle Tobin, François-Marie Bréon, Jean-Guy Devezeaux de Lavergne, Augustin Colette, P. Y. & P. M. R. (2014). Time series analysis : forecasting and control: George E. P. Jenkins, Gwilym M., Box: Amazon.com: Books. Rose, S., & Apt, J. (2015). What can reanalysis data tell us about wind power? Renewable Energy, 83, 963–969. https://doi.org/10.1016/j.renene.2015.05.027 Rudenko, N. V., Ershov, V. V., & Evstafiev, V. V. (2017). The Environment Friendly Power Source for Power Supply of Mobile Communication Base Stations. IOP Conference Series: Earth and Envi- ronmental Science, 66(1). https://doi.org/10.1088/1755-1315/66/1/012024 Saeed, M. A., Ahmed, Z., & Zhang, W. (2020). Wind energy potential and economic analysis with a com- parison of different methods for determining the optimal distribution parameters. Renewable Energy, 161, 1092–1109. https://doi.org/https://doi.org/10.1016/j.renene.2020.07.064 Saha, S., Moorthi, S., Pan, H.-L., Wu, X., Wang, J., Nadiga, S., … Goldberg, M. (2010). The NCEP Cli- mate Forecast System Reanalysis. Bulletin of the American Meteorological Society, 91(8), 1015– 1057. https://doi.org/10.1175/2010BAMS3001.1 Sailor, D. J., Smith, M., & Hart, M. (2008). Climate change implications for wind power resources in the 124 Northwest United States. Renewable Energy, 33(11), 2393–2406. https://doi.org/10.1016/j.renene.2008.01.007 Salcedo-Sanz, S., Ortiz-García, E. G., Portilla-Figueras, A., Prieto, L., & Paredes, D. (2009). Hybridizing the fifth generation mesoscale model with artificial neural networks for short-term wind speed pre- diction. Renewable Energy, 34(6), 1451–1457. https://doi.org/10.1016/j.renene.2008.10.017 Saleh Asheghabadi, M., Sahafnia, M., Bahadori, A., & Bakhshayeshi, N. (2019). Seismic behavior of suction caisson for offshore wind turbine to generate more renewable energy. International Journal of Environmental Science and Technology, 16(7), 2961–2972. https://doi.org/10.1007/s13762-018- 2150-8 Sample, J. E., Duncan, N., Ferguson, M., & Cooksley, S. (2015). Scotland׳s hydropower: Current capacity, future potential and the possible impacts of climate change. Renewable and Sustainable Energy Re- views, 52, 111–122. https://doi.org/10.1016/j.rser.2015.07.071 Sarrias-Mena, R., Fernández-Ramírez, L. M., García-Vázquez, C. A., & Jurado, F. (2014). Fuzzy logic based power management strategy of a multi-MW doubly-fed induction generator wind turbine with battery and ultracapacitor. Energy, 70, 561–576. https://doi.org/10.1016/j.energy.2014.04.049 Satkin, M., Noorollahi, Y., Abbaspour, M., & Yousefi, H. (2014). Multi criteria site selection model for wind-compressed air energy storage power plants in Iran. Renewable and Sustainable Energy Re- views, 32, 579–590. https://doi.org/10.1016/j.rser.2014.01.054 Schleisner, L. (2000). Life cycle assessment of a wind farm and related externalities. Renewable Energy, 20(3), 279–288. https://doi.org/10.1016/S0960-1481(99)00123-8 Sebastián, R., & Peña-Alzola, R. (2015). Control and simulation of a flywheel energy storage for a wind diesel power system. International Journal of Electrical Power & Energy Systems, 64, 1049–1056. https://doi.org/10.1016/j.ijepes.2014.08.017 Seljom, P., Rosenberg, E., Fidje, A., Haugen, J. E., Meir, M., Rekstad, J., & Jarlset, T. (2011). Modelling the effects of climate change on the energy system—A case study of Norway. Energy Policy, 39(11), 7310–7321. https://doi.org/10.1016/j.enpol.2011.08.054 Shamshirband, S., Keivani, A., Mohammadi, K., Lee, M., Hamid, S. H. A., & Petkovic, D. (2016). As- sessing the proficiency of adaptive neuro-fuzzy system to estimate wind power density: Case study of Aligoodarz, Iran. Renewable and Sustainable Energy Reviews, 59, 429–435. https://doi.org/10.1016/j.rser.2015.12.269 Sharma, N., Sharma, R., & Jindal, N. (2021). Machine Learning and Deep Learning Applications-A Vi- sion. Global Transitions Proceedings, 0–8. https://doi.org/10.1016/j.gltp.2021.01.004 Shen, N., Wang, Y., Peng, H., & Hou, Z. (2020). Renewable energy green innovation, fossil energy con- sumption, and air pollution-spatial empirical analysis based on China. Sustainability (Switzerland), 12(16). https://doi.org/10.3390/SU12166397 125 Shen, S., Sadoughi, M., Chen, X., Hong, M., & Hu, C. (2019). A deep learning method for online capaci- ty estimation of lithium-ion batteries. Journal of Energy Storage, 25, 100817. https://doi.org/10.1016/J.EST.2019.100817 Sheng Shen, Mohammadkazem Sadoughi, Xiangyi Chen, Mingyi Hong, C. H. (2019). IEEE Xplore: IEEE Transactions on Pattern Analysis and Machine Intelligence. Retrieved November 1, 2019, from Journal of Energy Storage website: https://ieeexplore.ieee.org/xpl/RecentIssue.jsp?punumber=34 Shi, J., Guo, J., & Zheng, S. (2012). Evaluation of hybrid forecasting approaches for wind speed and power generation time series. Renewable and Sustainable Energy Reviews, 16(5), 3471–3480. https://doi.org/10.1016/j.rser.2012.02.044 Shin, D., Ha, E., Kim, T., & Kim, C. (2021). Short-term photovoltaic power generation predicting by input/output structure of weather forecast using deep learning. Soft Computing, 25(1), 771–783. https://doi.org/10.1007/s00500-020-05199-7 Shin, Y., Koo, W. Y., Kim, T. H., Jung, S., & Kim, H. (2015). Capacity design and operation planning of a hybrid PV–wind–battery–diesel power generation system in the case of Deokjeok Island. Applied Thermal Engineering, 89, 514–525. https://doi.org/10.1016/j.applthermaleng.2015.06.043 Shukur, O. B., & Lee, M. H. (2015). Daily wind speed forecasting through hybrid KF-ANN model based on ARIMA. Renewable Energy, 76, 637–647. https://doi.org/10.1016/j.renene.2014.11.084 Simoes, M. G., Bose, B. K., & Spiegel, R. J. (1997a). Design and performance evaluation of a fuzzy- logic-based variable-speed wind generation system. IEEE Transactions on Industry Applications, 33(4), 956–965. https://doi.org/10.1109/28.605737 Simoes, M. G., Bose, B. K., & Spiegel, R. J. (1997b). Fuzzy logic based intelligent control of a variable speed cage machine wind generation system. IEEE Transactions on Power Electronics, 12(1), 87– 95. https://doi.org/10.1109/63.554173 Soman, S. S., Zareipour, H., Malik, O., & Mandal, P. (2010). A review of wind power and wind speed forecasting methods with different time horizons. North American Power Symposium 2010, 1–8. https://doi.org/10.1109/NAPS.2010.5619586 Soukissian, T. H., & Papadopoulos, A. (2015). Effects of different wind data sources in offshore wind power assessment. Renewable Energy, 77, 101–114. https://doi.org/10.1016/j.renene.2014.12.009 Staid, A., & Guikema, S. D. (2013). Statistical analysis of installed wind capacity in the United States. Energy Policy, 60, 378–385. https://doi.org/10.1016/j.enpol.2013.05.076 Swider, D. J. (2007). Compressed Air Energy Storage in an Electricity System With Significant Wind Power Generation. IEEE Transactions on Energy Conversion, 22(1), 95–102. https://doi.org/10.1109/TEC.2006.889547 Taraft, S., Rekioua, D., & Aouzellag, D. (2013). Wind Power Control System Associated to the Flywheel 126 Energy Storage System Connected to the Grid. Energy Procedia, 36, 1147–1157. https://doi.org/10.1016/j.egypro.2013.07.130 Taylor, J. W., & Buizza, R. (2002). Neural network load forecasting with weather ensemble predictions. IEEE Transactions on Power Systems, 17(3), 626–632. https://doi.org/10.1109/TPWRS.2002.800906 Tong, Z., Cheng, Z., & Tong, S. (2021). A review on the development of compressed air energy storage in China: Technical and economic challenges to commercialization. Renewable and Sustainable Ener- gy Reviews, 135(September 2019), 110178. https://doi.org/10.1016/j.rser.2020.110178 Trenberth Kevin E. (1990). Recent Observed Interdecadal Climate Changes in the Northern Hemisphere. Bulletin of the American Meteorological Society, 71(7), 988–993. Trianni, A., Cagno, E., Bertolotti, M., Thollander, P., & Andersson, E. (2019). Energy management : A practice-based assessment model. Applied Energy, 235(December 2018), 1614–1636. https://doi.org/10.1016/j.apenergy.2018.11.032 Ucar, A., & Balo, F. (2010). Assessment of wind power potential for turbine installation in coastal areas of Turkey. Renewable and Sustainable Energy Reviews, 14(7), 1901–1912. https://doi.org/10.1016/j.rser.2010.03.021 Uddin, M. S., & Kumar, S. (2014). Energy, emissions and environmental impact analysis of wind turbine using life cycle assessment technique. Journal of Cleaner Production, 69, 153–164. https://doi.org/10.1016/j.jclepro.2014.01.073 Uppala, S. M., KÅllberg, P. W., Simmons, A. J., Andrae, U., Bechtold, V. D. C., Fiorino, M., … Woollen, J. (2005). The ERA-40 re-analysis. Quarterly Journal of the Royal Meteorological Society, 131(612), 2961–3012. https://doi.org/10.1256/qj.04.176 Vijay M, D., Singh, B., & Bhuvaneswari, G. (2020). A High-Performance Microgrid with a Mechanical Sensorless SynRG Operated Wind Energy Generating System. IEEE Transactions on Industrial In- formatics, 16(12), 7349–7359. https://doi.org/10.1109/TII.2020.2976906 Wang, B., Ke, R.-Y., Yuan, X.-C., & Wei, Y.-M. (2014). China׳s regional assessment of renewable energy vulnerability to climate change. Renewable and Sustainable Energy Reviews, 40, 185–195. https://doi.org/10.1016/j.rser.2014.07.154 Wang, J., Qin, S., Zhou, Q., & Jiang, H. (2015). Medium-term wind speeds forecasting utilizing hybrid models for three different sites in Xinjiang, China. Renewable Energy, 76, 91–101. https://doi.org/10.1016/j.renene.2014.11.011 Wang, P., Fan, E., & Wang, P. (2021). Comparative analysis of image classification algorithms based on traditional machine learning and deep learning. Pattern Recognition Letters, 141, 61–67. https://doi.org/10.1016/j.patrec.2020.07.042 Wang, R., Li, C., Fu, W., & Tang, G. (2020). Deep Learning Method Based on Gated Recurrent Unit and 127 Variational Mode Decomposition for Short-Term Wind Power Interval Prediction. IEEE Transac- tions on Neural Networks and Learning Systems, 31(10), 3814–3827. https://doi.org/10.1109/TNNLS.2019.2946414 Wang, X. Y., Mahinda Vilathgamuwa, D., & Choi, S. S. (2008). Determination of Battery Storage Capaci- ty in Energy Buffer for Wind Farm. IEEE Transactions on Energy Conversion, 23(3), 868–878. https://doi.org/10.1109/TEC.2008.921556 Wu, J., Zhang, B., Li, H., Li, Z., Chen, Y., & Miao, X. (2014). Statistical distribution for wind power forecast error and its application to determine optimal size of energy storage system. International Journal of Electrical Power & Energy Systems, 55, 100–107. https://doi.org/10.1016/j.ijepes.2013.09.003 Xydis, G. (2013). On the exergetic capacity factor of a wind – solar power generation system. Journal of Cleaner Production, 47, 437–445. https://doi.org/10.1016/j.jclepro.2012.07.014 Yang, H., Lu, L., & Zhou, W. (2007). A novel optimization sizing model for hybrid solar-wind power generation system. Solar Energy, 81(1), 76–84. https://doi.org/10.1016/j.solener.2006.06.010 Yang, M., Patiño-Echeverri, D., & Yang, F. (2012). Wind power generation in China: Understanding the mismatch between capacity and generation. Renewable Energy, 41, 145–151. https://doi.org/10.1016/j.renene.2011.10.013 Zahedi, A. (2010). Australian renewable energy progress. Renewable and Sustainable Energy Reviews, 14(8), 2208–2213. https://doi.org/10.1016/j.rser.2010.03.026 Zahedi, Ahmad. (2014). Sustainable Power Supply Using Solar Energy and Wind Power Combined with Energy Storage. Energy Procedia, 52, 642–650. https://doi.org/10.1016/j.egypro.2014.07.120 Zhang, D., Qian, L., Mao, B., Huang, C., Huang, B., & Si, Y. (2018). A Data-Driven Design for Fault Detection of Wind Turbines Using Random Forests and XGboost. IEEE Access, 6, 21020–21031. https://doi.org/10.1109/ACCESS.2018.2818678 Zhang, J., Chowdhury, S., & Messac, A. (2014). A comprehensive measure of the energy resource: Wind power potential (WPP). Energy Conversion and Management, 86, 388–398. https://doi.org/10.1016/j.enconman.2014.04.083 Zhang, Y., Pan, G., Zhao, Y., Li, Q., & Wang, F. (2020). Short-term wind speed interval prediction based on artificial intelligence methods and error probability distribution. Energy Conversion and Man- agement, 224(June), 113346. https://doi.org/10.1016/j.enconman.2020.113346 Zhao, P., Dai, Y., & Wang, J. (2014). Design and thermodynamic analysis of a hybrid energy storage sys- tem based on A-CAES (adiabatic compressed air energy storage) and FESS (flywheel energy stor- age system) for wind power application. Energy, 70, 674–684. https://doi.org/10.1016/j.energy.2014.04.055 Zhao, P., Wang, M., Wang, J., & Dai, Y. (2015). A preliminary dynamic behaviors analysis of a hybrid 128 energy storage system based on adiabatic compressed air energy storage and flywheel energy stor- age system for wind power application. Energy, 84, 825–839. https://doi.org/10.1016/j.energy.2015.03.067 Zhao, X., Wang, C., Su, J., & Wang, J. (2019). Research and application based on the swarm intelligence algorithm and artificial intelligence for wind farm decision system. Renewable Energy, 134, 681– 697. https://doi.org/10.1016/j.renene.2018.11.061 Zhao, Z.-Y., Zhu, J., & Zuo, J. (2014). Sustainable development of the wind power industry in a complex environment: a flexibility study. Energy Policy, 75, 392–397. https://doi.org/10.1016/j.enpol.2014.10.017 Zhou, J., Shi, J., & Li, G. (2011). Fine tuning support vector machines for short-term wind speed forecast- ing. Energy Conversion and Management, 52(4), 1990–1998. https://doi.org/10.1016/j.enconman.2010.11.007 Zhou, W., Yang, H., & Fang, Z. (2006). Wind power potential and characteristic analysis of the Pearl River Delta region, China. Renewable Energy, 31(6), 739–753. https://doi.org/10.1016/j.renene.2005.05.006 129 Appendix 2. Paper 2: The impact of climate change on wind power enter- prises Yang, W., Liu, Y., and Yang, G. (2014): The impact of climate change on wind power enterprises. Proceedings of the 11th International conference on Innova- tion & Management. It has been published in Proceedings of the 11th International conference on Innova- tion & Management. Wenshan Yang: Defining of the research design, methodology and research methods, verification and analysis of findings, writing the body text of the original article, editing the article at different stages. Yang Liu: Editing the article at different stages. Gengsheng Yang: Acquisition of research resources. 130 The impact of climate change on wind power enterprises Wenshan YANG1,2, Yang LIU*1, Gengsheng YANG2 1 Department of Production, University of Vaasa, Vaasa, Finland, 65200 2 Hubei Meteorological Bureau, Wuhan, P. R. China, 430074 (E-mail: yangw@student.uwasa.fi, yli@uva.fi, yanggengsheng11@163.com) Abstract The aim of this study is to survey impact of climate change on wind power enterprises (WPEs). So the main work is to develop a new framework to determine the impacts climate change on WPEs. A huge database of climate data record (CDR) in used with meteorological and geographic variables for period 1979-present three times per day for 10 meter U wind component, 10 meter V wind component, total column water vapour, 2 meter temperature, medium cloud cover, mean sea level pressure, gravity wave dissipation, etc. Among these meteorological data, air density and wind speed can be used to predict the wind power density for wind farm site selecting. And wind speed, relative humidity, rainfall, snowfall over few decades can be chosen to draw graphs in order to analyze the significant correlation with wind power. Keywords wind power forecast, WPEs, climate change, global wind energy resource distribution, CDR. 1 Introduction Changes in temperature, precipitation, sea level, and the frequency and severity of extreme events will likely affect how much energy is produced, delivered, and consumed. It can be noticed that quite few researches regard climate change as a long-time measurement factor for renewable energy plant. Today, wind energy is widely used to produce electricity in many countries all over the world, such as China, United States, Germany, Spain, India, and Denmark. For wind energy, mostly wind power enterprises consider results of weather conditions as a factor for helping managers to make decisions which rely on historical and prevailing meteorological data. For example, they are more interested in the predicted and actual data of wind speed, wind direction and the rated power of the wind turbine. Long-term wind pro- jects do not include the yet unknown impacts of climate changes on wind power (Pereira, Martins, Pes, da Cruz Segundo, & Lyra, 2013). However, climate change is another concept which is not as same as weather condition. Climate change is a long-term accumulated effect of weather condition caused by many factors, and it directly results ex- treme weather events (EWEs) which has significant impacts on power generation. EWEs may generate rainstorm, typhoon, mudslide, extreme temperatures, hailstone across most parts of the world. With rising concerns about climate change, a recent similar climate change research report, man-made climate change contributed to some of 2012’s most extreme weather, including the spring and summer heat waves that baked parts of the United States and Hurricane Sandy, which devastated coastal communities along the 131 eastern coast of the country. Understanding the climate change phenomenon and its impact on wind pow- er system is of increasing importance all over the world. Wind power generations depend on the natural environment especially under extreme wind condition, which means the wind speed is near or over the cut-out speed (Lin et al., 2012). In addition to this, other disaster events including, lightning, strong wind, extreme temperatures, haze, fog acid rain and hail will result in sudden power drop under an extreme condition. However, lack of good quality data, of sufficient record length and spatial coverage usually restricts mod- el development and performance geared towards assessing the effects of climate change in these areas (Kenabatho, Parida, & Moalafhi, 2012). This research mainly aims at figuring out how climate change over decades can influence the energy power through comparing and analyzing climate data record. More specifically, the objective is to explore the impact of climate change on weather condition especially extreme weather events which influence the power output in wind power enterprises from the sight of meteorology. Meanwhile, make summarize typical climatical characteristics over three selected meteorological stations. The specific goals are described here, • Observe and analyze climate data record from approximately 1970 to recent days, and identify the main developing trend of climate change for three selected dissimilar typical meteorological stations. • Explicit how climate data record can be used to eliminate economic loss in wind power enter- prises. Section 2 will give a literature review about the current research situation and also summarize the art-of- state. Section 3 will draw a framework which assessing wind resources and finding correlation between wind power and some meteorological elements. Section 4 presents discussions and finally summary, this part also gives future research prospects. 2 Literature review Kenabatho et al. (2012) present an analysis of rainfall and climate data in order to determine the time of change in rainfall series and identify possible correlations between rainfall and temperature. They use historical rainfall, climate data from rainfall stations and large-scale CDR from 1965 to 2008. The results indicate that temperature is a significant rainfall predictor in Botswana. Meanwhile, they make predic- tions of future rainfall patterns in Botswana(Kenabatho et al., 2012). According to Birgit Mannig et al. (2013), central Asia has already implemented the high-resolution re- gional climate REMO and they use REMO simulations to get higher accuracy results which are closer to observational data(Mannig et al., 2013). 132 From geographic aspect, Turkey is a country which located between Europe and Asia, bordering the Med- iterranean, Aegean and Black Seas. One of the most apparent characteristics of this country is that its location between the colder European and warmer Asian and African systems also cause a wide variety of temperature and climate difference. Turkey has a rich potential of wind energy with 1002MW due to that this country surrounded by many mountains, and its unique geographical character creates a regular and moderate air inflow through its mountainous valley structures. A research investigated the renewable energy situation including hydropower, wind and geothermal potential in Turkey(Çapik, Yılmaz, & Çavuşoğlu, 2012). A research review and explicit the trends of observed terrestrial near-surface wind speeds for many coun- tries all over the world, and study the observed rates of atmospheric evaporative evaporation. In this study, they separately describe the trends of near-surface terrestrial wind speed and the trends in evaporative demand, then analyze the importance of wind speed to the evaporative process. It is not a review paper about wind resource but also a relevant reference to water resource assessment. The result show that near- surface terrestrial wind speeds are declining in both hemispheres for both the tropical and mid-latitudes. Four primary meteorological variables, which including wind speed, atmospheric humidity, radiation and air temperature, were also assessed. This paper also highlight the important role that wind speed trends play in governing evaporative demand trends (McVicar et al., 2012). Another similar literature provides global and seasonal estimate of the “practical” wind power, which defined as delivered from wind turbines in high-wind locations over land and near-shore, obtained with a 3-D numerical model. They found that the global practical wind power potential varies significantly with season and hemisphere. Such as the highest wind power output are generated in the season of winter and oppositely the lowest are in summer (Archer & Jacobson, 2013). One relevant research focuses on studying wind energy, solar energy, bio-energy resource separately in Mali by using modelling, satellite imagines and existing global datasets. The methods applied make ex- tensive use of satellite remote sensing and meteorological mesoscale modeling. In this study, the prelimi- nary wind resource map produced show that the North of Mali has more potential of wind energy (Ny- gaard et al., 2010). 3 Towards a conceptual framework Some changes associated with climate evolution will likely benefit the wind energy industry while other changes may negatively impact wind energy developments, and expansion of wind energy installed ca- pacity is poised to play a key role in climate change mitigation (Pryor & Barthelmie, 2010). Various wind power stations have different terrain feature which can be seen from global wind energy resource distribu- tion. This research mainly aims at figuring out how climate change over decades can influence the energy power through comparing and analyzing climate data record. More Specifically, classify all relative mete- orological phenomenon which are closely related to power output of wind power enterprises after analyz- 133 ing climate data record, and determine what factors influence them. Whether seasons effect? Whether time periods effect? Whether regional differences exist? On one hand, there are different ways to estimate the wind resources at a site and wind resource varies significantly from one location to another. Wind power density (WPD) (W/m2) can be predicted form wind speed and air density. WPD is a nonlinear function of the probability density function of wind speed [9]. Estimates of WPD are presented as wind class which ranging from 1 to 7 and the assessment of WPD play important role when site selecting. Accurate assessment of wind resource will not only reduce eco- nomic loss caused by EWEs but also increase the wind power output. On the other hand, some few meteorological elements may have relationship with wind power. Six major meteorological elements including air temperature, humidity, wind speed, wind direction, air pressure and rainfall are analyzed. Besides of these, there are evaporation, snow depth, snowfall, total cloud cover, sunshine duration and so on. Among these, we propose the assumption that meteorological elements such as wind speed, relative humidity, rainfall, snowfall over few decades have correlations with wind power, which needs to be proved by analysis with longitudinal data over decades, and then try to determine the function expression by using variety of CDR variables and draw graphs to analyze the relationship be- tween them. 4 Discussion and conclusion Robert Vautard and his colleagues used a sophisticated regional climate model (this model describes the interactions between wind turbines and the atmosphere) to determine the climate impacts on all current (2012) and near-future (2020) wind energy production according to European Union energy and climate policies. The team found that wind farms form a weak but stable anticyclonic flow over Europe but only in winter there will be a significant impact on daily temperature and rainfall, and this effect is weaker than what natural interannual changes result for the climate change. They use a regional climate model de- scribing the interactions between turbines and the atmosphere, and find limited impacts. However, the impacts remain much weaker than the natural climate interannual variability and changes expected from greenhouse gas emissions (Robert Vautard, Françoise Thais, Isabelle Tobin, François-Marie Bréon, Jean- Guy Devezeaux de Lavergne, Augustin Colette, 2014). This recent valuable research suggests a new and contrary direction to the impacts of climate change on WPEs. The exploitation of off-shore wind power has more potential prospects comparing to that of on-shore as the coastal wind speed is higher than inland. Meteorological and hydrological factors definitely influence the wind power output, especially typhoon and seawater corrosion. Sea waves can scour foundation of wind turbine tower, corrode undersea cables affecting and affect its stability. Meanwhile, the historical meteorological data of wind and tide directly influence wind power site selection. Therefore, the further research may be related to coastal wind power forecast, and focus on the impact of climate change on off- 134 shore wind power plants due to its typical climatic characteristics. This suggests that further research into the links between large-scale climate variability and wind power generation is necessary and important. In this paper, this work presents some of the most recent relevant research which combines climate change with wind power predicting. The outcome of this research is to help decision-makers of wind power enterprise on making strategic decisions which including site selection, management and mainte- nance of wind power station, and long-term wind power generation forecast etc. Furthermore, it also develops a proposal framework which to determine the relationship between and wind power density and wind power. And the further research should explore deeply in CDR so as to help the researchers in the field develop more effective wind speed and power forecasting methods by finding specific function expression. References Abazari, A., Babaei, M., Muyeen, S. M., & Kamwa, I. (2020). Learning adaptive fuzzy droop of PV con- tribution to frequency excursion of hybrid micro-grid during parameters uncertainties. International Journal of Electrical Power and Energy Systems, 123(December 2019), 106305. https://doi.org/10.1016/j.ijepes.2020.106305 Adeyeye, K., Ijumba, N., & Colton, J. (2020). Exploring the environmental and economic impacts of wind energy: a cost-benefit perspective. International Journal of Sustainable Development and World Ecology, 27(8), 718–731. https://doi.org/10.1080/13504509.2020.1768171 Akyuz, E., Oktay, Z., & Dincer, I. (2012). Performance investigation of hydrogen production from a hy- brid wind-PV system. International Journal of Hydrogen Energy, 37(21), 16623–16630. https://doi.org/10.1016/j.ijhydene.2012.02.149 Albrecht, J., Laleman, R., & Vulsteke, E. (2015). Balancing demand-pull and supply-push measures to support renewable electricity in Europe. Renewable and Sustainable Energy Reviews, 49, 267–277. https://doi.org/10.1016/j.rser.2015.04.078 Alexiadis, M. C., Dokopoulos, P. S., & Sahsamanoglou, H. S. (1999). Wind speed and power forecasting based on spatial correlation models. IEEE Transactions on Energy Conversion, 14(3), 836–842. https://doi.org/10.1109/60.790962 Alpaydin, E. (2019). Introduction to machine learning. Ammari, C., Belatrache, D., Touhami, B., & Makhloufi, S. (2021). Sizing, optimization, control and ener- gy management of hybrid renewable energy system- a review. Energy and Built Environment. https://doi.org/10.1016/j.enbenv.2021.04.002 Anurag Bhardwaj, Wei Di, J. W. (2018). Deep Learning Essentials: Your hands-on guide to the fundamen- tals of deep ... - Anurag Bhardwaj, Wei Di, Jianing Wei - Google Böcker. Retrieved April 14, 2021, from https://books.google.se/books/about/Deep_Learning_Essentials.html?id=ISBKDwAAQBAJ&redir_ 135 esc=y Archer, C. L., & Jacobson, M. Z. (2013). Geographical and seasonal variability of the global “practical” wind resources. Applied Geography, 45, 119–130. https://doi.org/10.1016/j.apgeog.2013.07.006 Azcárate, C., Blanco, R., Mallor, F., Garde, R., & Aguado, M. (2012). Peaking strategies for the manage- ment of wind-H 2 energy systems. Renewable Energy, 47, 103–111. https://doi.org/10.1016/j.renene.2012.04.016 Bhattacharjee, S., & Acharya, S. (2015). PV–wind hybrid power option for a low wind topography. Ener- gy Conversion and Management, 89, 942–954. https://doi.org/10.1016/j.enconman.2014.10.065 Bonou, A., Skelton, K., & Olsen, S. I. (2016). Ecodesign framework for developing wind turbines. Jour- nal of Cleaner Production. https://doi.org/10.1016/j.jclepro.2016.02.093 Bórawski, P., Bełdycka-Bórawska, A., Jankowski, K. J., Dubis, B., & Dunn, J. W. (2020). Development of wind energy market in the European Union. Renewable Energy, 161, 691–700. https://doi.org/https://doi.org/10.1016/j.renene.2020.07.081 Box, G. E. P., & Jenkins, G. M. (1976). Time series analysis: forecasting and control, revised ed. Holden- Day. Breslow, P. B., & Sailor, D. J. (2002). Vulnerability of wind power resources to climate change in the continental United States. Renewable Energy, 27(4), 585–598. https://doi.org/10.1016/S0960- 1481(01)00110-0 Cadenas, E., & Rivera, W. (2010). Wind speed forecasting in three different regions of Mexico, using a hybrid ARIMA–ANN model. Renewable Energy, 35(12), 2732–2738. https://doi.org/10.1016/j.renene.2010.04.022 Çapik, M., Yılmaz, A. O., & Çavuşoğlu, İ. (2012). Present situation and potential role of renewable ener- gy in Turkey. Renewable Energy, 46, 1–13. https://doi.org/10.1016/j.renene.2012.02.031 Carolin Mabel, M., & Fernandez, E. (2008). Analysis of wind power generation and prediction using ANN: A case study. Renewable Energy, 33(5), 986–992. https://doi.org/10.1016/j.renene.2007.06.013 Carta, J. A., & Mentado, D. (2007). A continuous bivariate model for wind power density and wind tur- bine energy output estimations. Energy Conversion and Management, 48(2), 420–432. https://doi.org/10.1016/j.enconman.2006.06.019 Cassola, F., & Burlando, M. (2012). Wind speed and wind energy forecast through Kalman filtering of Numerical Weather Prediction model output. Applied Energy, 99, 154–166. https://doi.org/10.1016/j.apenergy.2012.03.054 Çetin, O., & Ziya Sogut, M. (2021). A new strategic approach of energy management onboard ships sup- ported by exergy and economic criteria: A case study of a cargo ship. Ocean Engineering, 219(March 2020), 108137. https://doi.org/10.1016/j.oceaneng.2020.108137 136 Chadee, X. T., & Clarke, R. M. (2014). Large-scale wind energy potential of the Caribbean region using near-surface reanalysis data. Renewable and Sustainable Energy Reviews, 30, 45–58. https://doi.org/10.1016/j.rser.2013.09.018 Chang, T.-J., Chen, C.-L., Tu, Y.-L., Yeh, H.-T., & Wu, Y.-T. (2015). Evaluation of the climate change impact on wind resources in Taiwan Strait. Energy Conversion and Management, 95, 435–445. https://doi.org/10.1016/j.enconman.2015.02.033 Chávez-Ramírez, A. U., Vallejo-Becerra, V., Cruz, J. C., Ornelas, R., Orozco, G., Muñoz-Guerrero, R., & Arriaga, L. G. (2013). A hybrid power plant (Solar–Wind–Hydrogen) model based in artificial intel- ligence for a remote-housing application in Mexico. International Journal of Hydrogen Energy, 38(6), 2641–2655. https://doi.org/10.1016/j.ijhydene.2012.11.140 Chen, C., Liu, Y., Kumar, M., Qin, J., & Ren, Y. (2019). Energy consumption modelling using deep learn- ing embedded semi-supervised learning. Computers & Industrial Engineering, 135, 757–765. https://doi.org/10.1016/J.CIE.2019.06.052 Chen, K., & Yu, J. (2014). Short-term wind speed prediction using an unscented Kalman filter based state-space support vector regression approach. Applied Energy, 113, 690–705. https://doi.org/10.1016/j.apenergy.2013.08.025 Christensen, P., & Lund, H. (1998). Conflicting views of sustainability: The case of wind power and na- ture conservation in Denmark. European Environment, 8(1), 1–6. https://doi.org/10.1002/(SICI)1099-0976(199801/02)8:1<1::AID-EET139>3.0.CO;2-T Colak, I., Sagiroglu, S., & Yesilbudak, M. (2012). Data mining and wind power prediction: A literature review. Renewable Energy, 46, 241–247. https://doi.org/10.1016/j.renene.2012.02.015 Crossan, F. (2003). Research philosophy: towards an understanding. Nurse Researcher, 11(1), 46–55. https://doi.org/10.7748/nr2003.10.11.1.46.c5914 Cui, Q., Liu, Y., Ali, T., Gao, J., & Chen, H. (2020). Economic and climate impacts of reducing China’s renewable electricity curtailment: A comparison between CGE models with alternative nesting structures of electricity. Energy Economics, 91. https://doi.org/10.1016/j.eneco.2020.104892 Dashwood, J. (2012). Australia ’ s Energy Options : Renewables and efficiency. In Australia’s Energy Options: Renewables and efficiency (pp. 45–52). De Giorgi, M. G., Ficarella, A., & Tarantino, M. (2011). Assessment of the benefits of numerical weather predictions in wind power forecasting based on statistical methods. Energy, 36(7), 3968–3978. https://doi.org/10.1016/j.energy.2011.05.006 Demir, N., & Taşkın, A. (2013). Life cycle assessment of wind turbines in Pınarbaşı-Kayseri. Journal of Cleaner Production, 54, 253–263. https://doi.org/10.1016/j.jclepro.2013.04.016 Demolli, H., Dokuz, A. S., Ecemis, A., & Gokcek, M. (2019). Wind power forecasting based on daily wind speed data using machine learning algorithms. Energy Conversion and Management, 198, 137 111823. https://doi.org/10.1016/J.ENCONMAN.2019.111823 Díaz-González, F., Hau, M., Sumper, A., & Gomis-Bellmunt, O. (2015). Coordinated operation of wind turbines and flywheel storage for primary frequency control support. International Journal of Elec- trical Power & Energy Systems, 68, 313–326. https://doi.org/10.1016/j.ijepes.2014.12.062 Dihrab, S. S., & Sopian, K. (2010). Electricity generation of hybrid PV/wind systems in Iraq. Renewable Energy, 35(6), 1303–1307. https://doi.org/10.1016/j.renene.2009.12.010 Dincer, I. (2000). Renewable energy and sustainable development: a crucial review. Renewable and Sus- tainable Energy Reviews, 4(2), 157–175. https://doi.org/10.1016/S1364-0321(99)00011-8 Duan, Y., Edwards, J. S., & Dwivedi, Y. K. (2019). Artificial intelligence for decision making in the era of Big Data – evolution, challenges and research agenda. International Journal of Information Man- agement, 48, 63–71. https://doi.org/10.1016/j.ijinfomgt.2019.01.021 Dudovskiy, J. (2018). An Ultimate Guide to Writing A Dissertation in Business Studies A Step-by-Step Assistance. Research Methodology, Multiple. Retrieved from https://research-methodology.net Emeis, S. (2018). Wind Energy Meteorology: Atmospheric Physics for Wind Power Generation - Stefan Emeis - Google Books. Retrieved April 14, 2021, from https://books.google.fi/books?hl=en&lr=&id=6r1TDwAAQBAJ&oi=fnd&pg=PR5&dq=wind+for mation+wind+energy+kinetic+energy&ots=tL0AtcgUsX&sig=DVNSFnTmKsqc5vWRu- OYaI4Q7Yk&redir_esc=y#v=onepage&q=wind formation wind energy kinetic energy&f=false Eriksson, S., Bernhoff, H., & Leijon, M. (2008). Evaluation of different turbine concepts for wind power. Renewable and Sustainable Energy Reviews, 12(5), 1419–1434. https://doi.org/10.1016/j.rser.2006.05.017 Fant, C., Adam Schlosser, C., & Strzepek, K. (2015). The impact of climate change on wind and solar resources in southern Africa. Applied Energy. https://doi.org/10.1016/j.apenergy.2015.03.042 Fertig, E., & Apt, J. (2011). Economics of compressed air energy storage to integrate wind power: A case study in ERCOT. Energy Policy, 39(5), 2330–2342. https://doi.org/10.1016/j.enpol.2011.01.049 Fiesler, E., Caulfield, H. J., Choudry, A., & Ryan, J. P. (1990). Maximum capacity topologies for fully connected layered neural networks with bidirectional connections. 827–831. https://doi.org/10.1109/ijcnn.1990.137672 Flora, R., Marques, A. C., & Fuinhas, J. A. (2014). Wind power idle capacity in a panel of European countries. Energy, 66, 823–830. https://doi.org/10.1016/j.energy.2013.12.061 Friedman, J. H. (2001). Greedy Function Approximation: A Gradient Boosting Machine. Annals of Statis- tics, 29(5), 1189–1232. Retrieved from http://marefateadyan.nashriyat.ir/node/150 Gao, D., Jiang, D., Liu, P., Li, Z., Hu, S., & Xu, H. (2014). An integrated energy storage system based on hydrogen storage: Process configuration and case studies with wind power. Energy, 66, 332–341. https://doi.org/10.1016/j.energy.2014.01.095 138 Ge, B., Wang, W., Bi, D., Rogers, C. B., Peng, F. Z., de Almeida, A. T., & Abu-Rub, H. (2013). Energy storage system-based power control for grid-connected wind power farm. International Journal of Electrical Power & Energy Systems, 44(1), 115–122. https://doi.org/10.1016/j.ijepes.2012.07.021 Gil-García, I. C., García-Cascales, M. S., Fernández-Guillamón, A., & Molina-García, A. (2019). Catego- rization and analysis of relevant factors for optimal locations in onshore and offshorewind power plants: A taxonomic review. Journal of Marine Science and Engineering, 7(11), 1–21. https://doi.org/10.3390/jmse7110391 González-Aparicio, I., & Zucker, A. (2015). Impact of wind power uncertainty forecasting on the market integration of wind energy in Spain. Applied Energy, 159, 334–349. https://doi.org/10.1016/j.apenergy.2015.08.104 González-Mingueza, C., & Muñoz-Gutiérrez, F. (2014). Wind prediction using Weather Research Fore- casting model (WRF): A case study in Peru. Energy Conversion and Management, 81, 363–373. https://doi.org/10.1016/j.enconman.2014.02.024 Griffith, A. K. (1974). A comparison and evaluation of three machine learning procedures as applied to the game of checkers. Artificial Intelligence, 5(2), 137–148. https://doi.org/10.1016/0004- 3702(74)90027-7 Güler, Ö. (2009). Wind energy status in electrical energy production of Turkey. Renewable and Sustaina- ble Energy Reviews, 13(2), 473–478. https://doi.org/10.1016/j.rser.2007.03.015 Guo, Z., Wu, J., Lu, H., & Wang, J. (2011). A case study on a hybrid wind speed forecasting method using BP neural network. Knowledge-Based Systems, 24(7), 1048–1056. https://doi.org/10.1016/j.knosys.2011.04.019 Hamilton, S. H., Fu, B., Guillaume, J. H. A., Badham, J., Elsawah, S., Gober, P., … Zare, F. (2019). A framework for characterising and evaluating the effectiveness of environmental modelling. Envi- ronmental Modelling and Software, 118, 83–98. https://doi.org/10.1016/j.envsoft.2019.04.008 Han, S., Li, J., & Liu, Y. (2011). Tabu Search Algorithm Optimized ANN Model for Wind Power Predic- tion with NWP. Energy Procedia, 12, 733–740. https://doi.org/10.1016/j.egypro.2011.10.099 He, Y., Li, M., Meng, Z., Chen, S., Huang, S., Hu, Y., & Zou, X. (2021). An overview of acoustic emis- sion inspection and monitoring technology in the key components of renewable energy systems. Mechanical Systems and Signal Processing, 148. https://doi.org/10.1016/j.ymssp.2020.107146 Hennessey, J. P. (1977). Some Aspects of Wind Power Statistics. Journal of Applied Meteorology, 16(2), 119–128. https://doi.org/10.1175/1520-0450(1977)016<0119:SAOWPS>2.0.CO;2 Hochreiter, S., & Schmidhuber, J. (1997). Long Short-Term Memory. Neural Computation, 9(8), 1735– 1780. https://doi.org/10.1162/neco.1997.9.8.1735 Holttinen, H. (2006). Handling of wind power forecast errors in the Nordic power market. 2006 Interna- tional Conference on Probabilistic Methods Applied to Power Systems, 1–6. 139 https://doi.org/10.1109/PMAPS.2006.360288 Holttinen, H., Miettinen, J., & Sillanpää, S. (2013). Wind power forecasting accuracy and uncertainty in Finland. Retrieved from http://www.vtt.fi/publications/index.js Hosseini-Firouz, M. (2013). Optimal offering strategy considering the risk management for wind power producers in electricity market. International Journal of Electrical Power & Energy Systems, 49, 359–368. https://doi.org/10.1016/j.ijepes.2013.01.015 Hu, H., Wang, L., & Tao, R. (2021). Wind speed forecasting based on variational mode decomposition and improved echo state network. Renewable Energy, 164, 729–751. https://doi.org/10.1016/j.renene.2020.09.109 Hu, Q., Zhang, S., Xie, Z., Mi, J., & Wan, J. (2014). Noise model based ν-support vector regression with its application to short-term wind speed forecasting. Neural Networks : The Official Journal of the International Neural Network Society, 57C, 1–11. https://doi.org/10.1016/j.neunet.2014.05.003 Hua, Y., Oliphant, M., & Hu, E. J. (2016). Development of renewable energy in Australia and China: A comparison of policies and status. Renewable Energy, 85, 1044–1051. https://doi.org/10.1016/j.renene.2015.07.060 Huang, Q., Shi, Y., Wang, Y., Lu, L., & Cui, Y. (2015). Multi-turbine wind-solar hybrid system. Renewa- ble Energy, 76, 401–407. https://doi.org/10.1016/j.renene.2014.11.060 Huang, X., Zhang, D., & Zhang, X. (2021). Energy management of intelligent building based on deep reinforced learning. Alexandria Engineering Journal, 60(1), 1509–1517. https://doi.org/10.1016/j.aej.2020.11.005 Huang, Z., & Chalabi, Z. S. (1995). Use of time-series analysis to model and forecast wind speed. Journal of Wind Engineering and Industrial Aerodynamics, 56(2–3), 311–322. https://doi.org/10.1016/0167- 6105(94)00093-S Jacobsson, S., & Karltorp, K. (2012). Formation of competences to realize the potential of offshore wind power in the European Union. Energy Policy, 44, 374–384. https://doi.org/10.1016/j.enpol.2012.01.069 Jannati, M., Hosseinian, S. H., Vahidi, B., & Li, G. (2016). A significant reduction in the costs of battery energy storage systems by use of smart parking lots in the power fluctuation smoothing process of the wind farms. Renewable Energy, 87, 1–14. https://doi.org/10.1016/j.renene.2015.09.054 Jiang, P., Liu, Z., Niu, X., & Zhang, L. (2021). A combined forecasting system based on statistical method, artificial neural networks, and deep learning methods for short-term wind speed forecasting. Energy, 217. https://doi.org/10.1016/j.energy.2020.119361 Jung, J., & Broadwater, R. P. (2014). Current status and future advances for wind speed and power fore- casting. Renewable and Sustainable Energy Reviews, 31, 762–777. https://doi.org/10.1016/j.rser.2013.12.054 140 Kaldellis, J. K., Kapsali, M., & Kavadias, K. A. (2010). Energy balance analysis of wind-based pumped hydro storage systems in remote island electrical networks. Applied Energy, 87(8), 2427–2437. https://doi.org/10.1016/j.apenergy.2010.02.016 Kamal, L., & Jafri, Y. Z. (1997). Time series models to simulate and forecast hourly averaged wind speed in Quetta, Pakistan. Solar Energy, 61(1), 23–32. https://doi.org/10.1016/S0038-092X(97)00037-6 Kamel, R. M., Chaouachi, A., & Nagasaka, K. (2010). Wind power smoothing using fuzzy logic pitch controller and energy capacitor system for improvement Micro-Grid performance in islanding mode. Energy, 35(5), 2119–2129. https://doi.org/10.1016/j.energy.2010.01.030 Kandpal, T. C., & Broman, L. (2014). Renewable energy education: A global status review. Renewable and Sustainable Energy Reviews, 34, 300–324. https://doi.org/10.1016/j.rser.2014.02.039 Kariniotakis, G. N., Stavrakakis, G. S., & Nogaret, E. F. (1996). Wind power forecasting using advanced neural networks models. IEEE Transactions on Energy Conversion, 11(4), 762–767. https://doi.org/10.1109/60.556376 Kavasseri, R. G., & Seetharaman, K. (2009). Day-ahead wind speed forecasting using f-ARIMA models. Renewable Energy, 34(5), 1388–1393. https://doi.org/10.1016/j.renene.2008.09.006 Kenabatho, P. K., Parida, B. P., & Moalafhi, D. B. (2012). The value of large-scale climate variables in climate change assessment: The case of Botswana’s rainfall. Physics and Chemistry of the Earth, Parts A/B/C, 50–52, 64–71. https://doi.org/10.1016/j.pce.2012.08.006 Khalid, M., & Savkin, A. V. (2010). A model predictive control approach to the problem of wind power smoothing with controlled battery storage. Renewable Energy, 35(7), 1520–1526. https://doi.org/10.1016/j.renene.2009.11.030 Khalid, M., & Savkin, A. V. (2014). Minimization and control of battery energy storage for wind power smoothing: Aggregated, distributed and semi-distributed storage. Renewable Energy, 64, 105–112. https://doi.org/10.1016/j.renene.2013.09.043 Kimura, Y., Onai, Y., & Ushiyama, I. (1996). A demonstrative study for the wind and solar hybrid power system. Renewable Energy, 9(1–4), 895–898. https://doi.org/10.1016/0960-1481(96)88424-2 Klir, G. J., & Folger, T. A. (1988). Fuzzy Sets, Uncertainty and Information [Hardcover]. Retrieved from http://www.amazon.com/Fuzzy-Sets-Uncertainty-Information- George/dp/0133459845/ref=sr_1_1?ie=UTF8&qid=1403952802&sr=8- 1&keywords=Fuzzy+sets,+uncertainty,+and+information Kothari, C. R. (2004). Research Methodology: Methods and Techniques. Retrieved from https://books.google.fi/books?id=8c6gkbKi-F4C Laberge, Y. (2011). Advising on Research Methods: A Consultant’s Companion. Journal of Applied Sta- tistics, 38(12), 2991–2991. https://doi.org/10.1080/02664763.2011.559375 LeCun, Y., Bengio, Y., & Hinton, G. (2015). Deep learning. Nature, 521(7553), 436–444. 141 https://doi.org/10.1038/nature14539 Lee, K. M., Yoo, J., Kim, S.-W., Lee, J.-H., & Hong, J. (2019). Autonomic machine learning platform. International Journal of Information Management, 49, 491–501. https://doi.org/10.1016/J.IJINFOMGT.2019.07.003 Lei, M., Shiyan, L., Chuanwen, J., Hongling, L., & Yan, Z. (2009). A review on the forecasting of wind speed and generated power. Renewable and Sustainable Energy Reviews, 13(4), 915–920. https://doi.org/10.1016/j.rser.2008.02.002 Lim, S., & Chi, S. (2019). Xgboost application on bridge management systems for proactive damage estimation. Advanced Engineering Informatics, 41(October 2018), 100922. https://doi.org/10.1016/j.aei.2019.100922 Lin, J., Sun, Y., Cheng, L., & Gao, W. (2012). Assessment of the power reduction of wind farms under extreme wind condition by a high resolution simulation model. Applied Energy, 96, 21–32. https://doi.org/10.1016/j.apenergy.2011.10.028 Liu, H. of two new A.-A. and A.-K. hybrid methods for wind speed prediction, Tian, H., & Li, Y. (2012). Comparison of two new ARIMA-ANN and ARIMA-Kalman hybrid methods for wind speed pre- diction. Applied Energy, 98, 415–424. https://doi.org/10.1016/j.apenergy.2012.04.001 Liu, P., & Li, Y. (2021). An improved failure mode and effect analysis method for multi-criteria group decision-making in green logistics risk assessment. Reliability Engineering & System Safety, 215(May), 107826. https://doi.org/10.1016/j.ress.2021.107826 Liu, Y., Yu, S., Zhu, Y., Wang, D., & Liu, J. (2018). Modeling, planning, application and management of energy systems for isolated areas: A review. Renewable and Sustainable Energy Reviews, 82(June 2017), 460–470. https://doi.org/10.1016/j.rser.2017.09.063 Louka, P., Galanis, G., Siebert, N., Kariniotakis, G., Katsafados, P., Pytharoulis, I., & Kallos, G. (2008). Improvements in wind speed forecasts for wind power prediction purposes using Kalman filtering. Journal of Wind Engineering and Industrial Aerodynamics, 96(12), 2348–2362. https://doi.org/10.1016/j.jweia.2008.03.013 Lu, L., Yang, H., & Burnett, J. (2002). Investigation on wind power potential on Hong Kong islands—an analysis of wind power and wind turbine characteristics. Renewable Energy, 27(1), 1–12. https://doi.org/10.1016/S0960-1481(01)00164-1 Lund, H. (2007). Renewable energy strategies for sustainable development. Energy, 32(6), 912–919. https://doi.org/10.1016/j.energy.2006.10.017 Ma, T., Yang, H., Lu, L., & Peng, J. (2014). Optimal design of an autonomous solar–wind-pumped stor- age power supply system. Applied Energy. https://doi.org/10.1016/j.apenergy.2014.11.026 Maleki, A., & Askarzadeh, A. (2014). Optimal sizing of a PV/wind/diesel system with battery storage for electrification to an off-grid remote region: A case study of Rafsanjan, Iran. Sustainable Energy 142 Technologies and Assessments, 7, 147–153. https://doi.org/10.1016/j.seta.2014.04.005 Mannig, B., Müller, M., Starke, E., Merkenschlager, C., Mao, W., Zhi, X., … Paeth, H. (2013). Dynam- ical downscaling of climate change in Central Asia. Global and Planetary Change, 110, 26–39. https://doi.org/10.1016/j.gloplacha.2013.05.008 Maouedj, R., Mammeri, A., Draou, M. D., & Benyoucef, B. (2014). Performance Evaluation of Hybrid Photovoltaic-wind Power Systems. Energy Procedia, 50, 797–807. https://doi.org/10.1016/j.egypro.2014.06.098 Martínez, E., Sanz, F., Pellegrini, S., Jiménez, E., & Blanco, J. (2009). Life cycle assessment of a multi- megawatt wind turbine. Renewable Energy, 34(3), 667–673. https://doi.org/10.1016/j.renene.2008.05.020 McVicar, T. R., Roderick, M. L., Donohue, R. J., Li, L. T., Van Niel, T. G., Thomas, A., … Dinpashoh, Y. (2012). Global review and synthesis of trends in observed terrestrial near-surface wind speeds: Im- plications for evaporation. Journal of Hydrology, 416–417, 182–205. https://doi.org/10.1016/j.jhydrol.2011.10.024 Mens, M., Minnema, B., Overmars, K., & van den Hurk, B. (2021). Dilemmas in developing models for long-term drought risk management: The case of the National Water Model of the Netherlands. En- vironmental Modelling and Software, 143, 105100. https://doi.org/10.1016/j.envsoft.2021.105100 Mesinger, F., DiMego, G., Kalnay, E., Mitchell, K., Shafran, P. C., Ebisuzaki, W., … Shi, W. (2010). North American Regional Reanalysis. Retrieved from http://journals.ametsoc.org/doi/abs/10.1175/BAMS-87-3-343 Metternicht, G. (2001). Assessing temporal and spatial changes of salinity using fuzzy logic, remote sens- ing and GIS. Foundations of an expert system. Ecological Modelling, 144(2–3), 163–179. https://doi.org/10.1016/S0304-3800(01)00371-4 Mohamed, A. Z., Eskander, M. N., & Ghali, F. A. (2001). Fuzzy logic control based maximum power tracking of a wind energy system. Renewable Energy, 23(2), 235–245. https://doi.org/10.1016/S0960-1481(00)00099-9 Mohandes, M. A., Halawani, T. O., Rehman, S., & Hussain, A. A. (2004). Support vector machines for wind speed prediction. Renewable Energy, 29(6), 939–947. https://doi.org/10.1016/j.renene.2003.11.009 Murage, M. W., & Anderson, C. L. (2014). Contribution of pumped hydro storage to integration of wind power in Kenya: An optimal control approach. Renewable Energy, 63, 698–707. https://doi.org/10.1016/j.renene.2013.10.026 Necefer, L., Wong-Parodi, G., Small, M. J., & Begay-Campbell, S. (2018). Integrating technical, econom- ic and cultural impacts in a decision support tool for energy resource management in the Navajo Nation. Energy Strategy Reviews, 22(May 2017), 136–146. https://doi.org/10.1016/j.esr.2018.06.004 143 Novak, P., Ekelund, T., Jovik, I., & Schmidtbauer, B. (1995). Modeling and control of variable-speed wind-turbine drive-system dynamics. IEEE Control Systems Magazine, 15(4), 28–38. https://doi.org/10.1109/37.408463 Nygaard, I., Rasmussen, K., Badger, J., Nielsen, T. T., Hansen, L. B., Stisen, S., … Togola, I. (2010). Using modeling, satellite images and existing global datasets for rapid preliminary assessments of renewable energy resources: The case of Mali. Renewable and Sustainable Energy Reviews, 14(8), 2359–2371. https://doi.org/10.1016/j.rser.2010.04.001 Olaofe, Z. O. (2014). A 5-day wind speed & power forecasts using a layer recurrent neural network (LRNN). Sustainable Energy Technologies and Assessments, 6, 1–24. https://doi.org/10.1016/j.seta.2013.12.001 On the Vertical Extrapolation of Mean Wind Power DensitySedefian, L. (1980). On the Vertical Extrapo- lation of Mean Wind Power Density. Journal of Applied Meteorology, 19(4), 488–493. https://doi.org/10.1175/1520-0450(1980)019<0488:OTVEOM>2.0.CO;2 Oñederra, O., Asensio, F. J., Saldaña, G., Martín, J. I. S., & Zamora, I. (2020). Wind energy harnessing in a railway infrastructure: Converter topology and control proposal. Electronics (Switzerland), 9(11), 1–14. https://doi.org/10.3390/electronics9111943 Ortegon, K., Nies, L. F., & Sutherland, J. W. (2013). Preparing for end of service life of wind turbines. Journal of Cleaner Production, 39, 191–199. https://doi.org/10.1016/j.jclepro.2012.08.022 Ortiz-García, E. G., Salcedo-Sanz, S., Pérez-Bellido, Á. M., Gascón-Moreno, J., Portilla-Figueras, J. A., & Prieto, L. (2011). Short-term wind speed prediction in wind farms based on banks of support vec- tor machines. Wind Energy, 14(2), 193–207. https://doi.org/10.1002/we.411 Ozcan, M. (2014). Assessment of renewable energy incentive system from investors’ perspective. Renew- able Energy, 71, 425–432. https://doi.org/10.1016/j.renene.2014.05.053 Peng, H., Liu, F., & Yang, X. (2013). A hybrid strategy of short term wind power prediction. Renewable Energy, 50, 590–595. https://doi.org/10.1016/j.renene.2012.07.022 Pereira, E. B., Martins, F. R., Pes, M. P., da Cruz Segundo, E. I., & Lyra, A. de A. (2013). The impacts of global climate changes on the wind power density in Brazil. Renewable Energy, 49, 107–110. https://doi.org/10.1016/j.renene.2012.01.053 Persson, A. (1991). Kalman filtering a new approach to adaptive statistical interpretation of numerical meteorological forecasts. WMO Technical Document, (421). Pinson, P., & Kariniotakis, G. (2004). On-line assessment of prediction risk for wind power production forecasts. Wind Energy, 7(2), 119–132. https://doi.org/10.1002/we.114 Poncela, M., Poncela, P., & Perán, J. R. (2013). Automatic tuning of Kalman filters by maximum likeli- hood methods for wind energy forecasting. Applied Energy, 108, 349–362. https://doi.org/10.1016/j.apenergy.2013.03.041 144 Pryor, S. C., & Barthelmie, R. J. (2010). Climate change impacts on wind energy: A review. Renewable and Sustainable Energy Reviews, 14(1), 430–437. https://doi.org/10.1016/j.rser.2009.07.028 Rezaeian-Marjani, S., Masoumzadehasl, S., Galvani, S., & Talavat, V. (2020). Probabilistic assessment of D-STATCOM operation considering correlated uncertain variables. International Transactions on Electrical Energy Systems, 30(10). https://doi.org/10.1002/2050-7038.12522 Rienecker, M. M., Suarez, M. J., Gelaro, R., Todling, R., Bacmeister, J., Liu, E., … Woollen, J. (2011). MERRA: NASA’s Modern-Era Retrospective Analysis for Research and Applications. Journal of Climate, 24(14), 3624–3648. https://doi.org/10.1175/JCLI-D-11-00015.1 Robert Vautard, Françoise Thais, Isabelle Tobin, François-Marie Bréon, Jean-Guy Devezeaux de Lavergne, Augustin Colette, P. Y. & P. M. R. (2014). Time series analysis : forecasting and control: George E. P. Jenkins, Gwilym M., Box: Amazon.com: Books. Rose, S., & Apt, J. (2015). What can reanalysis data tell us about wind power? Renewable Energy, 83, 963–969. https://doi.org/10.1016/j.renene.2015.05.027 Rudenko, N. V., Ershov, V. V., & Evstafiev, V. V. (2017). The Environment Friendly Power Source for Power Supply of Mobile Communication Base Stations. IOP Conference Series: Earth and Envi- ronmental Science, 66(1). https://doi.org/10.1088/1755-1315/66/1/012024 Saeed, M. A., Ahmed, Z., & Zhang, W. (2020). Wind energy potential and economic analysis with a com- parison of different methods for determining the optimal distribution parameters. Renewable Energy, 161, 1092–1109. https://doi.org/https://doi.org/10.1016/j.renene.2020.07.064 Saha, S., Moorthi, S., Pan, H.-L., Wu, X., Wang, J., Nadiga, S., … Goldberg, M. (2010). The NCEP Cli- mate Forecast System Reanalysis. Bulletin of the American Meteorological Society, 91(8), 1015– 1057. https://doi.org/10.1175/2010BAMS3001.1 Sailor, D. J., Smith, M., & Hart, M. (2008). Climate change implications for wind power resources in the Northwest United States. Renewable Energy, 33(11), 2393–2406. https://doi.org/10.1016/j.renene.2008.01.007 Salcedo-Sanz, S., Ortiz-García, E. G., Portilla-Figueras, A., Prieto, L., & Paredes, D. (2009). Hybridizing the fifth generation mesoscale model with artificial neural networks for short-term wind speed pre- diction. Renewable Energy, 34(6), 1451–1457. https://doi.org/10.1016/j.renene.2008.10.017 Saleh Asheghabadi, M., Sahafnia, M., Bahadori, A., & Bakhshayeshi, N. (2019). Seismic behavior of suction caisson for offshore wind turbine to generate more renewable energy. International Journal of Environmental Science and Technology, 16(7), 2961–2972. https://doi.org/10.1007/s13762-018- 2150-8 Sample, J. E., Duncan, N., Ferguson, M., & Cooksley, S. (2015). Scotland׳s hydropower: Current capacity, future potential and the possible impacts of climate change. Renewable and Sustainable Energy Re- views, 52, 111–122. https://doi.org/10.1016/j.rser.2015.07.071 145 Sarrias-Mena, R., Fernández-Ramírez, L. M., García-Vázquez, C. A., & Jurado, F. (2014). Fuzzy logic based power management strategy of a multi-MW doubly-fed induction generator wind turbine with battery and ultracapacitor. Energy, 70, 561–576. https://doi.org/10.1016/j.energy.2014.04.049 Satkin, M., Noorollahi, Y., Abbaspour, M., & Yousefi, H. (2014). Multi criteria site selection model for wind-compressed air energy storage power plants in Iran. Renewable and Sustainable Energy Re- views, 32, 579–590. https://doi.org/10.1016/j.rser.2014.01.054 Schleisner, L. (2000). Life cycle assessment of a wind farm and related externalities. Renewable Energy, 20(3), 279–288. https://doi.org/10.1016/S0960-1481(99)00123-8 Sebastián, R., & Peña-Alzola, R. (2015). Control and simulation of a flywheel energy storage for a wind diesel power system. International Journal of Electrical Power & Energy Systems, 64, 1049–1056. https://doi.org/10.1016/j.ijepes.2014.08.017 Seljom, P., Rosenberg, E., Fidje, A., Haugen, J. E., Meir, M., Rekstad, J., & Jarlset, T. (2011). Modelling the effects of climate change on the energy system—A case study of Norway. Energy Policy, 39(11), 7310–7321. https://doi.org/10.1016/j.enpol.2011.08.054 Shamshirband, S., Keivani, A., Mohammadi, K., Lee, M., Hamid, S. H. A., & Petkovic, D. (2016). As- sessing the proficiency of adaptive neuro-fuzzy system to estimate wind power density: Case study of Aligoodarz, Iran. Renewable and Sustainable Energy Reviews, 59, 429–435. https://doi.org/10.1016/j.rser.2015.12.269 Sharma, N., Sharma, R., & Jindal, N. (2021). Machine Learning and Deep Learning Applications-A Vi- sion. Global Transitions Proceedings, 0–8. https://doi.org/10.1016/j.gltp.2021.01.004 Shen, N., Wang, Y., Peng, H., & Hou, Z. (2020). Renewable energy green innovation, fossil energy con- sumption, and air pollution-spatial empirical analysis based on China. Sustainability (Switzerland), 12(16). https://doi.org/10.3390/SU12166397 Shen, S., Sadoughi, M., Chen, X., Hong, M., & Hu, C. (2019). A deep learning method for online capaci- ty estimation of lithium-ion batteries. Journal of Energy Storage, 25, 100817. https://doi.org/10.1016/J.EST.2019.100817 Sheng Shen, Mohammadkazem Sadoughi, Xiangyi Chen, Mingyi Hong, C. H. (2019). IEEE Xplore: IEEE Transactions on Pattern Analysis and Machine Intelligence. Retrieved November 1, 2019, from Journal of Energy Storage website: https://ieeexplore.ieee.org/xpl/RecentIssue.jsp?punumber=34 Shi, J., Guo, J., & Zheng, S. (2012). Evaluation of hybrid forecasting approaches for wind speed and power generation time series. Renewable and Sustainable Energy Reviews, 16(5), 3471–3480. https://doi.org/10.1016/j.rser.2012.02.044 Shin, D., Ha, E., Kim, T., & Kim, C. (2021). Short-term photovoltaic power generation predicting by input/output structure of weather forecast using deep learning. Soft Computing, 25(1), 771–783. https://doi.org/10.1007/s00500-020-05199-7 146 Shin, Y., Koo, W. Y., Kim, T. H., Jung, S., & Kim, H. (2015). Capacity design and operation planning of a hybrid PV–wind–battery–diesel power generation system in the case of Deokjeok Island. Applied Thermal Engineering, 89, 514–525. https://doi.org/10.1016/j.applthermaleng.2015.06.043 Shukur, O. B., & Lee, M. H. (2015). Daily wind speed forecasting through hybrid KF-ANN model based on ARIMA. Renewable Energy, 76, 637–647. https://doi.org/10.1016/j.renene.2014.11.084 Simoes, M. G., Bose, B. K., & Spiegel, R. J. (1997a). Design and performance evaluation of a fuzzy- logic-based variable-speed wind generation system. IEEE Transactions on Industry Applications, 33(4), 956–965. https://doi.org/10.1109/28.605737 Simoes, M. G., Bose, B. K., & Spiegel, R. J. (1997b). Fuzzy logic based intelligent control of a variable speed cage machine wind generation system. IEEE Transactions on Power Electronics, 12(1), 87– 95. https://doi.org/10.1109/63.554173 Soman, S. S., Zareipour, H., Malik, O., & Mandal, P. (2010). A review of wind power and wind speed forecasting methods with different time horizons. North American Power Symposium 2010, 1–8. https://doi.org/10.1109/NAPS.2010.5619586 Soukissian, T. H., & Papadopoulos, A. (2015). Effects of different wind data sources in offshore wind power assessment. Renewable Energy, 77, 101–114. https://doi.org/10.1016/j.renene.2014.12.009 Staid, A., & Guikema, S. D. (2013). Statistical analysis of installed wind capacity in the United States. Energy Policy, 60, 378–385. https://doi.org/10.1016/j.enpol.2013.05.076 Swider, D. J. (2007). Compressed Air Energy Storage in an Electricity System With Significant Wind Power Generation. IEEE Transactions on Energy Conversion, 22(1), 95–102. https://doi.org/10.1109/TEC.2006.889547 Taraft, S., Rekioua, D., & Aouzellag, D. (2013). Wind Power Control System Associated to the Flywheel Energy Storage System Connected to the Grid. Energy Procedia, 36, 1147–1157. https://doi.org/10.1016/j.egypro.2013.07.130 Taylor, J. W., & Buizza, R. (2002). Neural network load forecasting with weather ensemble predictions. IEEE Transactions on Power Systems, 17(3), 626–632. https://doi.org/10.1109/TPWRS.2002.800906 Tong, Z., Cheng, Z., & Tong, S. (2021). A review on the development of compressed air energy storage in China: Technical and economic challenges to commercialization. Renewable and Sustainable Ener- gy Reviews, 135(September 2019), 110178. https://doi.org/10.1016/j.rser.2020.110178 Trenberth Kevin E. (1990). Recent Observed Interdecadal Climate Changes in the Northern Hemisphere. Bulletin of the American Meteorological Society, 71(7), 988–993. Trianni, A., Cagno, E., Bertolotti, M., Thollander, P., & Andersson, E. (2019). Energy management : A practice-based assessment model. Applied Energy, 235(December 2018), 1614–1636. https://doi.org/10.1016/j.apenergy.2018.11.032 147 Ucar, A., & Balo, F. (2010). Assessment of wind power potential for turbine installation in coastal areas of Turkey. Renewable and Sustainable Energy Reviews, 14(7), 1901–1912. https://doi.org/10.1016/j.rser.2010.03.021 Uddin, M. S., & Kumar, S. (2014). Energy, emissions and environmental impact analysis of wind turbine using life cycle assessment technique. Journal of Cleaner Production, 69, 153–164. https://doi.org/10.1016/j.jclepro.2014.01.073 Uppala, S. M., KÅllberg, P. W., Simmons, A. J., Andrae, U., Bechtold, V. D. C., Fiorino, M., … Woollen, J. (2005). The ERA-40 re-analysis. Quarterly Journal of the Royal Meteorological Society, 131(612), 2961–3012. https://doi.org/10.1256/qj.04.176 Vijay M, D., Singh, B., & Bhuvaneswari, G. (2020). A High-Performance Microgrid with a Mechanical Sensorless SynRG Operated Wind Energy Generating System. IEEE Transactions on Industrial In- formatics, 16(12), 7349–7359. https://doi.org/10.1109/TII.2020.2976906 Wang, B., Ke, R.-Y., Yuan, X.-C., & Wei, Y.-M. (2014). China׳s regional assessment of renewable energy vulnerability to climate change. Renewable and Sustainable Energy Reviews, 40, 185–195. https://doi.org/10.1016/j.rser.2014.07.154 Wang, J., Qin, S., Zhou, Q., & Jiang, H. (2015). Medium-term wind speeds forecasting utilizing hybrid models for three different sites in Xinjiang, China. Renewable Energy, 76, 91–101. https://doi.org/10.1016/j.renene.2014.11.011 Wang, P., Fan, E., & Wang, P. (2021). Comparative analysis of image classification algorithms based on traditional machine learning and deep learning. Pattern Recognition Letters, 141, 61–67. https://doi.org/10.1016/j.patrec.2020.07.042 Wang, R., Li, C., Fu, W., & Tang, G. (2020). Deep Learning Method Based on Gated Recurrent Unit and Variational Mode Decomposition for Short-Term Wind Power Interval Prediction. IEEE Transac- tions on Neural Networks and Learning Systems, 31(10), 3814–3827. https://doi.org/10.1109/TNNLS.2019.2946414 Wang, X. Y., Mahinda Vilathgamuwa, D., & Choi, S. S. (2008). Determination of Battery Storage Capaci- ty in Energy Buffer for Wind Farm. IEEE Transactions on Energy Conversion, 23(3), 868–878. https://doi.org/10.1109/TEC.2008.921556 Wu, J., Zhang, B., Li, H., Li, Z., Chen, Y., & Miao, X. (2014). Statistical distribution for wind power forecast error and its application to determine optimal size of energy storage system. International Journal of Electrical Power & Energy Systems, 55, 100–107. https://doi.org/10.1016/j.ijepes.2013.09.003 Xydis, G. (2013). On the exergetic capacity factor of a wind – solar power generation system. Journal of Cleaner Production, 47, 437–445. https://doi.org/10.1016/j.jclepro.2012.07.014 Yang, H., Lu, L., & Zhou, W. (2007). A novel optimization sizing model for hybrid solar-wind power generation system. Solar Energy, 81(1), 76–84. https://doi.org/10.1016/j.solener.2006.06.010 148 Yang, M., Patiño-Echeverri, D., & Yang, F. (2012). Wind power generation in China: Understanding the mismatch between capacity and generation. Renewable Energy, 41, 145–151. https://doi.org/10.1016/j.renene.2011.10.013 Zahedi, A. (2010). Australian renewable energy progress. Renewable and Sustainable Energy Reviews, 14(8), 2208–2213. https://doi.org/10.1016/j.rser.2010.03.026 Zahedi, Ahmad. (2014). Sustainable Power Supply Using Solar Energy and Wind Power Combined with Energy Storage. Energy Procedia, 52, 642–650. https://doi.org/10.1016/j.egypro.2014.07.120 Zhang, D., Qian, L., Mao, B., Huang, C., Huang, B., & Si, Y. (2018). A Data-Driven Design for Fault Detection of Wind Turbines Using Random Forests and XGboost. IEEE Access, 6, 21020–21031. https://doi.org/10.1109/ACCESS.2018.2818678 Zhang, J., Chowdhury, S., & Messac, A. (2014). A comprehensive measure of the energy resource: Wind power potential (WPP). Energy Conversion and Management, 86, 388–398. https://doi.org/10.1016/j.enconman.2014.04.083 Zhang, Y., Pan, G., Zhao, Y., Li, Q., & Wang, F. (2020). Short-term wind speed interval prediction based on artificial intelligence methods and error probability distribution. Energy Conversion and Man- agement, 224(June), 113346. https://doi.org/10.1016/j.enconman.2020.113346 Zhao, P., Dai, Y., & Wang, J. (2014). Design and thermodynamic analysis of a hybrid energy storage sys- tem based on A-CAES (adiabatic compressed air energy storage) and FESS (flywheel energy stor- age system) for wind power application. Energy, 70, 674–684. https://doi.org/10.1016/j.energy.2014.04.055 Zhao, P., Wang, M., Wang, J., & Dai, Y. (2015). A preliminary dynamic behaviors analysis of a hybrid energy storage system based on adiabatic compressed air energy storage and flywheel energy stor- age system for wind power application. Energy, 84, 825–839. https://doi.org/10.1016/j.energy.2015.03.067 Zhao, X., Wang, C., Su, J., & Wang, J. (2019). Research and application based on the swarm intelligence algorithm and artificial intelligence for wind farm decision system. Renewable Energy, 134, 681– 697. https://doi.org/10.1016/j.renene.2018.11.061 Zhao, Z.-Y., Zhu, J., & Zuo, J. (2014). Sustainable development of the wind power industry in a complex environment: a flexibility study. Energy Policy, 75, 392–397. https://doi.org/10.1016/j.enpol.2014.10.017 Zhou, J., Shi, J., & Li, G. (2011). Fine tuning support vector machines for short-term wind speed forecast- ing. Energy Conversion and Management, 52(4), 1990–1998. https://doi.org/10.1016/j.enconman.2010.11.007 Zhou, W., Yang, H., & Fang, Z. (2006). Wind power potential and characteristic analysis of the Pearl River Delta region, China. Renewable Energy, 31(6), 739–753. https://doi.org/10.1016/j.renene.2005.05.006 149 150 Appendix 3. Paper 3: Meteorological information service support system in wind park application Liu, Y. and Yang, W. (2015): Meteorological information service support system in wind park application. International journal of production economics, no. 2, pp. 222-237. Jufo-category of the publication 1, year 2015, and impact factor 2.60. Article has ben published in a peer-reviewed publication 2015.03.02. Yang Liu: Defining of the research design, acquisition of research resources, program- ming, guidance in the research process, verification and analysis of findings, editing the article at different stages. Wenshan Yang: Collecting and managing research material, data analysis, methodology and research methods, visualization of results, writing the body text of the original arti- cle. 151 Meteorological information service support sys- tem in wind park application Abstract Purpose – This paper introduces a holistic decision support system based on condi- tion-based maintenance which utilizes meteorological forecasting information to sup- port decision-making process in services of wind power enterprises. Design/methodology/approach – A pilot conceptual system combining with meteoro- logical information and operations management has been formulated in this study. The proposed system provides benchmarking to support decision-making directly and indi- rectly basing on processing meteorological information and evaluating its impact on service operations. It collects meteorological data to predict failure probabilities in different areas which need corresponding maintenance service and schedule the opti- mal maintenance periods. In addition, it provides meteorological forecasting and deci- sion support in case of extreme weather events. Findings – The conceptual study shows that there is a connection between the meteor- ological conditions and failures, and it is feasible to make service decisions based on the predictions of weather conditions and their impacts to failures. Research limitations/implications – The research presented at the present phase is not much beyond a conceptual framework. The actual implementation and all possible related practical issues will be dealt with in future research. Practical implications – It helps decision-makers to predict and identify possible cat- egories of faults in wind turbine, make optimal service decisions to enhance the output performance of wind power generation, and take in advance emergency counteractions in case of extreme weather events. Originality/value – It presents a novel concept and provides a roadmap to achieve op- timal operations in wind park application through combining meteorological infor- mation system with service decision-making. Keywords – information system, meteorological service, extreme weather events, de- cision-making, strategic management, condition-based maintenance. Paper type – Research paper 1. Introduction 152 According to Global Wind Energy Council (2012), the number of wind turbines spinning around the world by the end of 2011 is 199,064. Among that, the amount of wind turbines up and running in China is 45,894 which count 23% of the total amount. That is to say, China’s leadership in wind energy deployment is both an opportunity and a challenge for European and American companies to compete in this market and internationally. Europe remains a technology leader and is carving out the next frontier of wind energy with onshore and offshore deployments (Global Wind Energy Council, 2012). The wind energy potential of the Earth is huge and enough, in principle, to meet all the world’s electricity needs. Virtually every country has sites with average wind speeds of more than 5 m/s measured at a height of 10 m, which are sufficient for using wind power to generate electricity (Sesto and Ancona, 1995). Many of critical wind turbine faults are directly or indirectly related to weather condi- tions and extreme weather events (EWEs). This research intends to propose a pilot ser- vice support system which utilizes meteorological information to predict such situations which may lead to breakdowns and make it possible to take precautions in advance, and in addition to suggest other service related decisions based on condition-based mainte- nance (CBM), such as deciding the optimal time for maintenance during the predicted idle period. CBM is defined by a set of maintenance actions taken as a consequence of knowing the current operating status of equipment. Recent study considers it is a form of proactive equipment maintenance that forecasts incipient failures based on a real- time assessment of equipment condition obtained from embedded sensors and or exter- nal tests and measurements that are extracted directly from the equipment (Gulledge, Hiroshige and Iyer, 2010). Many recent studies show there is direct connection between service and business performance in wind power systems and demonstrate business po- tential analysis that optimal service decisions based on CBM in wind park application can significantly cut down operation and maintenance costs (El-Thalji and Jantunen, 2012; Tian et al., 2011; Nielsen and Sørensen, 2011), and by implementing a successful CBM strategy can also achieve higher level of cost effectiveness (El-Thalji and Jan- tunen, 2012), thus improve the operation and business performance. In many recent studies, the relationship between wind speed modelling and electricity generation from wind turbines is also studied. In fact, wind park investors are interested in long-range forecasts and simulation of wind speed for two main reasons: to evaluate 153 the profitability of building a wind farm in a given location, and to offset the risks asso- ciated with the variability of wind speed for an already operating wind farm (Caporin and Preś, 2012). The percentage of the world’s electricity that could be produced from offshore devices is estimated to be around 7% by 2050, and this would employ a signif- icant amount of people by this time, possibly around 1 million, mostly in the mainte- nance of existing installations (Esteban and Leary, 2012). Optimal maintenance are affected by various factors, such as availability of re- sources, dependency on meteorological surrounding conditions, as well as a complex logistical process chain (Tracht, Westerholt and Schuh, 2013), and failure probability can be predicted based on condition monitoring data of wind energy systems (Tracht et al., 2013). Based on these theories, this paper develops a holistic system which com- bines with meteorological information and operations management. The proposed sys- tem provides benchmarking to support decision-making basing on processing meteoro- logical information and evaluating its impact in service operations of wind power en- terprises. The concept of such decision support system is built based on years of well- established previous studies utilizing sense and respond type of continuous adjust- ments in decision-making to achieve sustainable competitive advantage in operations strategy implementation (Liu, 2013). The structure of this paper is as follows. Section 2 reviews the latest related studies. Section 3 introduces the research methodology. Section 4 describes the system structure and process in a conceptual framework. Section 5 discusses the managerial implications, research limitations and also recommendations for future research. Section 6 draws con- clusions. 2. Review of related studies 2.1 Wind power as an energy source Energy is the main intermediate strategic resource for economic development and growth in any country. This usually translates to better quality of life, and therefore it leads to higher primary energy consumption in all sectors, transportation, industry, services, household, etc. (Abulfotuh, 2007). Nowadays, the world faces a great chal- lenge of saving our future in terms of developing renewable energy. Until now, a huge amount of the energy requirements all over the world is supplied originally from con- ventional energy sources like coal, crude oil, natural gas, etc. However, these patterns 154 of energy are limited and often lead to pollution. Therefore, renewable energy re- sources will play an important role in our daily life in the world’s future. Renewable energy sources are those resources which can be used to produce energy again and again, e.g. solar energy, wind energy, biomass energy, geothermal energy, etc. and are also often called alternative sources of energy (Rathore and Panwar, 2007). Among the renewable energy sources wind energy is currently viewed as one of the most significant and attractive sources, which is a clean energy rather than coal, crude oil and natural gas. The outstanding characteristic of wind power is to save energy and protect environment although the intermittent character is a very critical problem. The use of renewable energy sources is closely linked to sustainable development, because a sustainable supply of energy resources which must be used effectively and efficiently is required for it, as well as for progressing in environmental problems (To- lón-Becerra, Lastra-Bravo and Bienvenido-Bárcena, 2011). It is undoubtedly that sus- tainable development will definitely let managers handle with problems during the period of decision-making. On one hand, wind power generation is becoming more and more popular in many countries, but it differs from conventional thermal generation due to the stochastic na- ture of wind. Thus wind power forecasting plays a key role in dealing with the chal- lenges of balancing supply and demand in any electrical system, given the uncertainty associated with the wind park power output. Accurate wind power forecasting reduces the need for additional balancing energy and reserve power needed to utilize wind power (Foley et al., 2012). On the other hand, the Nordic countries particularly experienced a number of ex- treme weather events (EWEs) during recent years and a significant number of wind power businesses were affected as a result. With the intensity and frequency of ex- treme weather predicted in the future, enhancing the resilience of businesses, especial- ly wind power enterprises (WPEs) which are considered as highly vulnerable, has be- come necessary (Wedawatta et al., 2011). However, little research has been undertaken on how construction of WPEs is responding to the risk of EWEs. 2.2 Meteorological service and decision-making Traditional maintenance techniques, such as preventive maintenance is scheduled in advance of failure and usually at regular intervals which are typically determined by 155 the analysis of historical reliability data (Gulledge, Hiroshige and Iyer, 2010), and time-based maintenance (TBM) is labour intensive, ineffective in identifying problems that develop between scheduled inspections, and not cost-effective (Ahmad and Kama- ruddin, 2012). Whereas CBM is scheduled by predicting the future status of the equipment based on operational or other characteristics. (Gulledge, Hiroshige and Iyer, 2010). Recent studies develop optimal CBM strategy and decision for wind power applications systems (Tian et al., 2011; Nielsen and Sørensen, 2011; El-Thalji and Jan- tunen, 2012). CBM is more efficient compared to preventive maintenance in many ways, e.g. condition monitoring and diagnostic practices have become significantly important part of offshore wind farms in order to cut down operation and maintenance costs (El-Thalji and Jantunen, 2012), and more realistic and worthwhile to apply than time-based maintenance (Ahmad and Kamaruddin, 2012). However, in wind park ap- plication it is typically hard to accurately predict with standalone meteorological data and may lead to a failure prediction. The other challenge is to enable CBM strategy to provide maintenance decisions and services at the right time i.e. maintenance is per- formed when it is needed and not too early and not too late i.e. causing breakdown and downtime (El-Thalji and Jantunen, 2012). Therefore a holistic system combining the complete meteorological service and decision-making is needed to increase the predic- tion accuracy and work together with the traditional preventive/corrective measures to provide optimal maintenance decisions. Short-term prediction is mainly oriented to the spot (daily and intraday) market, sys- tem management and scheduling of some maintenance tasks, being of interest to sys- tem operators, electricity companies and wind park promoters (Costaa et al., 2008). Wind forecasting for energy generation and power system operations mainly focuses on the immediate short-term of seconds to minutes, the short-term of hours up to two days, and the medium term of two to seven days. This is because power system opera- tions such as regulation, load following, balancing, unit commitment and scheduling, are carried out within these time frames. The science of wind power prediction is de- scribed as the application of the theories and practices of both meteorology and clima- tology specifically to wind power generation (Petersen et al., 1997). - Numerical weather prediction 156 - In case of non-saturated power, because the wind power is equal to wind speed third cube and wind speed are much more regular than that of wind power, consequently a small wind speed error will amplify wind power errors much. It is wide and effective by using short-term wind power forecasting methods which combining numerical weather prediction (NWP) model with statistical models, so that we can develop operating mode for electric grid dispatching, provide support for arranging dispatch rationally, reduce the effects of intermit- tent power to wind power systems effectively. The wind data from now, yester- day, or last year in the same period cannot be used to predict wind in the next 24 hours, because wind is dependent on the weather, and the wind power out- put cannot be guaranteed at any particular time. Thus the integration of wind power into electrical grids can cause difficulties in the management of the power system (Marciukaitis, Katinas and Kavaliauskas, 2008).Meteorological service Climate change is predicted to have a significant effect on the frequency of EWEs and the occurrence of natural disasters, such as hail, flood, tornado and thunderstorm. There is a need for facility managers to mitigate potential disruption and prepare for future events caused by natural phenomenon. Meteorological sector sends out early warnings to WPEs, using the results of real-time monitoring and weather forecasting from satellite, radar, observation stations. The meteorological ensuring system is a derivative product which mainly involving EWE forecasting and warnings. This can prevent and mitigate climate change on cru- cial facilities and the impact of the project effectively. In current practice, however, that little risk assessment is undertaken by few organizations preparing integrated dis- aster management plans or business continuity plans to help them meet the challenge (Warren, 2010). As we learn more about possible climate change impacts, certain WPE protection strategies may become more desirable and feasible in management, and we can adopt strategies to minimize its negative impacts on wind power generation. After studying how climate will change by predictions with wind power production and provide guidance in facing of EWEs, then facility managers can prepare for risk assessment and disaster plans after collecting scientific data related to the potential effects of climate change (Warren, 2010). 157 - Decision-making There is relatively little research in the area of operations and service management in renewable energy sector such as wind power by utilizing meteorological information. Some notable studies connecting meteorological forecasting with renewable energy include e.g. Kaplan & Norton (2011) and Eckman & Stackhouse (2012). Changes in competitive environments have increased the importance of strategic management in corporations. Successful companies must be able to anticipate changes in operating en- vironments and be able to react faster than their competitors (Kaplan and Norton, 2011). Earth observations are critical in enhancing the implementation of renewable energy technologies and improving energy efficiency (Eckman and Stackhouse, 2012). Other related research has been implemented by Liu, et al. (2012). According to the research from this group, they proposed a novel wind turbine fault diagnostic method based on the local mean decomposition technology, which is a new iterative approach to demodu- late amplitude and frequency modulated signals, which is suitable for obtaining instan- taneous frequencies in wind turbine condition monitoring and fault diagnosis. Finally, the experimental analysis of the wind turbine vibration signal proves the validity and availability of the new method (Liu et al., 2012). Our research presented in this paper addresses this problem from a conceptual level to bridge the gap between meteorological information and decision-making in service operations management. Even though this whole concept is a huge research which is still in progress, nevertheless this paper can be a pilot which leads to new ideas and opens more research paths. 3. Research methodology 3.1 Overview There are various types of strategies for conducting research in management and social sciences. Reisman (1988) defines research strategies such as ripple, embedding, bridg- ing, transfer of technology, creative application, structuring, and empirical validation. This study uses mainly the following research strategies. Ripple is used to develop analytical models for assessing failure probabilities based on meteorological infor- mation and NWP. Embedding and bridging are used to associate the decision-making process in connection with the service needs which are based on the failure forecasts. Empirical validation is used to validate the developed theories by performing various 158 case studies in different countries. Arbnor & Bjerke (1997) introduce three methodo- logical approaches i.e. analytical, systems, and actors. The nature of this study is to create a holistic system which is a set of components and the relations among them. Holweg (2005) applies the systems approach and contingency theory to review exist- ing contributions and synthesizes them into a conceptual model, which is very similar to the nature of this work. Therefore, systems and contingency based methodological approach is proposed to carry out this work. As the main contribution of this study is the integration of meteorological information with decision-making in service opera- tions, it requires a new design in the research methodology to integrate the classic components. Kasanen, et al. (1993) describe the constructive approach as “problem- solving through the construction of organizational procedures or models”, and also propose a market-based validation for assessing this aspect of a construction. In this work the construct is the integrative holistic system and it is feasible to apply a weak market test to validate and implement the research objectives. In summary, the re- search methods include literature survey, descriptive conceptual analysis, analyzing qualitative data based on Silverman (2001) and also quantitative data, classification by simple statistics, and finally using Kasanen et al.’s (1993) the constructive research approach with weak market tests and pilots for implementation. 3.2 Case study To achieve the entire objectives of this conceptual research, the empirical studies are important and numerous case studies should be carried out from different countries, and analyzing them with the proposed existing analytical models and creating new analytical models for further evaluation. Therefore, the selection of the case companies must be mostly representative wind park applications. The case studies will be carried out in future research. 3.3 Data collection and analysis The data of cases in different countries are collected in the same manner: by asking senior managers or directors to answer the questionnaires. The interviewees are nor- mally decision makers and middle management groups, who have good knowledge about the operations of their own wind parks. The interviewed high competence ex- perts should be representative to know well the operations of the studied wind park. The data collected typically from limited and described application problems is mainly 159 qualitative in nature and its validity and reliability can be ensured by improving the required careful documentation of the cases (Sykes, 1990; 1991). Firstly, the managers or directors are trained to understand every item of the questionnaires correctly by interview, email or telephone. Secondly, after they finish the questionnaires, the an- swers are analyzed with software. Thirdly, the discussion with the managers or direc- tors reveals the results and verifies the validity and reliability of the data further. 4. System description This section develops a conceptual framework for service support in wind park applica- tion. The proposed system involves 3 major modules: meteorological information mod- ule, wind power prediction module and operations management decision-making mod- ule. The complete system structure and process are illustrated in Figure 1. 4.1 Meteorological information module This module includes meteorological data collection and NWP. First, meteorological data are collected by the wind speed sensor, wind direction sensor, temperature sensor, atmospheric pressure sensor, humidity sensor etc. installed on the wind-testing tower of the targeted wind park. Through wireless communicating module, original meteorologi- cal data from the wind-testing tower is converted into a digital signal and finally trans- mitted to the receiving terminal. Then NWP processes the meteorological data to pa- rameters related to wind power output. NWP is a special version tailored to predicting wind power output and is different from the version used for commercial public weather forecasts. On the other hand, it also sends early warning messages in case of EWEs and managers in WPEs can get alarming signals in advance and take countermeasures quickly. With the new forecasting system it effectively links up the NWP model geared towards very short-range forecast of severe weather system. Currently, the suit is proba- bly one of the few operational forecasting systems that effectively combine radar infor- mation, dense mesoscale NWP model prognoses for real-time EWEs risk assessment. The following example illustrates how this is done in reality. A mountain area site locat- ed in central China has been chosen to test the proposed theory. The site is located near- by a wind park in operation, also including a meteorological station with anemometers between 30 and 70 m. This wind-testing tower has been brought into operation since November 2011. The mountain top has a height of 700 meters and has a direct distance of 34 kilometers to the local meteorological station, which has good correlation to pre- 160 dict the weather conditions for the wind park. The common meteorological disasters in this location are thunderstorm, flood, drought, low-temperature freeze and continuous rain. - Measuring the maximum wind speeds in the meteorological station The meteorological station is used to predict the 50-year wind base on annual average 10-minute maximum wind speeds. Through T-test to inspect the consistency for se- quence of annual maximum wind speed from 1974 to 2011, it has been discovered that the values in 1982 experienced a mutation. It is necessary to correct references of max- imum wind speeds from 1982 to 2011due to the diversion of the meteorological station. According to National Wind Energy Resource Evaluation Technology Provision and the type I extreme value distribution, the 50-year average 10-min maximum wind speed is 28.38 m/s. - Predicting the 50-year maximum wind speed in the wind park 𝑉50_𝑚𝑎𝑥 = 𝑢 − 1 𝛼 𝑙𝑛 𝑙𝑛 ( 50 50−1 )] (1) 𝜇 = 1 𝑛 ∑ 𝑛𝑖=1 𝑉𝑖 (2) 𝜎 = √ 1 𝑛−1 ∑ 𝑛𝑖=1 (𝑉𝑖 − 𝜇) 2 (3) 𝛼 = 𝐶1 𝜎 (4) 𝑢 = 𝜇 − 𝐶2 𝛼 (5) Using Eqs. (1)-(5), it can be calculated that the 50-year maximum wind speed is 30.5 m/s. - Predicting the 50-year extreme wind speed in the wind park Gust factors are a ratio between a peak wind speed of some duration within a given data segment and the mean wind speed of the same segment. An optimum gust factor of about 1.4 is suggested for all types of fabric structures in general, which is an interna- 161 tional standard value. Then it can be calculated that 50-year extreme wind speed is 42.7 m/s. 4.2 Wind power prediction module This module utilizes the processed meteorological data to predict patterns of values re- lated to wind power output. It is an intermediate process to obtain parameters to evalu- ate failure probabilities and calculate the optimal service decisions which are crucial information for the next process - operations management decision-making. Figure 1. System structure and process The wind power prediction module calculates the predicted amount of power output during particular hours and days based on the real-time meteorological data. In order to have an accurate prediction, short-term weather forecast is important for the dynamic control of wind turbine and for minimizing the scheduling errors which impact on grid reliability and market service costs (Lerner, Grundmeyer & Garvert 2009). Depending on their inputs, the forecast models are classified as physical or statistical or hybrid ap- proaches. The best way is to use meteorological forecast data from NWP systems com- bining several prediction techniques (Giorgi, Ficarella & Tarantino 2011). 162 The module involves real-time wind measuring, NWP, and wind power forecasting. WPEs establish the forecasting model based on NWP and historical data related to wind, and they participate in prediction and report survey to dispatch center on time. Whether using ultra-short term wind power forecasting or long term wind power forecasting, they are all based on the foundation of real-time wind measuring data. According to predicting and actual wind speed in wind parks, a mixed model of time series method and back-propagation neural networks arithmetic combining with mete- orological data can be used in wind power prediction. The more accuracy of meteoro- logical data is, the better forecasting results can be obtained. In addition, meteorological ensuring service can be provided to wind parks in the meantime. This wind power predicting system mainly includes five parts: (i) data collection; (ii) NWP; (iii) wind power prediction; (iv) graphical user interface (GUI) software; (v) pre- dicting database. The structure is shown in Figure 2. Figure 2. Structure of wind power prediction module - Data collection aims to select a site to set up wind-testing tower and collect wind speed, wind direction, temperature, pressure etc. which are also the input varia- bles of the wind power prediction module. - NWP incorporates information representing the outer scale geophysical variabil- ity through evolving boundary conditions and by assimilating observations of the current state of the atmosphere to predict flow characteristics. In this re- search, NWP is responsible for dealing with all collected meteorological refer- ences, adapting rational mathematical models to calculate the results of future weather. 163 - Wind power prediction mainly focuses on predicting wind speed and wind pow- er output which are crucial information for the next decision-making process. Wind speed and wind direction are the most important variables. Wind power is equal to wind speed third cube and wind speed are much more regular than that of wind power, and therefore it requires accurate wind measurement. - GUI Software deals with data transforming and interactive interfaces. 4.3 Operations management decision-making module This module mainly utilizes the prediction data from previous process to evaluate fail- ure probabilities in different parts of the wind turbine and calculate the optimal service decisions for the wind park operations management. The significance of failure analysis and fault diagnosis for wind turbine results lower breakdown rate, reduced maintenance cost and time, and improves the operational effi- ciency and reliability (Ma, He and Feng, 2012). The wind turbine is a complex system which transforms kinetic energy from wind power to electrical power. (Kostandyan and Sørensen, 2012). It consists of electrical, mechanical, hydraulic, structural, and software subsystems. Many of critical wind turbine faults are directly or indirectly related to weather condi- tions and EWEs. Analysis to weather related faults can reveal the causes which can be even predicted, since the weather conditions resulting faults can be predicted with me- teorological information system, making it possible to take precautions in advance to prevent such situations from happening. Other service decisions such as the optimal time for maintenance during idle period can be also predicted and scheduled in advance basing on meteorological information. Statistics show that the determining time for the fault diagnosis takes up 70% to 90% of the total time, while the repair time takes up only about 10% to 30% (Wang and Fent, 2004). A wind turbine can be unavailable because of planned maintenance activities or because of unforeseen failures, incidents or accidents. Analysis of predictable sources of wind turbine failures such as weather conditions can help a lot in decision-making to optimize maintenance schedule and maximizes wind power output. Each component has different physics of failure behavior depending on structure, shape, operational environment and many other parameters (Kostandyan and Sørensen, 2012). From the current structural characteristics of wind turbine shown in Figure 3 and its 164 actual fault conditions, faults usually occur in parts such as gears, shafts, bearings, fas- tener and box. 165 Figure 3. Main structure of wind turbine According to real case statistics, Table 1 shows typical fault diagnosis related to weather conditions and the relevant actions need to be taken. The decision model can be built based on failure probabilities according to these conditions shown in Table 1. Table 1. Typical failures related to weather conditions Failure parts Possible reasons Weather conditions Actions Blade Blade drive not ready EWEs Emergency stop Rotor Result of imbalance, blades and hub corrosion etc., brake sensor failure Rain, snow and other hash meteorological condition Normal stop Gearbox Over temperature, gearbox oil pressure too low High temperature Normal stop Generator Over speed, over tempera- ture, bearing faults, current too high/low, frequency sen- sor failure High temperature and / or humidity Emergency stop Normal stop Yaw system Yaw brake set unintentionally Extreme changes in wind speed / direction Normal stop Tower Weather or other failure may cause excessive vibration EWEs Emergency stop 5. Discussion and future study Based on collecting official documents, analytical results, lab experiments, and hypoth- esis test result, this investigation discuss the possible causes of wind power system fail- ure from these four perspectives and presents practical suggestions for wind tower risk management and future action plans for the areas of structural design evaluation, con- struction and quality management, and engineering document review. By addressing study recommendations, project stakeholders can improve their risk management strate- gies. Construction firms can also utilize these findings to learn lessons for future refer- ence. In terms of risk management, identifying the major causes of failure, one must understand the risk associated with these causes, and generate action plans that allow project managers to mitigate risk or employ control measures (Chou and Tu, 2011). 166 In addition to the conceptual design, this study has provided new insight for practi- cal operations management in wind park application. It helps decision-makers to pre- dict and identify possible categories of faults in wind turbine and make optimal service decisions to enhance the output performance of wind power generation. Further research is needed related to sensitivity analysis of: (1). Wind surveys and installations have so far concerned mostly onshore sites. Howev- er, a very interesting wind potential seems to exist also in offshore, shallow water loca- tions, where there is the advantage of better wind conditions and less environmental restrictions, although the disadvantage of more difficult access and higher installation and maintenance costs must be taken into account (Sesto and Claudio, 1998). In that situation, the seawater salinity is one critical meteorological factor which will be studied in future research. (2). Accumulated plastic strain depending on the temperature Mean and temperature range factors. The proposed model is useful to predict damage values for solder joint in power electrical components. However, the real test data are required for the accurate model parameter estimation. (3). In addition, operation and maintenance strategies might be developed based on the proposed approach. Especially strategies for renewable and replacement systems, where reliability updating might be implemented based on failure times. 6. Conclusions This paper develops a conceptual system which utilizes the meteorological infor- mation for decision-making based on CBM in operations and service management for wind parks, which is a form of proactive equipment maintenance that forecasts incipient failures based on a real-time assessment of various external and internal conditions ob- tained from e.g. meteorological data and equipment monitoring system etc. The objec- tive is to design an optimal service decision-making system based on CBM in wind park application to significantly cut down operation and maintenance costs and also imple- ment a successful CBM strategy to achieve higher level of cost effectiveness, thus im- prove the operation and business performance. This paper bridges the gaps in current research of this area and opens up new research paths in the development of forecasting practices for service related decision-making, operations and risk management of EWEs in wind parks. It has shown that through the analysis of the meteorological information, 167 it is possible to predict hash weather conditions which are harmful to cause faults in wind turbines. Modern NWP can provide reliable forecasts for wind parks as accurate as per quarter hour basis in the next couple of hours and also useful trend forecast up to days. By analyzing the approximate time-period of the 50-year maximum wind speed and extreme wind speed through EWEs forecasting, WPEs can effectively reduce and even avoid a huge number of losses in maintenance, and schedule service operations in more optimal periods. The basic idea has been already tested in a wind park in central China as depicted, but still lacks of systematic theory construction to be used as a deci- sion support system. The implementation of this conceptual model will be dealt with in future research. References Abulfotuh, F. (2007) 'Energy efficiency and renewable technologies: the way to sustain- able energy future', Desalination, vol. 209, pp. 275-282. Ahmad, R. and Kamaruddin, S. (2012) 'An overview of time-based and condition-based maintenance in industrial application', Computers & Industrial Engineering, vol. 63, no. 1, pp. 135-149. Arbnor, I. and Bjerke, B. (1997) Methodology for creating business knowledge, 2nd edi- tion, Thousand Oaks: Sage. Caporin, M. and Preś, J. (2012) 'Modelling and forecasting wind speed intensity for weather risk management', Computational Statistics and Data Analysis, vol. 56, pp. 3459-3476. Chou, J.S. and Tu, W.T. (2011) 'Failure analysis and risk management of a collapsed large wind turbine tower', Engineering Failure Analysis, vol. 18, p. 295–313. Chung, C.Y., Chung, H.Y. and Ou, S.C. (2011) 'Computer Simulation Applied to a Bio- mathematic Model for Tuberculosis', Procedia Engineering, vol. 15, p. 3626 – 3631. Costaa, A., Crespob, A., Navarroa, J., Lizcanoc, G., Madsend, H. and Feitosae, E. (2008) 'A review on the young history of the wind power short-term prediction', vol. 12, pp. 1725-1744. Eckman, R.S. and Stackhouse, P.W. (2012) 'CEOS contributions to informing energy management and policy decision making using space-based earth observations', Ap- plied Energy, vol. 90, pp. 206-210. 168 El-Thalji, I. and Jantunen, E. (2012) 'On the Development of Condition Based Mainte- nance Strategy for Offshore Wind Farm: Requirement Elicitation Process', Energy Pro- cedia, vol. 24, pp. 328-339. Esteban, M. and Leary, D. (2012) 'Current developments and future prospects of off- shore wind and ocean energy', Applied Energy, vol. 90, p. 128–136. Foley, A.M., Leahy, P.G., Marvuglia, A. and McKeogh, E.J. (2012) 'Current methods and advances in forecasting of wind power generation', Renewable Energy, vol. 37, pp. 1-8. Global Wind Energy Council (2012) Global Wind Energy Outlook 2012, [Online], Availa- ble: http://www.gwec.net/wp-content/uploads/2012/11/GWEO_2012_lowRes.pdf [11 November 2012]. Gulledge, T., Hiroshige, S. and Iyer, R. (2010) 'Condition-based Maintenance and the product improvement process', Computers in Industry, vol. 61, no. 9, pp. 813-832. Holweg, M. (2005) 'The three dimensions of responsiveness', International Journal of Operations & Production Management, vol. 25, no. 7, pp. 603-622. Kaplan, R.S. and Norton, D.P. (2011) The Strategy-focused Organization: How Balanced Scorecard Companies Thrive in the New Business Environment, Boston,MA: Harvard Business School Press. Kasanen, E., Lukka, K. and Siitonen, A. (1993) 'The constructive approach in manage- ment accounting research', Journal of Management Accounting Research, vol. 5, no. Fall, p. 243–264. Kostandyan, E. and Sørensen, J.D. (2012) 'Physics of failure as a basis for solder ele- ments reliability assessment in windturbines', Reliability engineering and system safety, vol. 108, pp. 100-107. Liu, Y. (2013) 'Sustainable competitive advantage in turbulent business environments', International Journal of Production Research, vol. 51, no. 10, pp. 2821-2841. Liu, W.Y., Zhang, W.H., Han, J.G. and Wang, G.F. (2012) 'A new wind turbine fault diag- nosis method based on the local mean decomposition', Renewable Energy, vol. 48, pp. 411-415. Ma, Y., He, C.B. and Feng, X.X. (2012) 'Institutions function and failure statistic and analysis of wind turbine', Physics procedia, vol. 24, pp. 25-30. 169 Marciukaitis, M., Katinas, V. and Kavaliauskas, A. (2008) '(2008). Wind power usage and prediction prospects in Lithuania', Renewable and sustainable Energy Reviews, vol. 12, no. 1, p. 265–277. Nielsen, J.J. and Sørensen, J.D. (2011) 'On risk-based operation and maintenance of offshore wind turbine components', Reliability Engineering & System Safety, vol. 96, no. 1, pp. 218-229. Petersen, E.L., Mortensen, N.G., Landberg, L., Højstrup, J. and Frank, H.P. (1997) Wind power meteorology, Roskilde, Denmark: Risø National Laboratory. Rathore, N.S. and Panwar, N.L. (2007) Renewable energy sources for sustainable devel- opment, New Delhi, India: New India Publishing Agency. Razmi, J., Ghaderi, S.F., Zairi, M. and Keyno, H.R.S. (2008) 'Emerald Article: Compiling and prioritizing strategies for electrical', Benchmarking, vol. 15, no. 6, pp. 794-804. Reisman, A. (1988) 'On alternative strategies for doing research in the management and social sciences', IEEE Transactions on Engineering Management, vol. 35, no. 4, pp. 215-220. Sesto, E. and Ancona, D.F. (1995) Present and prospective role of wind energy in elec- tricity supply, in: Proc. of New Electricity 21, Paris, Paris, France: OECD Publications. Sesto, E. and Claudio, C. (1998) 'Exploitation of wind as an energy source to meet the world's electricity demand', Journal of Wind Engineering and Industrial Aerodynamics, vol. 76, no. 74, pp. 375-387. Silverman, D. (2001) Interpreting qualitative data: methods for analysing talk, text and interaction, 2nd edition, London: Sage. Sykes, W. (1990) 'Validity and reliability in qualitative market research: a review of the literature', Journal of the Market Research Society, vol. 32, no. 3, pp. 289-328. Sykes, W. (1991) 'Taking stock: issues from the literature on validity and reliability in qualitative research', Journal of the Market Research Society, vol. 33, no. 1, pp. 3-12. Tian, Z., Jin, T., Wu, B. and Ding, F. (2011) 'Condition based maintenance optimization for wind power generation systems under continuous monitoring', Renewable Energy, vol. 36, no. 5, pp. 1502-1509. 170 Tolón-Becerra, A., Lastra-Bravo, X. and Bienvenido-Bárcena, F. (2011) 'Proposal for terri- torial distribution of the EU 2020 political renewable energy goal', Renewable Energy, vol. 36, pp. 2067-2077. Tracht, K., Goch, G., Schuh, P., Sorg, M. and Westerkamp, J.F. (2013) 'Failure probability prediction based on condition monitoring data of wind energy systems for spare parts supply', CIRP Annals - Manufacturing Technology, vol. 62, no. 1, pp. 127-130. Tracht, K., Westerholt, J. and Schuh, P. (2013) 'Spare Parts Planning for Offshore Wind Turbines Subject to Restrictive Maintenance Conditions', Procedia CIRP, vol. 7, pp. 563- 568. Wang, Y. and Fent, X.Y. (2004) 'Electric locomotive fault diagnosis expert system based on fault tree.', Electric Locomotives & Mass Transit Vehicles, vol. 27, pp. 35-36. Warren, C.M.J. (2010) 'The facilities manager preparing for climate change related dis- aster', Facilities, vol. 28, no. 12, pp. 502-513. Wedawatta, G., Ingirige, B., Jones, K. and Proverbs, D. (2011) 'Extreme weather events and construction SMEs Vulnerability, impacts, and responses', Structural Survey, vol. 29, no. 2, p. 106–119. Wright, C.M. and Mechling, G. (2002) 'The importance of operations management problems in service organizations', The international journal of management science, vol. 30, pp. 77-87. Zajaczkowski, F.J., Haupt, S.E. and Schmehl, K.J. (2011) 'A preliminary study of assimilat- ing numerical weather prediction data into computational fluid dynamics models for wind prediction', Journal of Wind Engineering and Industrial Aerodynamics, vol. 99, pp. 320-329. 171 Appendix 4. Paper 4: A study on renewable energy potential based on the global atmospheric data Yang, W. and Liu, Y. (2015): A study on renewable energy potential based on the global atmospheric data. Global Cleaner Production & Sustainable Con- sumption Conference. It has been published in Global Cleaner Production & Sustainable Consumption Con- ference 1 November 2015 - 4 November 2015. Wenshan Yang: Defining of the research design, data analysis, methodology and research methods, verification and analysis of findings, writing the body text of the original arti- cle, editing the article at different stages. Yang Liu: Acquisition of research resources, programming, guidance in the research pro- cess, visulazation of results, editing the article at different stages. 172 Title: A study on renewable energy potential based on the global atmospheric data Authors & affiliations: Wenshan Yang1, Yang Liu*2,1 1University of Vaasa, Finland; 2Vaasa University of Applied Sciences, Finland yli@uva.fi Full Paper: (Your paper must use Normal style and must fit in this box. Your paper should be no longer than 2500-9000 words. The box will ‘expand’ as you add text/diagrams into it.) 173 Abstract According to U.S. EIA (Energy Information Administration) International Energy Statis- tics, nowadays fossil fuel is still the primary sources of energy, and the amount of fossil fuel power generation accounts around three-fifth in the world’s total annual electricity net generation. However, fossil fuel power generation based on coal, oil and natural gas is now gradually being substituted due to the limited availability and environmental aspects such as global warming and pollution. In response to the concerns of climate change, many policy makers are becoming keen to re-examine the use of fossil fuels and promote renewable energy. Hydro, wind, tide, photovoltaic, geothermal are com- mon renewable energy resources to generate electricity. In an effort to mitigate the pressure of burning fossil fuel on climate changes, it becomes more and more essential that renewable energy will eventually replace the conventional fuel. However, do we have sufficient renewable energy potential to replace conventional fuel and fulfil the world’s energy consumption demands? Since many types of renewable energy such as the wind, solar, hydro are directly or indirectly related to meteorological factors and largely affected by them, this study analyzes the existing ERA-20C global datasets de- scribing the state of the atmosphere as well as land-surface and ocean-wave conditions from 1900 to 2010 obtained from the European Centre for Medium-Range Weather Forecasts (ECMWF) and tries to identify the meteorological factors (wind speed, solar radiation, rainfall, evaporation etc.) with their effects on the overall utilization potential of these renewable energy resources. From this study, it can be found that a correlation between the meteorological factors and the renewable energy potential does exist, which implies forecasting models of renewable energy potential can be invented based on the global atmospheric data. The future study will focus on forecasting the global potential of these renewable energy resources in the next decades. Keywords: Renewable energy, climate change, reanalysis, ERA-20C data, resources potential. 1. Introduction 174 With an increasing energy demand, renewable energy is an appropriate way to satisfy energy consumption without environmental degradation (Hua, Oliphant, & Hu, 2016; A. Zahedi, 2010). Meanwhile, renewable energy technologies can play a crucial role in the transition towards a low-carbon economy (Albrecht, Laleman, & Vulsteke, 2015). Re- newable energy development is a major response to address the issues of climate change and energy security. The utilization of renewable resources, however, highly depends on the climate conditions, which may be impacted in the future due to global climate change. When discussing why climate changes occur and reduce its influence to human beings, it is essential to consider atmospheric dynamics rather than only focus on surface vari- ables, in particular, temperature and precipitation (Trenberth Kevin E., 1990). Jacobsson and Karltorp (2012) pointed out that, in response to the threat of climate changes, the European Union electricity sector has to undergo a large-scale transformation process to reduce loss. Many researchers have turned to use reanalysis data instead of historical data, for ex- ample, NARR, ERA-40, MERRA, and CFSR (Mesinger et al., 2010; Rienecker et al., 2011; Rose & Apt, 2015; Saha et al., 2010; Uppala et al., 2005). The data resource is ERA-20C archive Version 2.0 by an independent intergovernmen- tal organisation European Center for Medium range Weather Forecasting (ECMWF) which supported by 34 states. Long time period of measured meteorological data were revised by reanalysis of meteorological observations in ERA-20C project. ERA-20C is a global, high-resolution, coupled atmosphere–ocean–land surface–sea ice system to provide the best estimate of the state of these coupled domains over this period. The aim of this research is to find out a correlation between the meteorological fac- tors and the renewable energy potential, which implies that the forecasting models of renewable energy potential can be invented based on the global atmospheric data. In this paper, wind energy potential and solar energy potential in Vaasa region in Finland 175 were roughly assessed by analysing ERA-20C reanalysis data. 2. Literature Review Breslow and Sailor (2002) tried to find out the impacts of climate changes on wind speed and wind power output across the continent US. Two general circulation models provided similar trend until 2050 but various in the future 20 years. After six years, Sailor et al. (2008) put further this idea by investigating in wind statistics from models about five-state region within the Northwest US. The results showed that summertime wind speeds may decrease by 5–10% while wintertime wind speeds may decrease or increase slightly. In order to identify the changes in the future wind- and hydro-power resource potential in Norway, Seljom et al. (2011) evaluated the impact of climates changes with MARKAL Norway model. They found out that the reduction of heating demand will be signifi- cantly higher than the increase of cooling demand, and there may be lower cost of en- ergy system and electricity production. Wang et al. (2014) developed a general framework and applied grey cluster analysis method to compare analysis renewable energy vulnerability to climate change in China. The results depicted the distribution of areas rich in hydropower, wind power and solar energy potential, which helps to improve decision-making analysis. In response to promote the transition towards a low carbon economy in Scotland, Sample et al. (2015) reviewed the potential impacts climate change and presented state of knowledge regarding the resilience of Scotland’s hydropower resource to a changing climate. Chang et al. (2015) proposed a new statistical downscaling framework using simulated weather research and forecasting (WRF) model to evaluate the climate change impact 176 on wind resources in Taiwan Strait. It was found out that in the future wind energy den- sity distributions are higher in the eastern half of Taiwan Strait but will reduce slightly comparing to the past time period. Fant et al. (2015) presented a method that estimates the risk of climate-change on wind and solar resource potential. The assessment combines the risk-based climate projections from the Integrated Global Systems Model (IGSM), which considers emis- sions and global climate sensitivity uncertainty, with more regionally detailed climate information from 8 GCMs available from the Coupled Model Intercomparison Project phase 3 (CMIP-3). 3. Data analysis 3.1 Wind energy Instead of pick out continent area, we identify the specific coordinate of longitude and latitude of the interested area (in this case the Vaasa region in Finland), then retrieve the dataset in the format of NetCDF. In order to retrieve specific geographic site in the particular observation area, we choose the highest resolution grid (0.125 degree * 0.125 degree) while the lowest resolution is 3 degree * 3 degree. We retrieve everyday meteorological record midnight 00:00 from 1st January 1961 to 31st December 2010, and a total data of 18262 days in 50 years were analyzed. Wind speed V at the height of 100 meter can be calculated through 100 meter u wind component and 100 meter v wind component (Eq.1). Wind power directly depends on ambient natural resources and hence it is sensitive to climate variability. Wind power density is directly related with the electric power generation. It is proportional to the cube of speed and can be divided into different classes. 𝑉 = √𝑢2 + 𝑣2 (1) 𝐷𝑤𝑝 = 1 2𝑛 ∑ 𝑛 𝑖=1 (𝜌 ∙ 𝑉𝑖 3) (2) 177 In this research, we use MATLAB R2014a to calculate, analyze, and plot figures. Fig.1 shows the historical trend of wind power density in Vaasa region at the height of 100 meter from 1960 to 2010. Fig. 1 Wind Power Density for all past 50 year (Year 1961-2010) As can be seen from Fig. 1 that there are apparently fluctuations all through the past 50 years. We separately picked out maximum wind power density from every ten years and every five years, for example, time period 1971-1980 and 1971-1975. For each peak value in every ten years or every five years, the general variations are plotted in Fig. 2 and Fig. 3. 242,4574 252,4574 262,4574 272,4574 282,4574 292,4574 302,4574 312,4574 1962 1967 1973 1975 1983 1990 1992 1995 2000 2005 Peak values in every 5 years 178 Fig 2. Peak of Wind Power Density for every 5 years Fig 3. Peak of Wind Power Density for every 10 years Therefore, it can be judged from the wind power density in above figures. Fig. 3 shows a more apparent trend when comparing with Fig. 2. It reveals a trend of fast increase from 1961 to 1973 and after that keeps stable during 1973-1990. And it presents a gen- eral trend of dramatic increase in the short period of 1990-1992 then a trend of sharp decrease in long-term 1992-2010. 3.2 Solar energy Total radiation approximately equals to sum of direct radiation and diffuse radiation. Except of solar radiation, there are many other meteorological elements, which includ- ing cloud, sunny/rainy day, temperature and humidity, could influence solar power. These problems remain to be done in the future. 4. Conclusion This study demonstrates the great potential of using global atmospheric reanalysis data to analyze the potential of renewable energy sources which are related with climate change. It can effectively help decision-maker in macro level through analyzing long- term atmospheric data. For the 20th century, we observed and analyzed the wind power 252,5922 262,5922 272,5922 282,5922 292,5922 302,5922 312,5922 1967 1973 1990 1992 2005 Peak values in every 10 years 179 density of past fifty years 1961-2010. In this study, the correlation between the mete- orological factors and wind energy potential has been found out. The main outcomes of the present study are presented as follows: (a) As illustrated in Fig. 1, wind power density for all the past 50 years from 1961 to 2010 was in fluc- tuation all the time. (b) From Fig. 2 and Fig. 3, the characteristics of peak values about wind power density in every 5 and 10 years were separately depicted in the section of Data Analysis. And comparing peak values for each 10 years is easier than each 5 years. (c) According to Climate Change 2014 Synthesis Report, 2005 and 1998 were the warmest two years in the instrumental global surface air temperature record since 1850, and twelve years (1995 to 2006) ranked among the 12 warmest years on record since 1850. (d) The variation trend of wind power density is basically consistent with the changes in surface cli- mate, in particular with the temperature. (e) This study only focuses on peak values analysing as an exploratory study. More useful statistic in- formation are expected in the future. For instance, investigating the trend about annual sum of wind power density. Since ERA-20C global datasets includes the atmosphere, land-surface and ocean-wave reanalysis data from 1900 to 2010, we could also obtained the variation trend of nat- ural resources (wind energy, solar energy, tidal energy, etc.) and forecast resources potential in the future. The future study may focus on forecasting the global potential of these renewable energy resources in the next decades. References Abazari, A., Babaei, M., Muyeen, S. M., & Kamwa, I. (2020). Learning adaptive fuzzy droop of PV con- tribution to frequency excursion of hybrid micro-grid during parameters uncertainties. International Journal of Electrical Power and Energy Systems, 123(December 2019), 106305. https://doi.org/10.1016/j.ijepes.2020.106305 Adeyeye, K., Ijumba, N., & Colton, J. (2020). Exploring the environmental and economic impacts of wind energy: a cost-benefit perspective. International Journal of Sustainable Development and World Ecology, 27(8), 718–731. https://doi.org/10.1080/13504509.2020.1768171 Akyuz, E., Oktay, Z., & Dincer, I. (2012). Performance investigation of hydrogen production from a hy- brid wind-PV system. International Journal of Hydrogen Energy, 37(21), 16623–16630. https://doi.org/10.1016/j.ijhydene.2012.02.149 Albrecht, J., Laleman, R., & Vulsteke, E. (2015). Balancing demand-pull and supply-push measures to support renewable electricity in Europe. Renewable and Sustainable Energy Reviews, 49, 267–277. https://doi.org/10.1016/j.rser.2015.04.078 Alexiadis, M. C., Dokopoulos, P. S., & Sahsamanoglou, H. S. (1999). Wind speed and power forecasting based on spatial correlation models. IEEE Transactions on Energy Conversion, 14(3), 836–842. https://doi.org/10.1109/60.790962 180 Alpaydin, E. (2019). Introduction to machine learning. Ammari, C., Belatrache, D., Touhami, B., & Makhloufi, S. (2021). Sizing, optimization, control and ener- gy management of hybrid renewable energy system- a review. Energy and Built Environment. https://doi.org/10.1016/j.enbenv.2021.04.002 Anurag Bhardwaj, Wei Di, J. W. (2018). Deep Learning Essentials: Your hands-on guide to the fundamen- tals of deep ... - Anurag Bhardwaj, Wei Di, Jianing Wei - Google Böcker. Retrieved April 14, 2021, from https://books.google.se/books/about/Deep_Learning_Essentials.html?id=ISBKDwAAQBAJ&redir_ esc=y Archer, C. L., & Jacobson, M. Z. (2013). Geographical and seasonal variability of the global “practical” wind resources. Applied Geography, 45, 119–130. https://doi.org/10.1016/j.apgeog.2013.07.006 Azcárate, C., Blanco, R., Mallor, F., Garde, R., & Aguado, M. (2012). Peaking strategies for the manage- ment of wind-H 2 energy systems. Renewable Energy, 47, 103–111. https://doi.org/10.1016/j.renene.2012.04.016 Bhattacharjee, S., & Acharya, S. (2015). PV–wind hybrid power option for a low wind topography. Ener- gy Conversion and Management, 89, 942–954. https://doi.org/10.1016/j.enconman.2014.10.065 Bonou, A., Skelton, K., & Olsen, S. I. (2016). Ecodesign framework for developing wind turbines. Jour- nal of Cleaner Production. https://doi.org/10.1016/j.jclepro.2016.02.093 Bórawski, P., Bełdycka-Bórawska, A., Jankowski, K. J., Dubis, B., & Dunn, J. W. (2020). Development of wind energy market in the European Union. Renewable Energy, 161, 691–700. https://doi.org/https://doi.org/10.1016/j.renene.2020.07.081 Box, G. E. P., & Jenkins, G. M. (1976). Time series analysis: forecasting and control, revised ed. Holden- Day. Breslow, P. B., & Sailor, D. J. (2002). Vulnerability of wind power resources to climate change in the continental United States. Renewable Energy, 27(4), 585–598. https://doi.org/10.1016/S0960- 1481(01)00110-0 Cadenas, E., & Rivera, W. (2010). Wind speed forecasting in three different regions of Mexico, using a hybrid ARIMA–ANN model. Renewable Energy, 35(12), 2732–2738. https://doi.org/10.1016/j.renene.2010.04.022 Çapik, M., Yılmaz, A. O., & Çavuşoğlu, İ. (2012). Present situation and potential role of renewable energy in Turkey. Renewable Energy, 46, 1–13. https://doi.org/10.1016/j.renene.2012.02.031 Carolin Mabel, M., & Fernandez, E. (2008). Analysis of wind power generation and prediction using ANN: A case study. Renewable Energy, 33(5), 986–992. https://doi.org/10.1016/j.renene.2007.06.013 Carta, J. A., & Mentado, D. (2007). A continuous bivariate model for wind power density and wind tur- 181 bine energy output estimations. Energy Conversion and Management, 48(2), 420–432. https://doi.org/10.1016/j.enconman.2006.06.019 Cassola, F., & Burlando, M. (2012). Wind speed and wind energy forecast through Kalman filtering of Numerical Weather Prediction model output. Applied Energy, 99, 154–166. https://doi.org/10.1016/j.apenergy.2012.03.054 Çetin, O., & Ziya Sogut, M. (2021). A new strategic approach of energy management onboard ships sup- ported by exergy and economic criteria: A case study of a cargo ship. Ocean Engineering, 219(March 2020), 108137. https://doi.org/10.1016/j.oceaneng.2020.108137 Chadee, X. T., & Clarke, R. M. (2014). Large-scale wind energy potential of the Caribbean region using near-surface reanalysis data. Renewable and Sustainable Energy Reviews, 30, 45–58. https://doi.org/10.1016/j.rser.2013.09.018 Chang, T.-J., Chen, C.-L., Tu, Y.-L., Yeh, H.-T., & Wu, Y.-T. (2015). Evaluation of the climate change impact on wind resources in Taiwan Strait. Energy Conversion and Management, 95, 435–445. https://doi.org/10.1016/j.enconman.2015.02.033 Chávez-Ramírez, A. U., Vallejo-Becerra, V., Cruz, J. C., Ornelas, R., Orozco, G., Muñoz-Guerrero, R., & Arriaga, L. G. (2013). A hybrid power plant (Solar–Wind–Hydrogen) model based in artificial intel- ligence for a remote-housing application in Mexico. International Journal of Hydrogen Energy, 38(6), 2641–2655. https://doi.org/10.1016/j.ijhydene.2012.11.140 Chen, C., Liu, Y., Kumar, M., Qin, J., & Ren, Y. (2019). Energy consumption modelling using deep learn- ing embedded semi-supervised learning. Computers & Industrial Engineering, 135, 757–765. https://doi.org/10.1016/J.CIE.2019.06.052 Chen, K., & Yu, J. (2014). Short-term wind speed prediction using an unscented Kalman filter based state- space support vector regression approach. Applied Energy, 113, 690–705. https://doi.org/10.1016/j.apenergy.2013.08.025 Christensen, P., & Lund, H. (1998). Conflicting views of sustainability: The case of wind power and na- ture conservation in Denmark. European Environment, 8(1), 1–6. https://doi.org/10.1002/(SICI)1099-0976(199801/02)8:1<1::AID-EET139>3.0.CO;2-T Colak, I., Sagiroglu, S., & Yesilbudak, M. (2012). Data mining and wind power prediction: A literature review. Renewable Energy, 46, 241–247. https://doi.org/10.1016/j.renene.2012.02.015 Crossan, F. (2003). Research philosophy: towards an understanding. Nurse Researcher, 11(1), 46–55. https://doi.org/10.7748/nr2003.10.11.1.46.c5914 Cui, Q., Liu, Y., Ali, T., Gao, J., & Chen, H. (2020). Economic and climate impacts of reducing China’s renewable electricity curtailment: A comparison between CGE models with alternative nesting structures of electricity. Energy Economics, 91. https://doi.org/10.1016/j.eneco.2020.104892 Dashwood, J. (2012). Australia ’ s Energy Options : Renewables and efficiency. In Australia’s Energy 182 Options: Renewables and efficiency (pp. 45–52). De Giorgi, M. G., Ficarella, A., & Tarantino, M. (2011). Assessment of the benefits of numerical weather predictions in wind power forecasting based on statistical methods. Energy, 36(7), 3968–3978. https://doi.org/10.1016/j.energy.2011.05.006 Demir, N., & Taşkın, A. (2013). Life cycle assessment of wind turbines in Pınarbaşı-Kayseri. Journal of Cleaner Production, 54, 253–263. https://doi.org/10.1016/j.jclepro.2013.04.016 Demolli, H., Dokuz, A. S., Ecemis, A., & Gokcek, M. (2019). Wind power forecasting based on daily wind speed data using machine learning algorithms. Energy Conversion and Management, 198, 111823. https://doi.org/10.1016/J.ENCONMAN.2019.111823 Díaz-González, F., Hau, M., Sumper, A., & Gomis-Bellmunt, O. (2015). Coordinated operation of wind turbines and flywheel storage for primary frequency control support. International Journal of Elec- trical Power & Energy Systems, 68, 313–326. https://doi.org/10.1016/j.ijepes.2014.12.062 Dihrab, S. S., & Sopian, K. (2010). Electricity generation of hybrid PV/wind systems in Iraq. Renewable Energy, 35(6), 1303–1307. https://doi.org/10.1016/j.renene.2009.12.010 Dincer, I. (2000). Renewable energy and sustainable development: a crucial review. Renewable and Sus- tainable Energy Reviews, 4(2), 157–175. https://doi.org/10.1016/S1364-0321(99)00011-8 Duan, Y., Edwards, J. S., & Dwivedi, Y. K. (2019). Artificial intelligence for decision making in the era of Big Data – evolution, challenges and research agenda. International Journal of Information Man- agement, 48, 63–71. https://doi.org/10.1016/j.ijinfomgt.2019.01.021 Dudovskiy, J. (2018). An Ultimate Guide to Writing A Dissertation in Business Studies A Step-by-Step Assistance. Research Methodology, Multiple. Retrieved from https://research-methodology.net Emeis, S. (2018). Wind Energy Meteorology: Atmospheric Physics for Wind Power Generation - Stefan Emeis - Google Books. Retrieved April 14, 2021, from https://books.google.fi/books?hl=en&lr=&id=6r1TDwAAQBAJ&oi=fnd&pg=PR5&dq=wind+form ation+wind+energy+kinetic+energy&ots=tL0AtcgUsX&sig=DVNSFnTmKsqc5vWRu- OYaI4Q7Yk&redir_esc=y#v=onepage&q=wind formation wind energy kinetic energy&f=false Eriksson, S., Bernhoff, H., & Leijon, M. (2008). Evaluation of different turbine concepts for wind power. Renewable and Sustainable Energy Reviews, 12(5), 1419–1434. https://doi.org/10.1016/j.rser.2006.05.017 Fant, C., Adam Schlosser, C., & Strzepek, K. (2015). The impact of climate change on wind and solar resources in southern Africa. Applied Energy. https://doi.org/10.1016/j.apenergy.2015.03.042 Fertig, E., & Apt, J. (2011). Economics of compressed air energy storage to integrate wind power: A case study in ERCOT. Energy Policy, 39(5), 2330–2342. https://doi.org/10.1016/j.enpol.2011.01.049 Fiesler, E., Caulfield, H. J., Choudry, A., & Ryan, J. P. (1990). Maximum capacity topologies for fully connected layered neural networks with bidirectional connections. 827–831. 183 https://doi.org/10.1109/ijcnn.1990.137672 Flora, R., Marques, A. C., & Fuinhas, J. A. (2014). Wind power idle capacity in a panel of European coun- tries. Energy, 66, 823–830. https://doi.org/10.1016/j.energy.2013.12.061 Friedman, J. H. (2001). Greedy Function Approximation: A Gradient Boosting Machine. Annals of Statis- tics, 29(5), 1189–1232. Retrieved from http://marefateadyan.nashriyat.ir/node/150 Gao, D., Jiang, D., Liu, P., Li, Z., Hu, S., & Xu, H. (2014). An integrated energy storage system based on hydrogen storage: Process configuration and case studies with wind power. Energy, 66, 332–341. https://doi.org/10.1016/j.energy.2014.01.095 Ge, B., Wang, W., Bi, D., Rogers, C. B., Peng, F. Z., de Almeida, A. T., & Abu-Rub, H. (2013). Energy storage system-based power control for grid-connected wind power farm. International Journal of Electrical Power & Energy Systems, 44(1), 115–122. https://doi.org/10.1016/j.ijepes.2012.07.021 Gil-García, I. C., García-Cascales, M. S., Fernández-Guillamón, A., & Molina-García, A. (2019). Catego- rization and analysis of relevant factors for optimal locations in onshore and offshorewind power plants: A taxonomic review. Journal of Marine Science and Engineering, 7(11), 1–21. https://doi.org/10.3390/jmse7110391 González-Aparicio, I., & Zucker, A. (2015). Impact of wind power uncertainty forecasting on the market integration of wind energy in Spain. Applied Energy, 159, 334–349. https://doi.org/10.1016/j.apenergy.2015.08.104 González-Mingueza, C., & Muñoz-Gutiérrez, F. (2014). Wind prediction using Weather Research Fore- casting model (WRF): A case study in Peru. Energy Conversion and Management, 81, 363–373. https://doi.org/10.1016/j.enconman.2014.02.024 Griffith, A. K. (1974). A comparison and evaluation of three machine learning procedures as applied to the game of checkers. Artificial Intelligence, 5(2), 137–148. https://doi.org/10.1016/0004- 3702(74)90027-7 Güler, Ö. (2009). Wind energy status in electrical energy production of Turkey. Renewable and Sustaina- ble Energy Reviews, 13(2), 473–478. https://doi.org/10.1016/j.rser.2007.03.015 Guo, Z., Wu, J., Lu, H., & Wang, J. (2011). A case study on a hybrid wind speed forecasting method using BP neural network. Knowledge-Based Systems, 24(7), 1048–1056. https://doi.org/10.1016/j.knosys.2011.04.019 Hamilton, S. H., Fu, B., Guillaume, J. H. A., Badham, J., Elsawah, S., Gober, P., … Zare, F. (2019). A framework for characterising and evaluating the effectiveness of environmental modelling. Envi- ronmental Modelling and Software, 118, 83–98. https://doi.org/10.1016/j.envsoft.2019.04.008 Han, S., Li, J., & Liu, Y. (2011). Tabu Search Algorithm Optimized ANN Model for Wind Power Predic- tion with NWP. Energy Procedia, 12, 733–740. https://doi.org/10.1016/j.egypro.2011.10.099 He, Y., Li, M., Meng, Z., Chen, S., Huang, S., Hu, Y., & Zou, X. (2021). An overview of acoustic emission 184 inspection and monitoring technology in the key components of renewable energy systems. Me- chanical Systems and Signal Processing, 148. https://doi.org/10.1016/j.ymssp.2020.107146 Hennessey, J. P. (1977). Some Aspects of Wind Power Statistics. Journal of Applied Meteorology, 16(2), 119–128. https://doi.org/10.1175/1520-0450(1977)016<0119:SAOWPS>2.0.CO;2 Hochreiter, S., & Schmidhuber, J. (1997). Long Short-Term Memory. Neural Computation, 9(8), 1735– 1780. https://doi.org/10.1162/neco.1997.9.8.1735 Holttinen, H. (2006). Handling of wind power forecast errors in the Nordic power market. 2006 Interna- tional Conference on Probabilistic Methods Applied to Power Systems, 1–6. https://doi.org/10.1109/PMAPS.2006.360288 Holttinen, H., Miettinen, J., & Sillanpää, S. (2013). Wind power forecasting accuracy and uncertainty in Finland. Retrieved from http://www.vtt.fi/publications/index.js Hosseini-Firouz, M. (2013). Optimal offering strategy considering the risk management for wind power producers in electricity market. International Journal of Electrical Power & Energy Systems, 49, 359–368. https://doi.org/10.1016/j.ijepes.2013.01.015 Hu, H., Wang, L., & Tao, R. (2021). Wind speed forecasting based on variational mode decomposition and improved echo state network. Renewable Energy, 164, 729–751. https://doi.org/10.1016/j.renene.2020.09.109 Hu, Q., Zhang, S., Xie, Z., Mi, J., & Wan, J. (2014). Noise model based ν-support vector regression with its application to short-term wind speed forecasting. Neural Networks : The Official Journal of the International Neural Network Society, 57C, 1–11. https://doi.org/10.1016/j.neunet.2014.05.003 Hua, Y., Oliphant, M., & Hu, E. J. (2016). Development of renewable energy in Australia and China: A comparison of policies and status. Renewable Energy, 85, 1044–1051. https://doi.org/10.1016/j.renene.2015.07.060 Huang, Q., Shi, Y., Wang, Y., Lu, L., & Cui, Y. (2015). Multi-turbine wind-solar hybrid system. Renewa- ble Energy, 76, 401–407. https://doi.org/10.1016/j.renene.2014.11.060 Huang, X., Zhang, D., & Zhang, X. (2021). Energy management of intelligent building based on deep reinforced learning. Alexandria Engineering Journal, 60(1), 1509–1517. https://doi.org/10.1016/j.aej.2020.11.005 Huang, Z., & Chalabi, Z. S. (1995). Use of time-series analysis to model and forecast wind speed. Journal of Wind Engineering and Industrial Aerodynamics, 56(2–3), 311–322. https://doi.org/10.1016/0167- 6105(94)00093-S Jacobsson, S., & Karltorp, K. (2012). Formation of competences to realize the potential of offshore wind power in the European Union. Energy Policy, 44, 374–384. https://doi.org/10.1016/j.enpol.2012.01.069 Jannati, M., Hosseinian, S. H., Vahidi, B., & Li, G. (2016). A significant reduction in the costs of battery 185 energy storage systems by use of smart parking lots in the power fluctuation smoothing process of the wind farms. Renewable Energy, 87, 1–14. https://doi.org/10.1016/j.renene.2015.09.054 Jiang, P., Liu, Z., Niu, X., & Zhang, L. (2021). A combined forecasting system based on statistical meth- od, artificial neural networks, and deep learning methods for short-term wind speed forecasting. En- ergy, 217. https://doi.org/10.1016/j.energy.2020.119361 Jung, J., & Broadwater, R. P. (2014). Current status and future advances for wind speed and power fore- casting. Renewable and Sustainable Energy Reviews, 31, 762–777. https://doi.org/10.1016/j.rser.2013.12.054 Kaldellis, J. K., Kapsali, M., & Kavadias, K. A. (2010). Energy balance analysis of wind-based pumped hydro storage systems in remote island electrical networks. Applied Energy, 87(8), 2427–2437. https://doi.org/10.1016/j.apenergy.2010.02.016 Kamal, L., & Jafri, Y. Z. (1997). Time series models to simulate and forecast hourly averaged wind speed in Quetta, Pakistan. Solar Energy, 61(1), 23–32. https://doi.org/10.1016/S0038-092X(97)00037-6 Kamel, R. M., Chaouachi, A., & Nagasaka, K. (2010). Wind power smoothing using fuzzy logic pitch controller and energy capacitor system for improvement Micro-Grid performance in islanding mode. Energy, 35(5), 2119–2129. https://doi.org/10.1016/j.energy.2010.01.030 Kandpal, T. C., & Broman, L. (2014). Renewable energy education: A global status review. Renewable and Sustainable Energy Reviews, 34, 300–324. https://doi.org/10.1016/j.rser.2014.02.039 Kariniotakis, G. N., Stavrakakis, G. S., & Nogaret, E. F. (1996). Wind power forecasting using advanced neural networks models. IEEE Transactions on Energy Conversion, 11(4), 762–767. https://doi.org/10.1109/60.556376 Kavasseri, R. G., & Seetharaman, K. (2009). Day-ahead wind speed forecasting using f-ARIMA models. Renewable Energy, 34(5), 1388–1393. https://doi.org/10.1016/j.renene.2008.09.006 Kenabatho, P. K., Parida, B. P., & Moalafhi, D. B. (2012). The value of large-scale climate variables in climate change assessment: The case of Botswana’s rainfall. Physics and Chemistry of the Earth, Parts A/B/C, 50–52, 64–71. https://doi.org/10.1016/j.pce.2012.08.006 Khalid, M., & Savkin, A. V. (2010). A model predictive control approach to the problem of wind power smoothing with controlled battery storage. Renewable Energy, 35(7), 1520–1526. https://doi.org/10.1016/j.renene.2009.11.030 Khalid, M., & Savkin, A. V. (2014). Minimization and control of battery energy storage for wind power smoothing: Aggregated, distributed and semi-distributed storage. Renewable Energy, 64, 105–112. https://doi.org/10.1016/j.renene.2013.09.043 Kimura, Y., Onai, Y., & Ushiyama, I. (1996). A demonstrative study for the wind and solar hybrid power system. Renewable Energy, 9(1–4), 895–898. https://doi.org/10.1016/0960-1481(96)88424-2 Klir, G. J., & Folger, T. A. (1988). Fuzzy Sets, Uncertainty and Information [Hardcover]. Retrieved from 186 http://www.amazon.com/Fuzzy-Sets-Uncertainty-Information- George/dp/0133459845/ref=sr_1_1?ie=UTF8&qid=1403952802&sr=8- 1&keywords=Fuzzy+sets,+uncertainty,+and+information Kothari, C. R. (2004). Research Methodology: Methods and Techniques. Retrieved from https://books.google.fi/books?id=8c6gkbKi-F4C Laberge, Y. (2011). Advising on Research Methods: A Consultant’s Companion. Journal of Applied Statis- tics, 38(12), 2991–2991. https://doi.org/10.1080/02664763.2011.559375 LeCun, Y., Bengio, Y., & Hinton, G. (2015). Deep learning. Nature, 521(7553), 436–444. https://doi.org/10.1038/nature14539 Lee, K. M., Yoo, J., Kim, S.-W., Lee, J.-H., & Hong, J. (2019). Autonomic machine learning platform. International Journal of Information Management, 49, 491–501. https://doi.org/10.1016/J.IJINFOMGT.2019.07.003 Lei, M., Shiyan, L., Chuanwen, J., Hongling, L., & Yan, Z. (2009). A review on the forecasting of wind speed and generated power. Renewable and Sustainable Energy Reviews, 13(4), 915–920. https://doi.org/10.1016/j.rser.2008.02.002 Lim, S., & Chi, S. (2019). Xgboost application on bridge management systems for proactive damage estimation. Advanced Engineering Informatics, 41(October 2018), 100922. https://doi.org/10.1016/j.aei.2019.100922 Lin, J., Sun, Y., Cheng, L., & Gao, W. (2012). Assessment of the power reduction of wind farms under extreme wind condition by a high resolution simulation model. Applied Energy, 96, 21–32. https://doi.org/10.1016/j.apenergy.2011.10.028 Liu, H. of two new A.-A. and A.-K. hybrid methods for wind speed prediction, Tian, H., & Li, Y. (2012). Comparison of two new ARIMA-ANN and ARIMA-Kalman hybrid methods for wind speed predic- tion. Applied Energy, 98, 415–424. https://doi.org/10.1016/j.apenergy.2012.04.001 Liu, P., & Li, Y. (2021). An improved failure mode and effect analysis method for multi-criteria group decision-making in green logistics risk assessment. Reliability Engineering & System Safety, 215(May), 107826. https://doi.org/10.1016/j.ress.2021.107826 Liu, Y., Yu, S., Zhu, Y., Wang, D., & Liu, J. (2018). Modeling, planning, application and management of energy systems for isolated areas: A review. Renewable and Sustainable Energy Reviews, 82(June 2017), 460–470. https://doi.org/10.1016/j.rser.2017.09.063 Louka, P., Galanis, G., Siebert, N., Kariniotakis, G., Katsafados, P., Pytharoulis, I., & Kallos, G. (2008). Improvements in wind speed forecasts for wind power prediction purposes using Kalman filtering. Journal of Wind Engineering and Industrial Aerodynamics, 96(12), 2348–2362. https://doi.org/10.1016/j.jweia.2008.03.013 Lu, L., Yang, H., & Burnett, J. (2002). Investigation on wind power potential on Hong Kong islands—an 187 analysis of wind power and wind turbine characteristics. Renewable Energy, 27(1), 1–12. https://doi.org/10.1016/S0960-1481(01)00164-1 Lund, H. (2007). Renewable energy strategies for sustainable development. Energy, 32(6), 912–919. https://doi.org/10.1016/j.energy.2006.10.017 Ma, T., Yang, H., Lu, L., & Peng, J. (2014). Optimal design of an autonomous solar–wind-pumped storage power supply system. Applied Energy. https://doi.org/10.1016/j.apenergy.2014.11.026 Maleki, A., & Askarzadeh, A. (2014). Optimal sizing of a PV/wind/diesel system with battery storage for electrification to an off-grid remote region: A case study of Rafsanjan, Iran. Sustainable Energy Technologies and Assessments, 7, 147–153. https://doi.org/10.1016/j.seta.2014.04.005 Mannig, B., Müller, M., Starke, E., Merkenschlager, C., Mao, W., Zhi, X., … Paeth, H. (2013). Dynamical downscaling of climate change in Central Asia. Global and Planetary Change, 110, 26–39. https://doi.org/10.1016/j.gloplacha.2013.05.008 Maouedj, R., Mammeri, A., Draou, M. D., & Benyoucef, B. (2014). Performance Evaluation of Hybrid Photovoltaic-wind Power Systems. Energy Procedia, 50, 797–807. https://doi.org/10.1016/j.egypro.2014.06.098 Martínez, E., Sanz, F., Pellegrini, S., Jiménez, E., & Blanco, J. (2009). Life cycle assessment of a multi- megawatt wind turbine. Renewable Energy, 34(3), 667–673. https://doi.org/10.1016/j.renene.2008.05.020 McVicar, T. R., Roderick, M. L., Donohue, R. J., Li, L. T., Van Niel, T. G., Thomas, A., … Dinpashoh, Y. (2012). Global review and synthesis of trends in observed terrestrial near-surface wind speeds: Im- plications for evaporation. Journal of Hydrology, 416–417, 182–205. https://doi.org/10.1016/j.jhydrol.2011.10.024 Mens, M., Minnema, B., Overmars, K., & van den Hurk, B. (2021). Dilemmas in developing models for long-term drought risk management: The case of the National Water Model of the Netherlands. En- vironmental Modelling and Software, 143, 105100. https://doi.org/10.1016/j.envsoft.2021.105100 Mesinger, F., DiMego, G., Kalnay, E., Mitchell, K., Shafran, P. C., Ebisuzaki, W., … Shi, W. (2010). North American Regional Reanalysis. Retrieved from http://journals.ametsoc.org/doi/abs/10.1175/BAMS-87-3-343 Metternicht, G. (2001). Assessing temporal and spatial changes of salinity using fuzzy logic, remote sens- ing and GIS. Foundations of an expert system. Ecological Modelling, 144(2–3), 163–179. https://doi.org/10.1016/S0304-3800(01)00371-4 Mohamed, A. Z., Eskander, M. N., & Ghali, F. A. (2001). Fuzzy logic control based maximum power tracking of a wind energy system. Renewable Energy, 23(2), 235–245. https://doi.org/10.1016/S0960-1481(00)00099-9 Mohandes, M. A., Halawani, T. O., Rehman, S., & Hussain, A. A. (2004). Support vector machines for 188 wind speed prediction. Renewable Energy, 29(6), 939–947. https://doi.org/10.1016/j.renene.2003.11.009 Murage, M. W., & Anderson, C. L. (2014). Contribution of pumped hydro storage to integration of wind power in Kenya: An optimal control approach. Renewable Energy, 63, 698–707. https://doi.org/10.1016/j.renene.2013.10.026 Necefer, L., Wong-Parodi, G., Small, M. J., & Begay-Campbell, S. (2018). Integrating technical, econom- ic and cultural impacts in a decision support tool for energy resource management in the Navajo Na- tion. Energy Strategy Reviews, 22(May 2017), 136–146. https://doi.org/10.1016/j.esr.2018.06.004 Novak, P., Ekelund, T., Jovik, I., & Schmidtbauer, B. (1995). Modeling and control of variable-speed wind-turbine drive-system dynamics. IEEE Control Systems Magazine, 15(4), 28–38. https://doi.org/10.1109/37.408463 Nygaard, I., Rasmussen, K., Badger, J., Nielsen, T. T., Hansen, L. B., Stisen, S., … Togola, I. (2010). Using modeling, satellite images and existing global datasets for rapid preliminary assessments of renewable energy resources: The case of Mali. Renewable and Sustainable Energy Reviews, 14(8), 2359–2371. https://doi.org/10.1016/j.rser.2010.04.001 Olaofe, Z. O. (2014). A 5-day wind speed & power forecasts using a layer recurrent neural network (LRNN). Sustainable Energy Technologies and Assessments, 6, 1–24. https://doi.org/10.1016/j.seta.2013.12.001 On the Vertical Extrapolation of Mean Wind Power DensitySedefian, L. (1980). On the Vertical Extrapo- lation of Mean Wind Power Density. Journal of Applied Meteorology, 19(4), 488–493. https://doi.org/10.1175/1520-0450(1980)019<0488:OTVEOM>2.0.CO;2 Oñederra, O., Asensio, F. J., Saldaña, G., Martín, J. I. S., & Zamora, I. (2020). Wind energy harnessing in a railway infrastructure: Converter topology and control proposal. Electronics (Switzerland), 9(11), 1–14. https://doi.org/10.3390/electronics9111943 Ortegon, K., Nies, L. F., & Sutherland, J. W. (2013). Preparing for end of service life of wind turbines. Journal of Cleaner Production, 39, 191–199. https://doi.org/10.1016/j.jclepro.2012.08.022 Ortiz-García, E. G., Salcedo-Sanz, S., Pérez-Bellido, Á. M., Gascón-Moreno, J., Portilla-Figueras, J. A., & Prieto, L. (2011). Short-term wind speed prediction in wind farms based on banks of support vector machines. Wind Energy, 14(2), 193–207. https://doi.org/10.1002/we.411 Ozcan, M. (2014). Assessment of renewable energy incentive system from investors’ perspective. Renew- able Energy, 71, 425–432. https://doi.org/10.1016/j.renene.2014.05.053 Peng, H., Liu, F., & Yang, X. (2013). A hybrid strategy of short term wind power prediction. Renewable Energy, 50, 590–595. https://doi.org/10.1016/j.renene.2012.07.022 Pereira, E. B., Martins, F. R., Pes, M. P., da Cruz Segundo, E. I., & Lyra, A. de A. (2013). The impacts of global climate changes on the wind power density in Brazil. Renewable Energy, 49, 107–110. 189 https://doi.org/10.1016/j.renene.2012.01.053 Persson, A. (1991). Kalman filtering a new approach to adaptive statistical interpretation of numerical meteorological forecasts. WMO Technical Document, (421). Pinson, P., & Kariniotakis, G. (2004). On-line assessment of prediction risk for wind power production forecasts. Wind Energy, 7(2), 119–132. https://doi.org/10.1002/we.114 Poncela, M., Poncela, P., & Perán, J. R. (2013). Automatic tuning of Kalman filters by maximum likeli- hood methods for wind energy forecasting. Applied Energy, 108, 349–362. https://doi.org/10.1016/j.apenergy.2013.03.041 Pryor, S. C., & Barthelmie, R. J. (2010). Climate change impacts on wind energy: A review. Renewable and Sustainable Energy Reviews, 14(1), 430–437. https://doi.org/10.1016/j.rser.2009.07.028 Rezaeian-Marjani, S., Masoumzadehasl, S., Galvani, S., & Talavat, V. (2020). Probabilistic assessment of D-STATCOM operation considering correlated uncertain variables. International Transactions on Electrical Energy Systems, 30(10). https://doi.org/10.1002/2050-7038.12522 Rienecker, M. M., Suarez, M. J., Gelaro, R., Todling, R., Bacmeister, J., Liu, E., … Woollen, J. (2011). MERRA: NASA’s Modern-Era Retrospective Analysis for Research and Applications. Journal of Climate, 24(14), 3624–3648. https://doi.org/10.1175/JCLI-D-11-00015.1 Robert Vautard, Françoise Thais, Isabelle Tobin, François-Marie Bréon, Jean-Guy Devezeaux de Lavergne, Augustin Colette, P. Y. & P. M. R. (2014). Time series analysis : forecasting and control: George E. P. Jenkins, Gwilym M., Box: Amazon.com: Books. Rose, S., & Apt, J. (2015). What can reanalysis data tell us about wind power? Renewable Energy, 83, 963–969. https://doi.org/10.1016/j.renene.2015.05.027 Rudenko, N. V., Ershov, V. V., & Evstafiev, V. V. (2017). The Environment Friendly Power Source for Power Supply of Mobile Communication Base Stations. IOP Conference Series: Earth and Envi- ronmental Science, 66(1). https://doi.org/10.1088/1755-1315/66/1/012024 Saeed, M. A., Ahmed, Z., & Zhang, W. (2020). Wind energy potential and economic analysis with a com- parison of different methods for determining the optimal distribution parameters. Renewable Ener- gy, 161, 1092–1109. https://doi.org/https://doi.org/10.1016/j.renene.2020.07.064 Saha, S., Moorthi, S., Pan, H.-L., Wu, X., Wang, J., Nadiga, S., … Goldberg, M. (2010). The NCEP Cli- mate Forecast System Reanalysis. Bulletin of the American Meteorological Society, 91(8), 1015– 1057. https://doi.org/10.1175/2010BAMS3001.1 Sailor, D. J., Smith, M., & Hart, M. (2008). Climate change implications for wind power resources in the Northwest United States. Renewable Energy, 33(11), 2393–2406. https://doi.org/10.1016/j.renene.2008.01.007 Salcedo-Sanz, S., Ortiz-García, E. G., Portilla-Figueras, A., Prieto, L., & Paredes, D. (2009). Hybridizing the fifth generation mesoscale model with artificial neural networks for short-term wind speed pre- 190 diction. Renewable Energy, 34(6), 1451–1457. https://doi.org/10.1016/j.renene.2008.10.017 Saleh Asheghabadi, M., Sahafnia, M., Bahadori, A., & Bakhshayeshi, N. (2019). Seismic behavior of suction caisson for offshore wind turbine to generate more renewable energy. International Journal of Environmental Science and Technology, 16(7), 2961–2972. https://doi.org/10.1007/s13762-018- 2150-8 Sample, J. E., Duncan, N., Ferguson, M., & Cooksley, S. (2015). Scotland׳s hydropower: Current capacity, future potential and the possible impacts of climate change. Renewable and Sustainable Energy Re- views, 52, 111–122. https://doi.org/10.1016/j.rser.2015.07.071 Sarrias-Mena, R., Fernández-Ramírez, L. M., García-Vázquez, C. A., & Jurado, F. (2014). Fuzzy logic based power management strategy of a multi-MW doubly-fed induction generator wind turbine with battery and ultracapacitor. Energy, 70, 561–576. https://doi.org/10.1016/j.energy.2014.04.049 Satkin, M., Noorollahi, Y., Abbaspour, M., & Yousefi, H. (2014). Multi criteria site selection model for wind-compressed air energy storage power plants in Iran. Renewable and Sustainable Energy Re- views, 32, 579–590. https://doi.org/10.1016/j.rser.2014.01.054 Schleisner, L. (2000). Life cycle assessment of a wind farm and related externalities. Renewable Energy, 20(3), 279–288. https://doi.org/10.1016/S0960-1481(99)00123-8 Sebastián, R., & Peña-Alzola, R. (2015). Control and simulation of a flywheel energy storage for a wind diesel power system. International Journal of Electrical Power & Energy Systems, 64, 1049–1056. https://doi.org/10.1016/j.ijepes.2014.08.017 Seljom, P., Rosenberg, E., Fidje, A., Haugen, J. E., Meir, M., Rekstad, J., & Jarlset, T. (2011). Modelling the effects of climate change on the energy system—A case study of Norway. Energy Policy, 39(11), 7310–7321. https://doi.org/10.1016/j.enpol.2011.08.054 Shamshirband, S., Keivani, A., Mohammadi, K., Lee, M., Hamid, S. H. A., & Petkovic, D. (2016). As- sessing the proficiency of adaptive neuro-fuzzy system to estimate wind power density: Case study of Aligoodarz, Iran. Renewable and Sustainable Energy Reviews, 59, 429–435. https://doi.org/10.1016/j.rser.2015.12.269 Sharma, N., Sharma, R., & Jindal, N. (2021). Machine Learning and Deep Learning Applications-A Vi- sion. Global Transitions Proceedings, 0–8. https://doi.org/10.1016/j.gltp.2021.01.004 Shen, N., Wang, Y., Peng, H., & Hou, Z. (2020). Renewable energy green innovation, fossil energy con- sumption, and air pollution-spatial empirical analysis based on China. Sustainability (Switzerland), 12(16). https://doi.org/10.3390/SU12166397 Shen, S., Sadoughi, M., Chen, X., Hong, M., & Hu, C. (2019). A deep learning method for online capacity estimation of lithium-ion batteries. Journal of Energy Storage, 25, 100817. https://doi.org/10.1016/J.EST.2019.100817 Sheng Shen, Mohammadkazem Sadoughi, Xiangyi Chen, Mingyi Hong, C. H. (2019). IEEE Xplore: IEEE 191 Transactions on Pattern Analysis and Machine Intelligence. Retrieved November 1, 2019, from Journal of Energy Storage website: https://ieeexplore.ieee.org/xpl/RecentIssue.jsp?punumber=34 Shi, J., Guo, J., & Zheng, S. (2012). Evaluation of hybrid forecasting approaches for wind speed and pow- er generation time series. Renewable and Sustainable Energy Reviews, 16(5), 3471–3480. https://doi.org/10.1016/j.rser.2012.02.044 Shin, D., Ha, E., Kim, T., & Kim, C. (2021). Short-term photovoltaic power generation predicting by input/output structure of weather forecast using deep learning. Soft Computing, 25(1), 771–783. https://doi.org/10.1007/s00500-020-05199-7 Shin, Y., Koo, W. Y., Kim, T. H., Jung, S., & Kim, H. (2015). Capacity design and operation planning of a hybrid PV–wind–battery–diesel power generation system in the case of Deokjeok Island. Applied Thermal Engineering, 89, 514–525. https://doi.org/10.1016/j.applthermaleng.2015.06.043 Shukur, O. B., & Lee, M. H. (2015). Daily wind speed forecasting through hybrid KF-ANN model based on ARIMA. Renewable Energy, 76, 637–647. https://doi.org/10.1016/j.renene.2014.11.084 Simoes, M. G., Bose, B. K., & Spiegel, R. J. (1997a). Design and performance evaluation of a fuzzy- logic-based variable-speed wind generation system. IEEE Transactions on Industry Applications, 33(4), 956–965. https://doi.org/10.1109/28.605737 Simoes, M. G., Bose, B. K., & Spiegel, R. J. (1997b). Fuzzy logic based intelligent control of a variable speed cage machine wind generation system. IEEE Transactions on Power Electronics, 12(1), 87– 95. https://doi.org/10.1109/63.554173 Soman, S. S., Zareipour, H., Malik, O., & Mandal, P. (2010). A review of wind power and wind speed forecasting methods with different time horizons. North American Power Symposium 2010, 1–8. https://doi.org/10.1109/NAPS.2010.5619586 Soukissian, T. H., & Papadopoulos, A. (2015). Effects of different wind data sources in offshore wind power assessment. Renewable Energy, 77, 101–114. https://doi.org/10.1016/j.renene.2014.12.009 Staid, A., & Guikema, S. D. (2013). Statistical analysis of installed wind capacity in the United States. Energy Policy, 60, 378–385. https://doi.org/10.1016/j.enpol.2013.05.076 Swider, D. J. (2007). Compressed Air Energy Storage in an Electricity System With Significant Wind Power Generation. IEEE Transactions on Energy Conversion, 22(1), 95–102. https://doi.org/10.1109/TEC.2006.889547 Taraft, S., Rekioua, D., & Aouzellag, D. (2013). Wind Power Control System Associated to the Flywheel Energy Storage System Connected to the Grid. Energy Procedia, 36, 1147–1157. https://doi.org/10.1016/j.egypro.2013.07.130 Taylor, J. W., & Buizza, R. (2002). Neural network load forecasting with weather ensemble predictions. IEEE Transactions on Power Systems, 17(3), 626–632. https://doi.org/10.1109/TPWRS.2002.800906 192 Tong, Z., Cheng, Z., & Tong, S. (2021). A review on the development of compressed air energy storage in China: Technical and economic challenges to commercialization. Renewable and Sustainable Ener- gy Reviews, 135(September 2019), 110178. https://doi.org/10.1016/j.rser.2020.110178 Trenberth Kevin E. (1990). Recent Observed Interdecadal Climate Changes in the Northern Hemisphere. Bulletin of the American Meteorological Society, 71(7), 988–993. Trianni, A., Cagno, E., Bertolotti, M., Thollander, P., & Andersson, E. (2019). Energy management : A practice-based assessment model. Applied Energy, 235(December 2018), 1614–1636. https://doi.org/10.1016/j.apenergy.2018.11.032 Ucar, A., & Balo, F. (2010). Assessment of wind power potential for turbine installation in coastal areas of Turkey. Renewable and Sustainable Energy Reviews, 14(7), 1901–1912. https://doi.org/10.1016/j.rser.2010.03.021 Uddin, M. S., & Kumar, S. (2014). Energy, emissions and environmental impact analysis of wind turbine using life cycle assessment technique. Journal of Cleaner Production, 69, 153–164. https://doi.org/10.1016/j.jclepro.2014.01.073 Uppala, S. M., KÅllberg, P. W., Simmons, A. J., Andrae, U., Bechtold, V. D. C., Fiorino, M., … Woollen, J. (2005). The ERA-40 re-analysis. Quarterly Journal of the Royal Meteorological Society, 131(612), 2961–3012. https://doi.org/10.1256/qj.04.176 Vijay M, D., Singh, B., & Bhuvaneswari, G. (2020). A High-Performance Microgrid with a Mechanical Sensorless SynRG Operated Wind Energy Generating System. IEEE Transactions on Industrial In- formatics, 16(12), 7349–7359. https://doi.org/10.1109/TII.2020.2976906 Wang, B., Ke, R.-Y., Yuan, X.-C., & Wei, Y.-M. (2014). China׳s regional assessment of renewable energy vulnerability to climate change. Renewable and Sustainable Energy Reviews, 40, 185–195. https://doi.org/10.1016/j.rser.2014.07.154 Wang, J., Qin, S., Zhou, Q., & Jiang, H. (2015). Medium-term wind speeds forecasting utilizing hybrid models for three different sites in Xinjiang, China. Renewable Energy, 76, 91–101. https://doi.org/10.1016/j.renene.2014.11.011 Wang, P., Fan, E., & Wang, P. (2021). Comparative analysis of image classification algorithms based on traditional machine learning and deep learning. Pattern Recognition Letters, 141, 61–67. https://doi.org/10.1016/j.patrec.2020.07.042 Wang, R., Li, C., Fu, W., & Tang, G. (2020). Deep Learning Method Based on Gated Recurrent Unit and Variational Mode Decomposition for Short-Term Wind Power Interval Prediction. IEEE Transac- tions on Neural Networks and Learning Systems, 31(10), 3814–3827. https://doi.org/10.1109/TNNLS.2019.2946414 Wang, X. Y., Mahinda Vilathgamuwa, D., & Choi, S. S. (2008). Determination of Battery Storage Capaci- ty in Energy Buffer for Wind Farm. IEEE Transactions on Energy Conversion, 23(3), 868–878. 193 https://doi.org/10.1109/TEC.2008.921556 Wu, J., Zhang, B., Li, H., Li, Z., Chen, Y., & Miao, X. (2014). Statistical distribution for wind power fore- cast error and its application to determine optimal size of energy storage system. International Jour- nal of Electrical Power & Energy Systems, 55, 100–107. https://doi.org/10.1016/j.ijepes.2013.09.003 Xydis, G. (2013). On the exergetic capacity factor of a wind – solar power generation system. Journal of Cleaner Production, 47, 437–445. https://doi.org/10.1016/j.jclepro.2012.07.014 Yang, H., Lu, L., & Zhou, W. (2007). A novel optimization sizing model for hybrid solar-wind power generation system. Solar Energy, 81(1), 76–84. https://doi.org/10.1016/j.solener.2006.06.010 Yang, M., Patiño-Echeverri, D., & Yang, F. (2012). Wind power generation in China: Understanding the mismatch between capacity and generation. Renewable Energy, 41, 145–151. https://doi.org/10.1016/j.renene.2011.10.013 Zahedi, A. (2010). Australian renewable energy progress. Renewable and Sustainable Energy Reviews, 14(8), 2208–2213. https://doi.org/10.1016/j.rser.2010.03.026 Zahedi, Ahmad. (2014). Sustainable Power Supply Using Solar Energy and Wind Power Combined with Energy Storage. Energy Procedia, 52, 642–650. https://doi.org/10.1016/j.egypro.2014.07.120 Zhang, D., Qian, L., Mao, B., Huang, C., Huang, B., & Si, Y. (2018). A Data-Driven Design for Fault Detection of Wind Turbines Using Random Forests and XGboost. IEEE Access, 6, 21020–21031. https://doi.org/10.1109/ACCESS.2018.2818678 Zhang, J., Chowdhury, S., & Messac, A. (2014). A comprehensive measure of the energy resource: Wind power potential (WPP). Energy Conversion and Management, 86, 388–398. https://doi.org/10.1016/j.enconman.2014.04.083 Zhang, Y., Pan, G., Zhao, Y., Li, Q., & Wang, F. (2020). Short-term wind speed interval prediction based on artificial intelligence methods and error probability distribution. Energy Conversion and Man- agement, 224(June), 113346. https://doi.org/10.1016/j.enconman.2020.113346 Zhao, P., Dai, Y., & Wang, J. (2014). Design and thermodynamic analysis of a hybrid energy storage sys- tem based on A-CAES (adiabatic compressed air energy storage) and FESS (flywheel energy stor- age system) for wind power application. Energy, 70, 674–684. https://doi.org/10.1016/j.energy.2014.04.055 Zhao, P., Wang, M., Wang, J., & Dai, Y. (2015). A preliminary dynamic behaviors analysis of a hybrid energy storage system based on adiabatic compressed air energy storage and flywheel energy stor- age system for wind power application. Energy, 84, 825–839. https://doi.org/10.1016/j.energy.2015.03.067 Zhao, X., Wang, C., Su, J., & Wang, J. (2019). Research and application based on the swarm intelligence algorithm and artificial intelligence for wind farm decision system. Renewable Energy, 134, 681– 194 697. https://doi.org/10.1016/j.renene.2018.11.061 Zhao, Z.-Y., Zhu, J., & Zuo, J. (2014). Sustainable development of the wind power industry in a complex environment: a flexibility study. Energy Policy, 75, 392–397. https://doi.org/10.1016/j.enpol.2014.10.017 Zhou, J., Shi, J., & Li, G. (2011). Fine tuning support vector machines for short-term wind speed forecast- ing. Energy Conversion and Management, 52(4), 1990–1998. https://doi.org/10.1016/j.enconman.2010.11.007 Zhou, W., Yang, H., & Fang, Z. (2006). Wind power potential and characteristic analysis of the Pearl Riv- er Delta region, China. Renewable Energy, 31(6), 739–753. https://doi.org/10.1016/j.renene.2005.05.006 195