A Deep Learning-to-learning Based Control system for renewable microgrids

Abstract

In terms of microgrids (MGs) operation, optimal control and management are vital issues that must be addressed carefully. This paper proposes a practical framework for the optimal energy management and control of renewable MGs considering energy storage (ES) devices, wind turbines, and microturbines. Due to the non-linearity and complexity of operation problems in MGs, it is vital to use an accurate and robust optimization technique to control the power flow of units efficiently. To this end, in the proposed framework, teacher learning-based optimization (TLBO) is utilized to solve the power flow dispatch in the system efficiently. Moreover, a novel hybrid deep learning model based on principal component analysis (PCA), convolutional neural networks (CNN), and bidirectional long short-term memory (BLSTM) is proposed to address the short-term wind power forecasting problem. The feasibility and performance of the proposed framework and the effect of wind power forecasting on operation efficiency are examined using the IEEE 33-bus test system. Also, the Australian Woolnorth wind site data is utilized as a real-world dataset to evaluate the performance of the forecasting model. The results show that the proposed framework can be used to schedule MGs in the best way possible.