FPGA Validated Advanced Learning-Based Voltage Control of DC/DC Converter Feeding CPL in DC Microgrid Applications
Khan, Hussain Sarwar; Kauhaniemi, Kimmo (2023-08-31)
Katso/ Avaa
Tiedosto avautuu julkiseksi: : 31.08.2025
Khan, Hussain Sarwar
Kauhaniemi, Kimmo
IEEE
31.08.2023
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi-fe202401122570
https://urn.fi/URN:NBN:fi-fe202401122570
Kuvaus
vertaisarvioitu
The high penetration of renewable energy distribution generations enables the concept of microgrids and is widely accepted for future power systems. In this context, the DC microgrid is preferred due to easy integration, less system losses and offer high reliability and efficiency compared to its counterparts. However, the constant power loads (CPL) are a risk to the stability of the power electronics devices due to their negative impedance characteristics and also effects the voltage quality. To overcome these conditions, this paper proposes advanced artificial intelligence-based control of DC/DC converter to regulate the DC voltage in DC microgrid (MG) applications. At the start, model predictive control is implemented as an expert to control the studied converter to extract the dataset. The extracted dataset is used to train the proposed artificial neural network (ANN). The proposed controller is tested under various operating conditions while feeding the constant power loads. The proposed controller presents a superior transient response compared to conventional model predictive control (MPC). The experimental validation of the proposed scheme is carried out by implementing the controller on the FPGA ZYBO Z7-7020 board. The results are also compared with the conventional PI control. The proposed control technique has less computational burden and mitigates destabilizing effects caused by the CPLs.
The high penetration of renewable energy distribution generations enables the concept of microgrids and is widely accepted for future power systems. In this context, the DC microgrid is preferred due to easy integration, less system losses and offer high reliability and efficiency compared to its counterparts. However, the constant power loads (CPL) are a risk to the stability of the power electronics devices due to their negative impedance characteristics and also effects the voltage quality. To overcome these conditions, this paper proposes advanced artificial intelligence-based control of DC/DC converter to regulate the DC voltage in DC microgrid (MG) applications. At the start, model predictive control is implemented as an expert to control the studied converter to extract the dataset. The extracted dataset is used to train the proposed artificial neural network (ANN). The proposed controller is tested under various operating conditions while feeding the constant power loads. The proposed controller presents a superior transient response compared to conventional model predictive control (MPC). The experimental validation of the proposed scheme is carried out by implementing the controller on the FPGA ZYBO Z7-7020 board. The results are also compared with the conventional PI control. The proposed control technique has less computational burden and mitigates destabilizing effects caused by the CPLs.
Tiivistelmä
The high penetration of renewable energy distribution generations enables the concept of microgrids and is widely accepted for future power systems. In this context, the DC microgrid is preferred due to easy integration, less system losses and offer high reliability and efficiency compared to its counterparts. However, the constant power loads (CPL) are a risk to the stability of the power electronics devices due to their negative impedance characteristics and also effects the voltage quality. To overcome these conditions, this paper proposes advanced artificial intelligence-based control of DC/DC converter to regulate the DC voltage in DC microgrid (MG) applications. At the start, model predictive control is implemented as an expert to control the studied converter to extract the dataset. The extracted dataset is used to train the proposed artificial neural network (ANN). The proposed controller is tested under various operating conditions while feeding the constant power loads. The proposed controller presents a superior transient response compared to conventional model predictive control (MPC). The experimental validation of the proposed scheme is carried out by implementing the controller on the FPGA ZYBO Z7-7020 board. The results are also compared with the conventional PI control. The proposed control technique has less computational burden and mitigates destabilizing effects caused by the CPLs.
Kokoelmat
- Artikkelit [2625]