Resilient frequency stabilization of renewable-penetrated microgrids via deep learning and machine learning.

Area of Science:

• Electrical Engineering
• Renewable Energy Systems
• Control Systems

Background:

• Microgrids (MGs) require stable frequency for reliable operation.
• Wind power's inherent variability poses challenges to MG frequency regulation.
• Accurate wind speed forecasting is crucial for mitigating power imbalances.

Purpose of the Study:

• To investigate wind speed prediction using Support Vector Regression (SVR) and Deep Neural Networks (DNN).
• To analyze the impact of wind speed forecasting on MG frequency regulation.
• To propose an optimized Proportional-Integral-Derivative (PID) controller for enhanced frequency stability.

Main Methods:

• Wind speed prediction using SVR and DNN models at four distinct sites.
• Incorporating predicted wind data as disturbances in a complex MG system model.
• Optimizing a PID controller using a hybrid Particle Swarm Optimization (PSO) and Grey Wolf Optimization (GWO) algorithm.
• Validation through time- and frequency-domain analyses and Hardware-in-the-Loop (HIL) simulation.

Main Results:

• SVR demonstrated superior wind speed forecasting accuracy over DNN, reducing prediction errors significantly.
• The hybrid PSO-GWO tuned PID controller outperformed individual PSO and GWO optimized controllers.
• The proposed controller significantly reduced overshoot, settling time, and integral square error in MG frequency response.
• Real-time HIL implementation confirmed the effectiveness of the proposed approach.

Conclusions:

• SVR is a highly effective method for wind speed forecasting in MGs.
• The hybrid PSO-GWO optimized PID controller significantly enhances MG frequency stability.
• The validated approach improves the reliability of renewable energy-integrated microgrids.