Twin-delayed deep deterministic policy gradient for enhanced power optimization in solar PV-integrated DFIG wind...

Area of Science:

• Electrical Engineering
• Renewable Energy Systems
• Control Theory

Background:

• Modern power systems face stability challenges due to the integration of intermittent renewable energy sources.
• Conventional Proportional-Integral (PI) controllers struggle with the nonlinearities inherent in hybrid DFIG-solar PV systems.
• Deep Deterministic Policy Gradient (DDPG) algorithms can exhibit aggressive control actions leading to instability.

Purpose of the Study:

• To apply the Twin-Delayed Deep Deterministic Policy Gradient (TD3) algorithm for multi-objective control of a grid-connected DFIG-solar PV system.
• To develop a single, unified controller capable of managing the Rotor Side Converter (RSC), Grid Side Converter (GSC), and integrated solar PV system.
• To overcome the limitations of PI and DDPG controllers in handling system nonlinearities and aggressive control actions.

Main Methods:

• Implementation of a single TD3-based controller for simultaneous control of RSC, GSC, and solar PV integration.
• Utilizing dual critic networks within the TD3 algorithm to mitigate action overestimation and improve learning stability.
• Validation through OPAL-RT real-time hardware-in-the-loop (HIL) simulations.

Main Results:

• The TD3 controller demonstrated a 10.3% reduction in power overshoot compared to PI control.
• Achieved an 8% improvement in DC link voltage regulation and a 15.3% faster response time.
• Showcased a 16.9% faster settling time and superior performance over DDPG across all evaluated metrics.

Conclusions:

• The TD3 algorithm offers a robust and efficient solution for controlling hybrid DFIG-solar PV systems.
• TD3 provides enhanced stability, faster response, and improved regulation compared to conventional PI and DDPG methods.
• This unified controller approach is effective for managing complex renewable energy integration challenges in power systems.