Actor-Critic RL Approach to Stabilize Voltage in DC Microgrids for Linear, CPL, and PPL Load Profiles

Voltage regulation is essential for stable and efficient operation in DC microgrids, especially under varying load conditions, including pulsed power loads (PPL) and constant power loads (CPL). Nonlinear and dynamic load profiles cause significant voltage fluctuations, challenging the performance of conventional control techniques. This paper introduces an Actor-Critic reinforcement learning (RL)-based approach combined with a Proportional-Integral (PI) controller to enhance voltage regulation in DC microgrids. The proposed method dynamically adjusts the duty cycle of DC-DC converters, minimizing voltage variations and ensuring smooth transitions. Simulation results, validated through MATLAB Simulink, demonstrate the superiority of the proposed method over traditional PI controllers. For PPL conditions, the approach achieves a 60% reduction in voltage dip, 50% faster recovery time, and improved damping. Under CPL scenarios, voltage variation is reduced by 57%, steady-state error decreases by 57%, and power quality is enhanced with stable performance. During additional PPL events, the voltage spike is reduced by 55%, settling time improves by 52%, and overall system stability is significantly enhanced. These findings highlight the robustness and adaptability of the proposed hybrid Actor-Critic RL framework in addressing the challenges posed by nonlinear and dynamic load variations in DC microgrids, offering a practical solution for real-world applications.