A Deep Reinforcement Learning Control Strategy with Integrated Droop Control for Parallel DC-DC Buck Converters with CPLs

The establishment of DC microgrids presents difficulties in ensuring stability and optimizing transient-time control performance. In response to the various issues in bus voltage control of parallel DC-DC buck converters, a control strategy based on model-free deep reinforcement learning (DRL) combined with droop control is proposed in this paper. Utilizing DRL controllers to independently control multiple DC-DC buck converters, achieving bus voltage stability and precise power distribution. Compared to traditional controllers such as PID control, model predictive control (MPC), and sliding mode control (SMC), the proposed strategy, leveraging the fast adaptive capability of DRL, significantly reduces the performance recovery time when disturbances occur. The simulation results illustrate the effectiveness of the approach in the presence of significant signal disturbances.