Learning-Based Model Predictive Control of DC-DC Buck Converters in DC Microgrids: A Multi-Agent Deep Reinforcement Learning Approach

This paper proposes a learning-based finite control set model predictive (FCS-MPC) to improve the performance of DC-DC buck converters interfaced with constant power loads in DC microgrid (DC-MG). An approach based on deep reinforcement learning (DRL) is presented address one ongoing challenges FCS-MPC converters, i.e., optimal design weighting coefficients appearing objective function for each converter. A deterministic policy gradient method employed learn coefficient policy. Markov decision formulates DRL problem. The agent trained converter MG, and are obtained reward computation interactions between MG agent. proposed strategy wholly distributed, wherein agents exchange data other agents, implying multi-agent scheme offers several advantages, including preventing dependency system operating point conditions, plug-and-play capability, robustness against uncertainties unknown load dynamics.