Microgrid Optimization Strategy for Charging and Swapping Power Stations with New Energy Based on Multi-Agent Reinforcement Learning

Abstract

Aiming at the coordinated control of charging and swapping loads in complex environments, this research proposes an optimization strategy for microgrids with new energy charging and swapping stations based on adaptive multi-agent reinforcement learning. First, a microgrid model including charging and swapping loads, photovoltaic power generation, and wind power generation was constructed, and the Markov decision process was used to characterize the stochastic characteristics of new energy power generation, including charging and swapping loads. The deep relationship between uncertainty factors and charging and swapping laws was explored, and an adaptive multi-agent deep reinforcement learning method was used to optimize the random action selection process, improve the convergence speed of the coordinated optimization model, and realize coordinated control of multiple charging and swapping loads. Finally, through the analysis of different scenarios, the effectiveness of the proposed adaptive multi-agent reinforcement learning model for coordinated control of charging and swapping loads was verified. The results show that the proposed method has a faster convergence speed and can effectively optimize the charging process of charging and swapping loads, reducing power fluctuations of the newly connected energy grid.