Virtual Power Plant Adjustable Resource Aggregation Adjustment Optimization Strategy Based on Multi-agent Game

Abstract

Based on the virtual power plant platform, this paper studies the aggregation adjustment and optimization strategy of multiple agents, and uses the multi-agent reinforcement learning strategy to realize the game behavior among multiple power generation companies, energy storage companies and load users, which meets the adjustment in the virtual platform. Nash equilibrium of the overall income of resource participants. Based on the modeling of the adjustable resource aggregation of the virtual power plant, the game strategy is divided into the overall cooperative game and the partial non-cooperative game according to the game characteristics, and different game strategies are adopted respectively. The comparison results of field operation examples and methods prove that the strategy has advantages in terms of model training time-consuming, execution time-consuming, convergence, etc., and it has theoretical guidance and field promotion value.