Energy management strategies based on deep learning in grid-forming energy storage systems

Abstract This research focuses on the grid-forming energy storage system (ESS). The deep Q-network (DQN) method is employed to optimize the capacity configuration and operation strategy of the ESS. In this study, an isolated microgrid on a small island is selected as the research subject. By utilizing historical monitoring data, the performance of the DQN and the traditional Q-learning method in the optimization of the ESS is compared. These results indicate that the DQN model demonstrates outstanding performance in terms of the probability of power shortage and the energy curtailment rate of the microgrid.