Dispatch Optimization of Energy Management for Microgrid System Based on Firefly Moving Regression with Reinforcement Learning Strategy

The increasing penetration of renewable energy sources and the growing demand for reliable and the development of microgrids (MGs) has become more common. The utilization of Renewable Energy Sources (RES) in Microgrids is more advantageous. This study examines the integration of Distributed Generation (DG), Grid-connected PV systems into a Microgrid. However, the complexity of microgrid energy management systems poses significant challenges in optimizing energy production, consumption, and storage. This paper proposes a novel reinforcement learning approach for microgrid energy management, which integrates a Deep Q-network (DQN) with Moving Regression (MR). The DQN is used to learn the optimal energy management strategy, while the MR is employed to optimize the parameters of the DQN. the results show that it can effectively optimize energy production, consumption, and storage while minimizing energy costs and reducing emissions. The proposed approach also demonstrates improved performance compared with Firefly Moving Regression (FMR) and traditional methods, such as rules-based strategy. In this work, the viability of the proposed method in power dispatch strategy is visualized in Python Environment. The result shown in Grid-connected mode for efficient charge into battery storage when apply Ruled-base is 62.5 W/h, FMR is 107.5 and proposed method such as DQN-MR is 199.3 and for minimize the total cost when apply Ruled-base is NT$ 611, FMR is NT$ 513, and proposed method such as DQN-MR is NT$ 250.