Energy Management Strategy of PEMFC Hybrid Power Supply System Based on Q-Learning

In order to optimize the efficiency of PEMFC hybrid power supply system and maintain the stability of battery SOC, an energy management strategy based on Q-learning is proposed in this paper. Firstly, by selecting the output power of the battery as the action variable and using the <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\varepsilon$</tex>-greedy algorithm to select the output action, a reward function that comprehensively considers the PEMFC efficiency, battery SOC and other objectives is designed to satisfy the low hydrogen consumption level under the same load power and keep the fuel cell working in the high efficiency range. Reduce power fluctuations to extend fuel cell life. Then, a mode switching strategy of the three-port converter is obtained by interactive learning between the Q-learning agent and the hybrid power supply system. The simulation results show that compared with the state machine strategy, the proposed energy management strategy has obvious advantages in terms of high efficiency interval, SOC maintenance and equivalent hydrogen consumption, which validates the effectiveness of the proposed energy management strategy.