SOH-Disparity-Aware Energy Management for Multi-Stack Fuel Cells Using Enhanced Soft Actor-Critic Reinforcement Learning

Under stringent environmental regulations, multi-stack fuel cell commercial vehicles are emerging as a key technology for zero-emission transportation. However, due to the “barrel effect,” disparities in the state of health (SOH) among fuel cell stacks can accelerate system degradation and shorten the overall service life. In this study, we firstly propose an index to quantify the SOH disparity among fuel cell stacks and incorporate it into the objective function to be minimized. Then, an Enhanced Soft Actor-Critic (ESAC) reinforcement learning framework is developed, which embeds a three-layer rule-based strategy. Hardware-in-the-loop test results demonstrate that introducing the SOH-disparity-aware term into the objective function effectively mitigates the SOH imbalance phenomenon. Meanwhile, ESAC promotes long-term operation of the multi-stack system at identical constant power levels, which alleviates fuel cell degradation and further reduces SOH disparity among stacks. These findings provide a critical pathway for intelligent energy management in next-generation fuel cell trucks.