Learning-Based Energy Management Strategy of Fuel Cell City Buses Considering Power Source Degradation

<div>Fuel cell vehicles (FCVs) offer a promising solution for achieving environmentally friendly transportation and improving fuel economy. The energy management strategy (EMS), as a critical technology for FCVs, faces significant challenges of achieving a balanced coordination among the fuel economy, power battery life, and durability of fuel cell across diverse environments. To address these challenges, a learning-based EMS for fuel cell city buses considering power source degradation is proposed. First, a fuel cell degradation model and a power battery aging model from the literature are presented. Then, based on the deep Q-network (DQN), four factors are incorporated into the reward function, including comprehensive hydrogen consumption, fuel cell performance degradation, power battery life degradation, and battery state of charge deviation. The simulation results show that compared to the dynamic programming–based EMS (DP-EMS), the proposed EMS improves the fuel cell durability while approaching the control effectiveness of DP global optimization. In comparison to the back-propagation-based EMS (BP-EMS), the proposed EMS obtains a 0.37% reduction in the equivalent hydrogen consumption and a 4.72% increase in effective Ah-throughput; the fuel cell performance degradation reduces by 40.09%, balancing the degradation of fuel cell and power battery while ensuring low energy consumption and improving the overall performance of the system. Finally, the adaptability of the proposed strategy to driving conditions is validated in this article.</div>