Safety and Longevity-Enhanced Energy Management of Fuel Cell Hybrid Electric Vehicle With Machine Learning Approach

The safety, life expectancy and operating cost of fuel cell hybrid electric vehicle (FCHEV) are highly dependent on the power allocation among the onboard power sources. Motivated by this, this paper proposes a machine learning-based multi-physical-constrained energy management strategy to improve the driving economy, thermal safety, and durability of FCHEV. In particular, the fully-continues deep deterministic policy gradient (DDPG) algorithm is exploited to optimize the power distribution of FCHEV in a real-time fashion. Within the proposed framework, the thermal and aging behaviors of the hybrid power sources are scrutinized and optimized, for the first time, to enhance the safety and life performance of FCHEV. The proposed strategy is tested under typical road missions for validation. The unexpected temperature build-up of lithium-ion battery (LIB) and the degradation of hybrid system can be well suppressed to enhance the thermal safety and life performance. Moreover, comparative results suggest that the proposed strategy can optimize the hybrid sources split collaborated with the improvement of operating performance, economy performance and operating time.