Predictive Health-Conscious Energy Management Strategy of a Hybrid Multi-Stack Fuel Cell Vehicle
Mohammadreza Moghadari, Mohsen Kandidayeni, Loïc Boulon, Hicham Chaoui
Abstract
The use of a multi-stack fuel cell (FC) hybrid electric vehicle (HEV) has recently attracted a lot of attention as a key to enhance system modularity, efficiency, and redundancy. An essential challenge here is the high operating cost of the FC-HEV, which includes the cost of hydrogen, FC degradation, and energy storage system (ESS). For this purpose, this study proposes a predictive health-conscious (PHC) energy management strategy (EMS) to reduce the total operating cost of a multi-stack FC-HEV. The proposed EMS is a multi-layer strategy. In the upper layer, a rule-based strategy is devised to determine how many FCs should be activated considering the degradation of FCs, requested power, and battery state of charge (SOC) to decrease the hydrogen consumption and prevent extreme FC degradation. The lower layer is a real-time optimization algorithm called model predictive control (MPC) that employs its predictive ability to share the optimal power between FCs and the battery. Unlike the other studies, a new velocity predictor, stacked bidirectional long short-term memory (SBLSTM) networks, is utilized in the lower layer to provide accurate future velocity prediction for the MPC. The performance of the proposed strategy is benchmarked with dynamic programming (DP) and equivalent consumption minimization strategy (ECMS) under two different driving cycles. The obtained results indicate that the proposed PHC-EMS achieves close results to DP (with maximum of 3.728% difference) while outperforming ECMS up to 5.154%.