Real-Time Multiobjective EMS for Fuel Cell Vehicle Considering Energy Source Health Awareness Under Unknown Real-World Conditions

Economic consideration and service life have constrained the development of fuel cell vehicles (FCVs). Moreover, due to the complexity of operating conditions, the energy management strategy (EMS) must adapt to various unknown scenarios. This article proposes a novel health-aware real-time multiobjective (HR-MO) EMS based on vehicle-cloud architecture for medium and heavy-duty multigear FCV. Specifically, a robust energy source degradation model is incorporated into the multiobjective EMS. The vehicle speed prediction model trained on multisource data is finely designed based on bidirectional long short-term memory (BiLSTM) to achieve high accuracy, lightweight, and strong adaptability. The self-adjusting dynamic programming (DP) algorithm is employed as a solver to enhance computational efficiency and energy efficiency. Finally, in order to meet instantaneous power demand, a vehicle-cloud collaborative EMS is designed to optimize the output power of the fuel cell (FC) in the cloud and adaptively adjust the battery power in the vehicle. Simulation results show that the proposed strategy under different unknown scenarios can reduce hydrogen fuel consumption, the high-power/idle operation time of the FC, the number of start/stop cycles, and the variation in FC load. Furthermore, an experiment test is conducted for real-time performance verification.