A hierarchical optimization energy management strategy based on AECMS-MPC for heavy-duty fuel cell hybrid vehicles

Proton exchange membrane fuel cell (PEMFC) hybrid vehicles (FCHVs) present a promising solution for reducing vehicular emissions and improving fuel economy. However, their widespread adoption is constrained by high costs and limited system longevity. Conventional energy management strategies (EMS) focused on fuel efficiency are inadequate in effectively balancing operational costs. This paper introduces a hierarchical optimization adaptive energy management strategy (AECMS-MPC) tailored for heavy-duty fuel cell hybrid trucks. The top-level design integrates a Fuzzy Neural Network (FNN) to plan near-global optimal state-of-charge (SOC) trajectories and an Adaptive Particle Swarm Optimization-Back Propagation Neural Network (APSO-BPNN) for vehicle speed forecasting. At the lower control level, an adaptive Equivalent Consumption Minimization Strategy-Pontryagin's Minimum Principle (ECMS-PMP) algorithm is employed, utilizing predicted future driving conditions to enhance real-time performance and improve prediction accuracy. The integration of these components aims to improve the prediction precision of the Model Predictive Control (MPC) system and optimize real-time performance, ensuring efficient energy management under varying operational conditions. Simulation results demonstrate the effectiveness of the AECMS-MPC algorithm, achieving an 9.54 % reduction in hydrogen consumption compared to rule-based strategies while maintaining power balance. Furthermore, the strategy exhibits superior computational efficiency, adaptability, and robustness. The results indicate that AECMS-MPC can effectively adapt to environments with similar driving characteristics and overcome noise and instability. Additionally, hardware-in-the-loop (HIL) testing confirms the strategy's promising real-time applicability, underscoring its potential for practical implementation in heavy-duty fuel cell hybrid vehicles.