Enhancing adaptive health-conscious energy management control strategy with self-learning power correction for fuel cell hybrid electric vehicles

Optimizing hydrogen consumption and suppressing powertrain degradation are critical challenges in energy management for fuel cell hybrid electric vehicles. Conventional strategies are often hampered by static optimization formulations and an inability to autonomously adapt to dynamic operational environments. This study proposes a hierarchical framework that dynamically allocates power and adaptively corrects fuel cell power by integrating voltage degradation, achieving balanced multi-objective trade-offs. First, by targeting battery severity factors and hydrogen consumption rate, the Non-dominated Sorting Genetic Algorithm II is employed to generate a Pareto non-dominated solution set. Subsequently, an adaptive mechanism is introduced to dynamically select optimal power allocation coefficients, enhancing the adaptability under varying driving conditions. To mitigate fuel cell voltage degradation resulting from power fluctuations, a self-learning correction module is designed based on deep reinforcement learning. And the fuel cell protection strategy is proposed to regulate fuel cell power transients, reduce power system performance degradation. Results show that, compared to the Sequential Quadratic Programming strategy, the proposed strategy reduces fuel cell voltage degradation by 16.99 % and overall driving costs by 10.96 %. Both numerical experiments and hardware-in-the-loop experiments confirm the effectiveness of the proposed strategy.