Load-adaptive fuzzy energy management architecture for fuel cell hybrid electric heavy-duty trucks

Fuel cell systems offer an effective solution for decarbonizing heavy-duty applications. This study develops a two-layer EMS architecture using systems thinking principles. In its offline layer, dynamic programming generates global optimal solutions that are used to train a fuzzy logic controller. In the online layer, a load indicator parameter and a load adjuster algorithm are proposed to improve adaptability. The strategy is evaluated on a reference driving cycle and tested under three load cases. For generalization, three additional driving cycles are considered, with dynamic programming as the benchmark. The proposed strategy achieves near-optimal performance, with a maximum energy consumption deviation of 3.2% relative to the global optimum under a charge-sustaining constraint for the reference cycle. For unseen cycles, it maintains charge sustenance and limits the average energy consumption increase to 5.18%, demonstrating robustness and adaptability. A detailed root-cause error analysis is provided to identify the sources of the observed errors.