Robust Parameter Identification of Semi‐Empirical PEMFC Models Using a Stabilized Improved Tornado Optimizer

ABSTRACT This study presents an improved tornado optimizer with a Coriolis‐force‐inspired mechanism (ITOC) for robust parameter identification of semi‐empirical proton exchange membrane fuel cell (PEMFC) models. To alleviate premature convergence and unstable local refinement in highly nonlinear calibration problems, ITOC integrates chaotic initialization, cooperative evolution of multiple air masses, reverse‐flow perturbation, differential learning, and an adaptive trust‐region local refinement strategy. The Coriolis term is numerically stabilized to generate smoother search trajectories and improved convergence reliability. The proposed method is first examined on eight benchmark functions, demonstrating strong accuracy and robustness compared with representative methods considered in this work. It is then validated on four commercial PEMFC stacks (BCS 500 W, Nedstack PS‐6, a 250 W FC stack, and SR‐12). Across 30 independent runs, ITOC achieves lower fitting errors and smaller run‐to‐run variation. The identified parameters enable accurate reconstruction of polarization and power characteristics that agree well with experimental measurements; only minor deviations are observed at high current for SR‐12, whereas the overall trend is preserved. These results indicate that ITOC provides a reliable tool for PEMFC model calibration and performance prediction in hydrogen energy conversion applications.