Electrochemical Parameter Extraction of Proton Exchange Membrane Fuel Cells Using a Modified Chaos and Lévy Flight–Driven Kepler Optimization Algorithm

Proton exchangemembrane fuel cells (PEMFCs) are important electrochemical energy‐conversion devices whose predictive simulation depends on reliable identification of charge‐transport–related and polarization‐related parameters. Extracting these parameters from nonlinear polarization data remains challenging because semi‐empirical PEMFC models are strongly coupled, multimodal, and sensitive to operating conditions. To address this problem, this study develops a modified Kepler Optimization Algorithm driven by Modified Chaos and Lévy Flight (MCL‐KOA) for electrochemical parameter extraction. The proposed method enhances population initialization through a modified chaotic mapping and improves late‐stage local refinement through Lévy‐flight perturbation, thereby strengthening global exploration and numerical stability. The framework is validated on three PEMFC data sets, including a 12 W Horizon H‐12 micro‐stack, a 6000 W NedStack PS6 industrial stack, and the 1000 W noisy FCLAB FC2 dynamic aging data set from the IEEE PHM 2014 Data Challenge. Compared with five benchmark optimizers, MCL‐KOA consistently achieves the lowest sum of squared errors (SSE) while preserving the physical consistency of the reconstructed polarization characteristics. The results indicate that MCL‐KOA effectively mitigates premature convergence in complex parameter spaces and provides a useful numerical tool for PEMFC modeling, diagnosis, and control‐oriented analysis.