Computational Modeling and Intelligent Simulation of PEMFC Parameter Identification: Design and Technical Validation of a Virtual Teaching Experiment Using an Enhanced LRSAO Algorithm

Proton exchange membrane fuel cell (PEMFC) parameter identification is a nonlinear computational modeling problem involving strongly coupled electrochemical parameters and partially unobservable polarization processes. Its experimental teaching is further constrained by the cost of fuel cell stacks and the safety requirements associated with hydrogen operation. To address these challenges, this study develops a computational modeling and intelligent simulation framework for a virtual teaching experiment on PEMFC parameter identification. A semi-empirical output-voltage model is established, and the sum of squared errors (SSE) between measured and simulated voltages is formulated as the optimization objective. An enhanced Logistic–Tent reverse snow ablation optimizer (LRSAO), termed RLFDB-LRSAO, is introduced by integrating roulette-wheel-selection-enhanced fitness-distance balance and Lévy flight perturbation. Its methodological novelty lies in the stage-wise coordination of population-diversity enhancement, candidate-selection guidance, and search perturbation within the LRSAO framework, rather than in the individual component strategies themselves. The framework organizes the identification process into mechanism interpretation, model construction, algorithm implementation, parameter configuration, visualization, comparative evaluation, and reflective analysis. Case studies using the NedStack PS6 and Modular SR-12 stacks yield best SSE values of 1.2173340 and 6.13503904, respectively. A small-scale qualitative teaching evaluation involving 20 postgraduate students indicated that the framework supported programming practice, strengthened conceptual understanding of PEMFC parameter identification and intelligent optimization, and provided useful support for research-oriented skills such as engineering problem analysis, technical writing, and innovation-oriented project development. These results provide preliminary evidence of technical and educational feasibility while supporting a cautious, problem-dependent interpretation of the optimizer.