Benchmarking recently developed metaheuristic algorithms for PEM fuel cell modeling and parameter identification

Proton Exchange Membrane Fuel Cells (PEMFCs) have been recognized as a key technology for sustainable energy systems owing to their high energy conversion efficiency and low emissions. Nevertheless, accurate modeling of PEMFCs remains a difficult undertaking due to their highly nonlinear, complex, and multivariable behavior, making reliable parameter estimation essential for simulation, design optimization, performance analysis, and fault diagnosis. The inherent nonlinear dynamics and variability of PEMFC characteristics further complicate the parameter identification process. To address this challenge, this study presents a systematic investigation and comparative evaluation of five newly developed metaheuristic optimizers to determine a suitable optimization method for PEMFC modeling using Ballard Mark V, Horizon H-12, and Avista SR-12 fuel cell systems. The optimization objective is formulated as an SSE-based minimization problem aimed at reducing the mismatch between the experimentally measured data and the corresponding model predictions obtained using the investigated metaheuristic (MH) optimizers. The investigated algorithms are evaluated based on convergence characteristics, estimation accuracy, robustness, and statistical performance indicators. The results demonstrate that the Dhole Optimization Algorithm (DOA) and the Enterprise Development Algorithm (ED) exhibited superior overall performance, achieving enhanced convergence speed, estimation precision, and robustness. These findings provide valuable insights into the effectiveness of advanced metaheuristic optimization algorithms for PEMFC parameter estimation and facilitate the construction of higher-accuracy and reliable PEMFC models for PEMFC-based energy.