Multi-population mutative moth-flame optimization algorithm for modeling and the identification of PEMFC parameters

Proton Exchange Membrane Fuel Cells (PEMFCs) stand out as complex nonlinear multivariable systems, and developing a suitable model is crucial for designing the electrochemical conversion devices' redox reaction process. To tackle the issue of parameter identification in fuel cells, this paper proposes a "Multi-population Mutative Moth-Flame Optimization"(MM-MFO) algorithm. Inspired by the diversity found in natural species, this algorithm introduces a mutation strategy based on the fitness of population segments, applying distinct mutation operations to subgroups with varying fitness levels. Consequently, it can overcome the drawbacks of single-population searches that tend to get stuck in local optima. Through testing across eight benchmark functions, MM-MFO exhibits excellent performance in convergence speed and accuracy. Leveraging its strong capabilities, the algorithm is utilized for identifying the parameters of PEMFC models, yielding more suitable parameter values. Compared to other algorithms, MM-MFO can more accurately estimate model parameters.