A hybrid mutational Northern Goshawk and elite opposition learning artificial rabbits optimizer for PEMFC parameter estimation

For the purpose of simulating, controlling, evaluating, managing and optimizing PEMFCs it is necessary to develop accurate mathematical models. The present study develops a mathematical model which uses empirical or semi-empirical equations to estimate unknown model parameters through optimization techniques. This thesis calculates, analyzes and discusses the sum of squares error (SSE) between measured and estimated current and voltage values using parameters derived from multiple optimization techniques for six commercially available PEMFCs: BCS 500 W-PEMFC, 500 W SR-12 PEMFC, Nedstack PS6 PEMFC, H-12 PEMFC, HORIZON 500 W PEMFC and a 250 W-stack PEMFC. To minimize the SSE between measured and estimated current values under these new models we employ an advanced version of Artificial Rabbits Optimization called Mutational Northern goshawk and Elite opposition learning-based Artificial Rabbits Optimizer (MNEARO). Additionally SSE, Absolute Error (AE), and Mean Bias Error (MBE) are computed for different recent methods according to literature on voltage measurement. Other optimization algorithms including ARO, TLBO, DE and SSA are used for comparative analysis purposes. On top of that MNEARO outperforms others in terms of both computational cost as well as solution quality while experiments carried out using benchmark problems indicate its superiority over other meta-heuristics approaches.