An Enhanced Optimization Algorithm for Estimating PEM Fuel Cells Parameters: Performance Comparison and Real-World Applicability

Accurate modeling is critical for the efficient design, control, and optimization of a proton exchange membrane (PEM) fuel cell. This problem is regarded as a multi-modal, multi-variate, non-linear optimization problem, and several methods in the literature have been presented to tackle it. However, those methods still suffer from either being stuck into local optima or a slow convergence rate. To solve this problem more accurately in terms of final results and convergence speed, this study proposes a hybrid optimization algorithm based on enhancing the recently proposed crested porcupine optimizer (CPO) using two new optimization methods, namely ranking-based optimization and exploration-exploitation improvement strategies. While the latter strives to further reinforce the exploration and exploitation operators of CPO for escaping from local optima and improving the convergence speed, the former aims to replace solutions with poor performance with those that can avoid slipping into local optima. To show the effectiveness and efficiency of this improved CPO (ICPO), it is tested on five popular PEMFC stacks and contrasted with a variety of high-performance optimization techniques. Several performance measures, including worst fitness, average fitness, best fitness, Wilcoxon rank sum test, standard deviation, and Friedman mean rank, are used in this study to evaluate the superiority and significance of the proposed ICPO against the rival optimizers. According to the experimental findings, ICPO outperforms other competing algorithms across all performance metrics, implying that it is a robust substitute for identifying unknown PEMFC parameters.