Accurate parameter identification of proton exchange membrane fuel cell models using different metaheuristic optimization algorithms

Abstract

Recently, Numerous metaheuristic techniques have been utilized for the expedient identification of Proton Exchange Membrane Fuel Cells ‎ (PEMFCs) models. The reported techniques can inspect fickle in a wide search space for finding optimal solutions at the appropriate time. In this paper, recent optimization techniques are intended to better identify the unknown parameters of various PEMFCs. Three neoteric metaheuristic techniques of the Gazelle optimization algorithm (GOA), Prairie Dog Optimization Algorithm (PDO), and Reptile Search Algorithm (RSA) have been applied and evaluated. The proposed optimization algorithms have been validated for identifying the parameters of three PEMFCs‎: BCS 500 W PEMFC, SR-12 500 W PEMFC, and 250 W PEMFC stack‎. The sum of the squared errors (SSE) between the estimated voltage and the corresponding measured data was formulated as the objective function (OF). MATLAB/Simulink has been employed to validate the proposed optimization methods. The results showed that the three optimization techniques can solve the Fuel Cell ‎(FC) parameters identification optimization problem. Moreover, there are insignificant distinctions between the three applied methods with regard to their optimal value of the objective function. The finest technique considering the average value of the objective function is GOA for BCS 500 W-PEM with 0.0115, while the worst algorithm is PDO with 0.0112. Additionally, the statistical results prove that the three algorithms have 100%, 99.99%, and 100% tracking efficiencies for GOA, PDO, and RSA, respectively, according to 30 individual launches of BCS 500 W-PEM. The results have been evaluated via those of published articles. The I/V curves achieved employing GOA, PDO, and RSA methods provided a good agreement with the corresponding measured ones with the superiority of the GOA relating to the convergence speed, tracking efficiency, statistical metrics, and estimation accuracy.

Keywords

• Metaheuristic Optimization
• Parameter Identification
• System Modelling