Abstract
The artificial electric field algorithm (AEFA) is a recent physics population-based optimization approach inspired by Coulomb's law of electrostatic force and Newton's law of motion. In this paper, an alternative version of AEFA called mAEFA is proposed to boost the searchability and the balance between the explorations to the exploitation of the original AEFA. To escape dropping on the local points in the mAEFA, three efficient strategies for instance; modified local escaping operator (MLEO), levy flight (LF), and opposition-based learning (OBL), are in conjunction with the original AEFA. The convergence rate will be improved when the best agent is identified; thus, stagnation at a local solution can be efficiently avoided. To assess the performance of the proposed mAEFA, it has been evaluated over the CEC'2020 test functions. Furthermore, a robust methodology based on mAEEA is proposed to identify the best parameters of PEM fuel cell (PEMFC). The model of the PEMFC includes nonlinear characteristics that involve several unknown design variables. Thus, it is challenging to develop an accurate model. There are seven design variables to be tuned to reach the targeted dependable model. Two different types of PEMFCs: NedStack PS6 and SR-12 500 W were used to demonstrate the superiority of the mAEEA. Throughout the optimization process, the unidentified parameters of PEMFC are appointed to be decision variables. But the objective function, which necessary to be least is represented by the SSE between the calculated PEMFC voltage and the experimental one. Nine recent optimizers are used in the comparison with the proposed mAEEA. According to the main findings, the advantage of the proposed mAEEA in determining the best PEMFC parameters is verified compared to the other optimizers. Lowest SSE, lowest RMSE, minimum stranded deviation, maximum efficiency, and high coefficient of determination are achieved by the proposed mAEEA.