Model identification of the Proton Exchange Membrane Fuel Cells by Extreme Learning Machine and a developed version of Arithmetic Optimization Algorithm

Abstract The major purpose of this study is to provide a proper approach for model identification of the Proton Exchange Membrane Fuel Cells (PEMFCs) to use in different applications. The method introduces a modified version of the Extreme Learning Machine (ELM) model for identification purposes. The modification is based on proposing and designing a new improved metaheuristic, called the developed Arithmetic Optimization Algorithm (dAOA). The algorithm is then verified by various algorithms to demonstrate its effectiveness and is then employed for optimizing the ELM configuration with the aim of minimizing the sum of the square error between the output voltage of the real PEMFC data and the output voltage achieved by the network. Simulation results of the proposed system are validated based on four scenarios that their results are as follows: for 3/5bar with 353.15 K, the maximum training error for dAOA and AOA are 6.01 and 9.899, respectively. For 1/1bar with 343.15 K, the maximum training error for dAOA and AOA are 0.0045 and 0.012, respectively. For 2.5/3bar with 343.15 K, the maximum training error for dAOA and AOA are 0.0132 and 0.0177, respectively, and for 1.5/1.5 bar with 343.15 K, the maximum training error for dAOA and AOA are 0.01and 0.01584, respectively. This shows the better results of the proposed dAOA to provide accurate parameters of the PEMFC stack system.