Optimal model identification of the PEMFCs using optimized Rotor Hopfield Neural Network

Abstract

In this paper, a new effective technique is presented for model identification of the Proton-exchange membrane (PEM) fuel cells by a new hybrid model of Rotor Hopfield Neural Network (RHNN). The concept is to lessen the Error between the empirical output voltage and the modeled produced voltage of the PEM fuel cells based on the proposed hybrid RHNN. To improve the structure of the RHNN model, a new metaheuristic, called collective guidance factor-based Pathfinder algorithm (GFPF) has been proposed. To validate the performance of the proposed model, it is performed to a case study and its results are compared with the original RHNN as two Blackbox models. Simulation results indicate 0.5665% and 0.4141% maximum relative error for testing of the original RHNN-based model and the GFPF-RHNN-based model, respectively which is a good accuracy for PEM fuel cells modeling.

Keywords

• Collective guidance factor-based Pathfinder algorithm
• Output voltage
• PEM fuel cell
• Rotor Hopfield Neural Network
• System identification