A new method of parameter identification for proton exchange membrane fuel cell based on hybrid particle swarm optimization with differential evolution algorithm

Abstract

With the characteristics of high energy conversion efficiency, high energy den?sity and low operating temperature, the proton exchange membrane fuel cells (PEMFC) have become one of the green energy sources with broad prospects. The establishment of accurate mathematical model of the PEMFC is the basis of simulation and control strategy. At present, some intelligent algorithms have certain drawbacks, and can hardly find the balanced point between precision and computational time. In this study, a novel parameter identification approach com?bining the hybrid particle swarm optimization (PSO) algorithm with differential evolution, i.e. hybrid DEPSO, is proposed to obtain the unknown parameters in the PEMFC mathematical model and solve the problems of premature convergence of PSO and poor global search ability of differential evolution. Six benchmark functions are applied to verify the performance of the algorithm. The results prove that the hybrid DEPSO can evade local optimum preferably while having swifter convergence rate. Two PEMFC stacks are investigated and modeled. In order to evaluate the accuracy of model, the sum of squared errors between the measured voltage and the estimated output voltage are examined. Numerical results show higher accuracy of the hybrid DEPSO-based model comparing with other recently published optimization approaches. Furthermore, the simulation results indicate that the accuracy of the PEMFC model optimized by the hybrid DEPSO algorithm improves 0.19-1.86%, which can provide a new solution the multi-objective opti?mization problem and promote the practical application of the PEMFC.