Parameter identification of a PEMFC semi-empirical model based on improved crayfish optimization algorithm

Mathematical models of proton exchange membrane fuel cells (PEMFCs) are regarded as important tools for performance analysis, state evaluation, and optimal control. Some model parameters cannot be measured directly. Therefore, these parameters must be identified from experimental data. Various intelligent optimization algorithms have been used for PEMFC model parameter identification. However, limitations still exist in identification accuracy, result stability, and computational efficiency. To address these problems, an improved crayfish optimization algorithm (ICOA) is proposed in this paper. A semi-empirical steady-state PEMFC model is used. The polarization loss equations of the model are also optimized. Differential evolution strategy and elite retention mechanism are introduced into the original crayfish optimization algorithm. The unknown parameters of the PEMFC model are then optimized by the proposed ICOA. The proposed method is validated using data from a typical commercial NedStack and a 30 kW PEMFC stack. The results are compared with those obtained by traditional intelligent optimization algorithms, such as the grey wolf optimizer and genetic algorithm. The comparison results show that better parameter identification performance is achieved by ICOA. Higher identification accuracy and stronger global search ability are also obtained.