Experimental study of PEM fuel cell temperature characteristic and corresponding automated optimal temperature calibration model

This study thoroughly analyses and quantifies the operating temperature characterization of the proton exchange membrane (PEM) fuel cell and develops an automated temperature calibration model to precisely identify the optimal operating temperature point corresponding to different current densities. The results show that the automated temperature calibration model integrated with the metaheuristic optimization algorithms and CSO-SVR model has the best overall predictive performance, with the R2 of the predicted values obtained in the training phase and the test phase both exceeding 0.999 and the RMSE less than 2.29 × 10−3 V. In addition, the optimum operating temperature obtained by this model is basically consistent with the experiment value under different current densities, which indicates that the automatic calibration model of fuel cell temperature proposed in this paper has high accuracy and robustness. Therefore, a lot of time-consuming and high-cost experiments can be avoided by using the proposed automatic calibration model of fuel cell temperature. Furthermore, the model and analysis in this paper may provide theoretical support for the thermal control of the vehicle fuel cell system.