Prediction performance of PEM fuel cells by gene expression programming

Abstract

In the present study, gene expression programming has been utilized to evaluate the output voltage of different PEM fuel cells as the performance symbol of these structures. A total number of 843 data were collected from the literature, randomly divided into 682 and 161 sets, and then trained and tested, respectively by different models. The used data as input parameters were consisted of current density, fuel cell temperature, anode humidification temperature, cathode humidification temperature, operating pressures, fuel cell type, O2 flow rate, air flow rate and active surface area of the PEM fuel cells. According to these input parameters, in the gene expression programming models, the voltage of each PEM fuel cell in different conditions was predicted. The training and testing results in the gene expression programming model have shown an acceptable potential for predicting voltage values of the PEM fuel cells in the considered range.