In the present study, a genetic algorithm-polynomial neural network (GA-PNN) was used for modeling proton exchange membrane fuel cell (PEMFC) performance, based on some numerical results which were correlated with experimental data. Thus, the current density was modeled in respect of input (design) variables, i.e., the variation of pressure at the cathode side, voltage, membrane thickness, anode transfer coefficient, relative humidity of inlet fuel and relative humidity of inlet air. The numerical data set for the modeling was divided into train and test sections. The GA-PNN model was introduced with 80% of the numerically-validated data and the remaining data was used for testing the appropriateness of the GA-PNN model by means of two statistical criteria.

This content is only available via PDF.
You do not currently have access to this content.