Validation of genetic algorithm results in a fuel cell model

The target in this paper is to show how the parameter range, the validation strategy, and the selection of the algorithm effect on the performance of Genetic Algorithms in parameter identification of different fuel cells. The data originates in the current vs. voltage curves (polarization curves) from the published literature. It comes from one fuel cell, and consists on four independent data sets. The results seem promising – a real-coded Genetic Algorithm seems to provide with the model parameters that take the properties of the fuel cell into account. The parameter range is, in this case, the dominating variable, and it needs a careful consideration before identification trials. The histograms of the model parameters during identification can serve as guidelines for the selection of proper ranges for the identified parameters.