Dynamic model of a high power PEM fuel cell system on the basis of artificial neural networks

Polymeric electrolyte membrane fuel cell (PEMFC) systems are potentially promising candidates as alternative energy sources, modeling this kind of system is a difficult task due it is strongly dependent on many physicochemical parameters that cannot be easily measured on the real system. Artificial neural network (ANN) has become in a powerful modeling tool for performance prediction of complex systems where internal variable relationships are no well known. In this paper a commercial 5 kW PEMFC system is successfully modeled by training a multilayer perceptron network (MLP) just acquiring small amount of experimental data, this model is able to predict the behavior of the system without any physical equations achieving an acceptable degree of accuracy.