Model Based PEM Fuel Cell Health Monitoring

To identify important fuel cell performance parameters for predicting its performance and assessing its health from its operating data (without performing special tests), this paper describes a model based parameter identification approach. It employs a non-linear programming method to identify optimal values of parameters of a fuel cell model that minimize the difference between the voltage trajectory predicted by the model and that of an actual fuel cell when both are subjected to the same current draws. Using an existing fuel cell model, simulation studies were carried out to demonstrate the feasibility. Specifically, the membrane resistance and a coefficient of the concentration loss term were identified. The study shows that both can be identified accurately even poor initial guesses were used.