A Kalman filter based approach to PEM fuel cell fault detection

In this paper the online diagnosis of polymeric electrolyte membrane fuel cells is afforded. In the literature, frequency-domain approaches are widely treated. They are often based on the injection of a perturbation on the cell current, which allows determining the cell impedance through the joint analysis of the cell current and voltage. The perturbation can be sinusoidal, with its frequency varying into an assigned range, or more complex. The sinusoidal current and voltage signals used in classical electrochemical impedance spectroscopy can be used also for identifying the key parameters of an equivalent circuit of the fuel cell. This procedure, which is based on a suitable Kalman filter, allows to detect fault conditions affecting the cell, such as drying and flooding. The proposed approach allows to predict a possible fault before the long processing of the huge amount of data needed by the impedance spectroscopy is available.