Faults diagnosis between PEM fuel cell and DC/DC converter using neural networks for automotive applications

Fault tolerance in proton exchange membrane fuel cell (PEMFC) and power converters for automotive applications has become crucial in order to increase the reliability of the power train. As a matter of fact, the occurrence of faults in PEMFC and power converters has undesirable effects on the whole power train such as decreasing of the efficiency and lifetime of the components (PEMFC, converters). The purpose of this paper is to present a fault diagnosis method for PEMFC and DC/DC converter. This fault diagnosis is based on neural networks (NNs) modeling approach combined to numerical simulation in which a new developed sensitive model of PEMFC and an interleaved DC/DC converter have been especially used. Specifically, in this study drying and flooding faults that usually occured in PEMFC according to operations condition variation such as temperatue, humidity and pressure have been considered. Moreover, the power semiconductor failures in DC/DC converter have been taken into consideration in this study.