PEMFC water management fault diagnosis method based on principal component analysis and support vector data description

Abstract

A data-driven strategy for diagnosing the water management failure in a Proton Exchange Membrane Fuel Cell (PEMFC) is proposed in this paper. In the proposed diagnosis approach, individual cell voltages are used as the variables for diagnosis. A dimension reduction tool, named principal component analysis (PCA), is used to extract important feature information from diagnostic variables collected at different time points. The pattern recognition tool, named support vector data description (SVDD), is then used to construct hyperspheres, each of which tightly contains a certain kind of data in the feature space. A multi-classification decision strategy, which considers the size of the hypersphere and the distance from the sample to the hypersphere center, is finally proposed to realize fault detection. The experimental results show that the PEMFC stack water management fault can be successfully diagnosed and distinguished based on the PCA and SVDD multi-classification fault diagnosis strategy.