Fault Diagnosis of Solid Oxide Fuel Cell Based on a Supervised Self Organization Map Model

Abstract

Too high stack temperature and insufficient reactant gas flow may lead to severe and irreversible damages in a real solid oxide fuel cell (SOFC) power system. Thus, fault monitoring and diagnosis technology is indispensable to improve the SOFC system reliability. A supervised selforganization map (SOM) model is proposed to diagnose the faults of the SOFC system in this paper. Using the supervised SOM model, the multidimensional testing data of the SOFC is mapped into a twodimensional map, and the different region in the out map is represented for one fault mode. The method is evaluated using the data obtained from an SOFC mathematical model, and the results show that the supervised SOM analysis contributes on a very efficient way to the faults diagnosis of the SOFC system.