Evaluation Method for Feature Selection in Proton Exchange Membrane Fuel Cell Fault Diagnosis

Considering the fact that various features can be used in proton exchange membrane fuel cell (PEMFC) fault diagnosis, while lack of feature evaluation method brings great difficulty selecting appropriate at practical PEMFC applications, a generalized and selection is urgently required diagnosis. This article proposes novel method, where discrimination capacity robustness are evaluated. With proposed providing accurate consistent diagnostic performance determined. In this study, widely existing diagnosis utilized, which extracted from either voltage or multisensor signals, their effectiveness identifying faults different systems investigated. Results demonstrate with available test data ranked based on results. From findings, for Moreover, early stage also distinguished high ranking features. will beneficial systems, mitigation strategies taken to remove effect due faults.