Fault Diagnosis of Proton Exchange Membrane Fuel Cells

Failures caused by load cycling significantly limit the reliability and durability of proton exchange membrane fuel cells (PEMFC) in vehicular applications. Improper management of load cycling accelerates cell degradation and can lead to severe material attenuation. Integrating an effective fault diagnosis system for early fault detection is key to maintain high cell performance. This paper considers the failure events of flooding and catalytic degradation due to their significant impact on cell lifetime. The square root unscented kalman filter (SRUKF) is adopted to provide optimal state estimates based on a semi-empirical model of a PEMFC. Then posteriori probabilities of flooding and catalytic degradation are calculated based on the Bayes’ Rule and used as fault indication signals. Based on these signals, a decision logic is derived. Simulation results demonstrate that the proposed strategy successfully achieves the objective of early fault detection and isolation.