Online diagnostics of HTPEM fuel cells using small amplitude transient analysis for CO poisoning

Recent developments in materials have allowed PEM fuel cells to operate at higher temperatures and alleviate some of the problems that occur during operation. High-temperature PEM fuel cells are still under development, and very little has been done to study transient conditions, specifically the application of small amplitude load transients for diagnostic purposes. This paper presents the evolution of the fuel cell voltage transient for small current pulses over a range of operating conditions. A fault mechanism in the form of CO poisoning is introduced to further study and evaluate the transients for diagnostic purposes. A new two-stage diagnostic method is proposed based on the voltage transient. The first stage makes use of the discrete S-transform for fault marker identification and provides fast estimations on the fuel cell state of health. The second stage makes use of a population-based incremental learning (PBIL) algorithm for equivalent circuit parameter extraction, required for detailed diagnostics. The method is evaluated for both the healthy and the faulted CO poisoning condition in order to verify performance.