Identification of a PEMFC fractional order model

The present paper addresses the important issue of monitoring the operating state of the Polymer Electrolyte Membrane Fuel Cell systems. The monitoring system takes a model based approach. Its originality lies in adopting a fuel cell fractional order impedance model which permits to provide a better insight into the fuel cell physical phenomena without increasing the number of parameters. This article first validates experimentally the accuracy of the suggested model, using a frequency identification method carried out by nonlinear optimization using single fuel cell experimental impedance spectroscopy data. In a second phase, time series identification is achieved using a least square method specifically designed for fractional order models. The latter method is first verified on registered data which represents a basic tool for offline monitoring. Subsequently it is refined as a recursive tool permitting an online monitoring; it is validated on laboratory test bench. •The problem addressed: Diagnosis of water problems in a fuel cell.•Development of a new PEMFC fractional order model: more compact and accurate.•Model's parameters considered as good indicators of flooding and drying issues.•Identification of the model's parameters using experimental data: EIS (frequential data), time series current/voltage measurements.•Least square method adapted to fractional order models for offline diagnosis, and recursive least square method for online diagnosis.