Online Data-driven Fault Detection for the HTPEM Fuel Cells based on EIS: Equivalent Electrical Circuit Model Analysis

Online diagnosis of faults in the fuel supply system of the high temperature proton exchange membrane (HTPEM) fuel cell was studied based on electrochemical impedance spectrum (EIS). EIS data were collected under different faulty conditions with changing loading currents. The parameter identification of various equivalent electrochemical circuit models (ECMs) was implemented and analyzed in terms of the interpretability, fitting accuracy, fitting time, and parameter consistency. The results show that the accuracy and interpretability were improved with the increasing number of electrical parameters and the complexity, whereas it can reduce the parameter consistency and increase the fitting time. And parameter identification with frequency above 0.5 Hz using ECM8 can present ideal accuracy and consistency. The faults will be identified based on the support vector machine (SVM) method with features extracted from ECM8. This work aims to provide insights into a more robust strategy for EIS-based fault diagnosis in fuel cells.