A Novel In Situ Diagnostic Approach for Fuel Cell Health Diagnostics Using Onboard Sensor Data

Abstract

Fuel cell health diagnostics can help prevent potential failures and reduce maintenance costs. Traditional diagnostic methods usually require specialized equipment and are difficult to implement continuously in commercial fuel cell systems. This study proposes an in situ health diagnostic method based on routinely available onboard sensor signals. A one-dimensional steady-state proton exchange membrane fuel cell (PEMFC) model is first established to analyze the sensitivity of physical and operating parameters, based on which the effective reaction-area parameter ($A_{\mathrm{eff}}$) and high-frequency resistance (HFR) are selected as the primary diagnostic parameters. Cathode pressure and cathode stoichiometry are further selected as controllable perturbation variables, and a simplified parameter-identification model is developed by coupling cathodic reaction kinetics with oxygen transport. The target parameters are rapidly estimated using a genetic algorithm and quasi-steady-state onboard data. The method is validated through controlled short-stack experiments and further evaluated using operational datasets from commercial fuel cell vehicles and an open-air-cooled fuel cell stack system. For systems with available manufacturer specifications, the identified $A_{\mathrm{eff}}$ shows relative deviations of approximately −5.5% to −6.2% from the nominal geometrical active area, whereas the HFR estimation error ranges from −20% to −17%. The results demonstrate the feasibility and adaptability of the proposed method for non-invasive onboard fuel cell condition monitoring and maintenance decision support.