A multi-model ensemble and its portability for the prognostics of direct methanol fuel cells under different dynamic operating conditions

Accurate estimation of State Of Health (SOH) and prediction of Remaining Useful Life (RUL) are all-important to ensure the fuel cells’ reliable operation. During the ageing of Direct Methanol Fuel Cells (DMFCs), the output voltage can be monitored in real-time to acquire the degradation trend. However, traditional prognostic methods relying solely on observed voltage trend regression cannot cope with dynamic changes in operation. The deep-level degradation information, which is merely affected by environmental factors, can be accessed by the Electrochemical Impedance Spectroscopy (EIS) measurements. Using the Equivalent Circuit Model (ECM), degradation covariates such as internal impedances can be identified and tracked over time. This paper proposes an approach that combines internal characterization and direct observation to predict the RUL of DMFCs under dynamic operating conditions. The proposed approach is implemented on two DMFCs of the same type prepared for accelerated ageing tests under different scenarios: one DMFC following the China Light Vehicle Test Cycle (CLTC) for training, and the other following the World Light Vehicle Test Cycle (WLTC) to verify the method’s portability. Compared to the traditional data-driven prediction method, experimental results show that the proposed multi-level degradation indicator-based approach can provide more accurate SOH estimation and RUL predictions. © 2026 Elsevier Ltd.