Health Evaluation of Fuel Cells under Dynamic Conditions based on Negative Resistance Equivalent Circuit Model

Proton exchange membrane fuel cell (PEMFC) is considered as an alternative energy devise to reduce fossil energy usage. However, low durability still hinders the wide application for fuel cell electric vehicles. Degradation prognosis can help to prolong the lifetime of fuel cells, which requires the accurate health evaluation of internal state, especially under dynamic conditions. In this paper, a hybrid health evaluation and prognostic approach is proposed. Specifically, the non-physical negative resistance and capacitance is proposed to establish the equivalent circuit model and therefore capture the dynamic characteristic of fuel cell response. Then, the total polarization resistance is selected as the health indicator for degradation prediction. Finally, a gated recurrent unit (GRU) model with self-attention mechanism is used to predict the degradation behavior of health indicator. The proposed model and prediction method are validated by a long-term PEMFC stack durability test dataset including the nonlinear accelerated degradation.