A reconstructed fuel cell life-prediction model for a fuel cell hybrid city bus

Life prediction is a significant and difficult topic for a proton-exchange membrane fuel cell stack, especially a commercial fuel cell stack. This paper proposes a reconstructed fuel cell life-prediction model to estimate the fuel cell lifetime adopted in a city bus. Considering the temperature fluctuation and sensor errors, the voltage model is separated into three parts to simplify the fitting process, and the validation results show that the proposed degradation model is credible and robust. Furthermore, the 14-day training data show that the deviation of the predicted voltage is less than 1%. In this study, the most important innovation is separating the time into different categories. Compared to the traditional method without classification, the proposed model obtains an improved predicted deviation of 50-100%.