Proton Exchange Membrane Fuel Cell Degradation and Remaining Useful Life Prediction based on Artificial Neural Network

Abstract

the degradation and Remaining Useful Life (RUL) prediction are very important for proton exchange membrane fuel cell (PEMFC) operation. The degradation and RUL prediction of PEMFC are researched by using Artificial Neural Network (ANN) in this paper. A back propagation neural network is applied to construct the PEMFC model. The PEMFC model includes 4 neural layer networks with 2 hidden neural layers. The proposed PEMFC model is constructed based on the operating data of PEMFC. The PEMFC ANN model considers five following variables which affect the degradation of PEMFC: stack current, stack temperature, air pressure, hydrogen pressure and air humidity. The results have shown that the PEMFC ANN model can successfully predict the degradation considering the mean relative error (RE) of degradation prediction is 0.0695%, and the degradation prediction performance of proposed model is better than that of the other model which only consider the PEMFC current and temperature. The RUL prediction can be given with an RE smaller than 5% after learning 400 h.