Prognosis of fuel cell degradation under different applications using wavelet analysis and nonlinear autoregressive exogenous neural network

This paper presents the degradation prognosis of Proton Exchange Membrane Fuel Cell (PEMFC) operated under several conditions based on combination two types data: data from postal fuel cell hybrid electric vehicles equipped with PEMFC and carrying out their delivery missions laboratory. The is wavelet analysis Nonlinear Autoregressive Exogenous Neural Network (NARX). influences historical state, operating (load current, relative humidity, temperature, hydrogen pressure), global trend, recovery phenomena are considered. Firstly, raw voltage degraded waveform decomposed into multiple sub-waveforms by analysis. Then, each sub-waveform made NARX. Finally, overall gotten combing sub-waveform. Experimental results have shown that novel method which exploits in a reliable model covers over wide range conditions. proposed not only can make an accurate less learning but also use directly experimental large fluctuation.