Accurate Estimation of PEMFC State of Health using Modified Hybrid Artificial Neural Network Models

Abstract

In favor low emissions and high efficiency of fuel cell, Fuel cell is regarded as next generation power devices in smart cities and sustainable mobility. Fuel cells convert the chemical energy stored in fuels to electricity in an electrochemically way. A suitable diagnostic is required to identify the different faults that may occur in fuel cell systems. This paper aims at illustrating a novel technique to increase the service life and understand the aging mechanisms in fuel cell systems by modifying air flow rate (qwin) and humidifying gases to guarantee the proper operation of the PEMFC. In this paper, the artificial intelligence technology (i.e. neural network ANN) is used for determining the overall performance and resistance losses of PEMFC at numerous operating conditions. The proposed model in this study deals with the parameters of the electrochemical impedance spectroscopy and polarization curves, to estimate and diagnose the state of health of the fuel cell in both case flooding and drying out of the FC. This model identifies a set of three parameters of Randles model in different state of humidification, at either low or high relative humidity RH conditions. Simulation experiments show that the proposed technique enables to monitoring the water management in a simple way that helps to define the state of health (SOH) of the PEMFC.