Modeling of Cathode Pt /C Electrocatalyst Degradation and Performance of a PEMFC using Artificial Neural Network

Modeling is of a great importance in developing fuel cell technology with less possible expenses. A good understanding of chemical, physical and mechanical processes of the operating system and related equations are needed to model fuel cells, which are hard to determine in many cases. Artificial intelligence (AI) is a choice to overcome this difficulty instead of costly experiments. AI systems such as artificial neural networks (ANNs) have been employed to solve, predict and optimize the engineering problems in the last decade. In the present study, capabilities of ANN to predict the performance of proton exchange membrane fuel cell (PEMFC) considering the cathode electrocatlyst layer degradation is investigated. Experimental data are utilized for training and testing the networks. Current density, temperature, humidity, number of potential cycles, Platinum load and fuel/oxidant flow rates, potential cycle time step are considered as the inputs and the cell potential, Platinum mass loss percentage of the cathode and location of Platinum particles, which are diffused into membrane and deposited there, are regarded as outputs of ANNs. Back propagation (BP) algorithm has been used to train the network. It is observed that when the networks tuned finely, the obtained results from modeling are in good agreement with the experimental data and achieved responses of ANN are acceptable.