PEM Fuel Cell Design Simulation for Electric Vehicles Using Artificial Neural Networks

Abstract

The recent research on fuel cell design has shown the effectiveness of the simulation tools on saving time and money, so we propose a fuel cell design method using artificial neural networks (ANN) for the proton exchange membrane which is the most common and commercially used fuel cell type. We train an artificial neural network on previously performed fuel cell design experiments proposed in another research as a dataset, then we test the trained model by simulating the output power density that can be obtained from user input design data. The used dataset employs commonly used cathode, anode, and membrane types which allows the simulation process using the same materials which are commercially available. We show that the software simulation process using ANN is so beneficial and can produce accurate simulation results imitating the real-world design data.