Diagnosis of Faults in Proton Exchange Membrane Fuel Cells Using Shallow Neural-Based Approaches

PEMFCs (Proton Exchange Membrane Fuel Cells) are considered to be the best efficient and environmentally friendly energy generation systems. Researchers from academia, government agencies, and the private sector are actively studying these topics. Moreover, various industries in the transportation sector are investigating the application of PEMFC technology. This paper aims to model and diagnose faults in PEMFC systems. However, due to the complexity and risks associated with obtaining flooding and drying scenarios from an experimental setup, this paper will utilize the state space model of a healthy experimental setup to simulate flooding and drying scenarios. The healthy, flooding, and drying models will be developed using Matlab Simulink. After comparing the suggested simulation model with the experimental results, simulations of a real PEMFC system will be employed to diagnose faults in healthy and faulty systems. Additionally, neural network methods will be developed, implemented, and trained using both healthy and faulty datasets. Two neural network methods have been used in this study: the Multilayer Perceptron (MLP) and the Long Short-Term Memory (LSTM). Following the completion of this paper, an accuracy level of 95% to 99% is expected.