Performance Prediction of Solid Oxide Fuel Cells Based on a Neutral Network Model

Abstract

Predicting various parameters in a fast and accurate way plays an important role for the development of high-temperature sealed reactors such as the solid oxide fuel cells (SOFCs). However, the analysis of SOFC performance parameters mainly focuses on the development of multiphysics models, which is complicated and slow, causing inefficiency in the prediction of key parameters. In our study, a neural network prediction model of SOFCs was developed in combination with the deep learning method. The neural network model was trained by using the results from a multiphysics model validated by experiments. The trained SOFC neural network model was found to have good consistency with the simulation results of multiphysics model. The neural network model was further utilized in a system-level model to provide fast and accurate prediction of the system performance.Keywords: SOFC, Multiphysics model, Neural network model, System simulation