Optimization of electrochemical performance of a solid oxide fuel cell using Artificial Neural Network

A neural network model is developed for prediction of solid oxide fuel cell performance. The back propagation algorithm is used for the cell voltage and power prediction. As the model is developed, the neural network model's prediction is presented and compared with the physical non-linear model results. Hence, the neural network structure based on the Levenberg Marquardt back propagation algorithm has been concluded to be more appropriate for modeling the dependencies of the non-linearity on the performance of solid oxide fuel cell.