A neural network model in LaNi_xAl_1-xO₃ catalyst for methane reforming in SOFC

There has been a great interest in application of SOFC using hydrocarbon fuel, and the major issue associated with the improvement of steam reforming is the possibility to extend the area of operation conditions, and lead to low carbon formation. The improvement can be achieved by modifying the design of existing nickel catalysts. Generally, mathematical modeling can be considered as an important element in the methane reforming in terms of the avoidance of extensive experiments. In this article, a two layered feed-forward neural network has been trained with the back propagation algorithm to learn parameters in the process that the LaNi_xAl_1-xO₃(x = 0.1-0.9), reforms methane. The data used during the training procedure are generated by using of mass spectrometer after reforming. The average values of the errors for prediction are well below 5% The artificial neural network (ANN) model was capable of modeling the relationship between catalytic behaviors and the structure of catalytic well.