Application of neural network for the prediction of chemical exergy: application in exergy and economic analyses

Abstract

ABSTRACTChemical exergy values of various substances are one of the most important parameters in the exergonomic analysis of chemical processes. In this present contribution, artificial neural network coupled with radial basis function (RBF) neural network was utilized for the prediction of standard chemical exergy of materials. The numbers of overall, training, and testing data used for the development of the neural network model are 135, 113, and 22, respectively. To develop a model successfully and with high accuracy, the atom number, polarizability factor, and molar mass of substances were considered as the input variable and standard exergy of substances was assumed as the output parameter of the model. Statistical parameters such as coefficient of determination and average absolute relative deviation for the modeling results were reported. Moreover, scatter and histogram plots were used for the estimation of model accuracy and robustness. Model outputs reveal that the RBF neural network greatly predicts the chemical exergy values of organic substances. Hence, this model can help engineers in applying the exergonomic analysis of the chemical process.