Prediction of the flash points of alkanes by group bond contribution method using artificial neural networks

Abstract

A group bond contribution model using artificial neural networks, which had the high ability of nonlinear of prediction, was established to predict the flash points of alkanes. This model contained not only the information of group property but also connectivity in molecules. A set of 16 group bonds were used as input parameters of neural networks to study the correlation of molecular structures with flash points of 44 alkanes. The results showed that the predicted flash points were in good agreement with the experimental data that the absolute mean absolute error was 6.9 K and the absolute mean relative error was 2.29%, which were superior to those of traditional group contribution methods. The method can be used not only to reveal the quantitative correlation between flash points and molecular structures of alkanes but also to predict the flash points of organic compounds for chemical engineering.