Artificial neural network investigation of the structural group contribution method for predicting pure components auto ignition temperature

A theoretical method for predicting the auto ignition temperature (AIT) of pure components is presented. Artificial neural networks were used to investigate several structural group contribution (SGC) methods available in the literature. The networks were used to probe the structural groups that have significant contribution to the overall AIT property of pure components and arrive at the set of groups that can best represent the auto ignition temperature for about 490 substances. The 58 single and binary structural groups listed were derived from the Ambrose, Joback, and Chueh-Swanson definitions of group contributions and modified to account for the location of the functional groups in the molecule. The proposed method can predict the auto ignition temperature of pure components, from only the knowledge of the molecular structure, with an average error of 2.8% and a correlation coefficient of 0.98. The results are further compared to the more traditional approach of the SGC method along with other methods in the literature, and shown to be far more accurate. The method is notable for the absence of any method which has previously been used to estimate pure component AIT from their molecular structure alone.