The bond dissociation energy (BDE) of 80 diverse arylamines was modeled using the five descriptors calculated from the molecular structure along with a quantitative structure-property relationship (QSPR) technique. The heuristic method (HM) and back-propagation neural network method were utilized to construct the linear and nonlinear prediction models, leading to the correlation coefficients R of 0.823 and 0.976. By comparison of the stability with prediction ability of the models, it was found that BP network was a good method for predicting the N—H BDE values of arylamines.

Key words: heuristic method, BP artificial neural network, quantitative structure-property relationship (QSPR), bond dissociation energy (BDE)