Prediction of monomer reactivity parameters using quantum chemical descriptors

Abstract

To construct artificial neural network (ANN) models for the prediction of reactivity parameters (u, ν), density functional theory (DFT) calculations are carried out for 55 vinyl monomers, at the B3LYP level of theory with a 6–31G(d) basis set. After the generation of quantum chemical descriptors, the stepwise multiple linear regression (MLR) analysis and the ANN method are used to develop quantitative structure-property relationship (QSPR) models of parameters u and ν. The ANN models produced test set root-mean-square (rms) errors of 0.35 for the parameter u and 0.34 for the parameter ν. Research results indicate that the QSPR models based on DFT calculations and ANN techniques are accurate and possess the ability to generalize.