Artificial neural network prediction of glass transition temperature of polymers

Abstract

In this article, the molecular average polarizability α, the energy of the highest occupied molecular orbital EHOMO, the total thermal energy Ethermal, and the total entropy S were used to correlate with glass transition temperature Tg for 113 polymers. The quantum chemical descriptors obtained directly from polymer monomers can represent the essential factors that are governing the nature of glass transition in polymers. Stepwise multiple linear regression (MLR) analysis and back-propagation artificial neural network (ANN) were used to generate the model. The final optimum neural network with 4–[4–2]2–1 structure produced a training set root mean square error (RMSE) of 11 K (R = 0.973) and a prediction set RMSE of 17 K (R = 0.955). The results show that the ANN model obtained in this paper is accurate in the prediction of Tg values for polymers.