QSPR study of permeability coefficients through low-density polyethylene based on radial basis function neural networks and the heuristic method

Abstract Traditional quantitative structure–permeability relationship (QSPR) is performed for the study of permeability coefficients of various compounds through low-density polyethylene at 21.1 °C. Descriptors calculated from the molecular structures alone were used to represent the characteristics of the compounds. The three molecular descriptors selected by the heuristic method (HM) in CODESSA were used as inputs for radial basis function neural networks (RBFNNs). The results obtained by RBFNNs were compared with those by HM. The root-mean-squared errors (RMS) for the whole data set given by HM and RBFNNs were 0.4565 and 0.3461, respectively, which shows the RBFNNs model is better than the HM model. The prediction results are in agreement with the experimental values. This paper provided a potential method for predicting the permeability coefficient in polymer science.