科研之友
微信
新浪微博
Facebook
分享链接
摘要
An artificial neural network (ANN) implementing a back-propagation algorithm was applied for the prediction of the glass-transition temperature (T(g)) values of 84 polvacrylates and 21 polyvinyls. The experimental T(g) data of the polymers were divided into a training set (50 polyacrylates) and a testing set (34 polyacrylates and 21 polyvinyls). Three molecule descriptors (mean atomic van der Waals volume, bond information content, and three-dimensional molecule representation of structures based on electron diffraction descriptor for signal 13/weighted by atomic masses, Mor13m) were used as input parameters of the neural network. Simulated with the optimum back-propagation ANN 3-[3-2]-1, the root mean square (rms) error for the testing set was 17.7 K, and the correlation coefficient was 0.942, which were accurate in comparison with existing models. The ANN model could be used not only to reveal the quantitative relation between T(g) and the molecular structure but also to predict the T(g) values of the polyacrylates and polyvinyls.
