Prediction of the molecular weight for a vinylidene chloride/vinyl chloride resin during shear-related thermal degradation in air by using a back-propagation artificial neural network

Thermal degradation of a copolymer is characterized by time-dependent molecular weight change mainly resulting from syntheses of chain-scission and cross-linking reactions. The kinetics models for thermal degradation have been well established. However, the kinetics mathematical model is limited for shear-related thermal degradation of polymers in air because shear stress has uncertain effects on the degradation process. In this paper, the back-propagation artificial neural network (ANN) is adopted to predict both the number-average molecular weight and the weight-average molecular weight of vinylidene chloride (VDC)/vinyl chloride (VC) during shear-related thermal degradation in air. The optimum architectures of ANN, 3−9−9−1 and 3−10−1, have been found for predicting the number-average molecular weight and the weight-average molecular weight of VDC/VC during shear-related thermal degradation, respectively.