Predicting properties of EPDM vulcanizates by using artificial neural network

摘要

In this paper, artificial neural network (ANN) was applied to forecast the properties of EPDM Vulcanizates. Twenty groups of experiment results designed by the method of current rotational and combinatorial design of quadratic regression with three factors were used as the training samples of ANN. Back-Propagation (BP) neural network was established by the neural network tool of MATLAB (MATrix LABoratory software) version 6.5 and the optimum parameters of ANN were chosen. Through training the BP neural network, the well-trained ANN is expected to be very helpful for prediction of EPDM vulcanizates properties including oxygen indexes, tensile strength and elongation at break. The results show that the well-trained ANN can exactly forecast the EPDM vulcanizates properties. ANN based on MATLAB 6.5 also offers an efficient and credible method on analyzing the effect of EPDM vulcanizates' components.