Abstract: In the paper, adopt Taguchi DOE method and BP (Back Propagation) neural network technology, the effects of four factors, molding temperature, molding time, injection pressure, and molding pressure on the expansion ratio of EVA were studied with a four-level orthogonal experiment. It was concluded that the effect of molding temperature was the most obvious, that of molding time the second, and those of injection pressure and molding pressure are smaller. Taking molding temperature, molding time, and injection pressure as the parameters, another orthogonal experiment was carried out for different shapes and sizes of the standard sample molds. The experiment results were used as neural network sample data, and expansion ratio of EVA plastic could be predicted more accurately using after trained neural network.

Key words: Taguchi method of experimental design, Back Propagation neural network, ethylene-vinyl acetate copolymer, expansion ratio, prediction