dimensional prediction of injection molded parts using melt pressure trace and neural network

the variations in plastic injection molding process may lead to the inconsistency of molded parts' dimensions. furthermore, due to the speed of production as well as post-shrinkage of molded parts make the control of process difficult and give inaccurate molded parts. the objective of this research is to predict the dimensions of injection molded parts on the basis of cavity pressure during the molding process. at the first stage of experimentations, the variations of molding process were determined under a base setting condition by using cavity pressure measurement approach. at the second stage, the effects of molding parameters on the cavity pressure profile as well as part's dimensions were studied. following, an artificial neural-network model was implemented capable of predicting the molded part's dimensions based on the cavity pressure. to increase the efficiency of proposed model, three features of the cavity pressure trace encompassing maximum cavity pressure, cavity pressure-time integral value, as well as time to reach the maximum pressure were selected as neural-network inputs.