Taguchi-based Methodology for Determining Process Model of Injection Molding Using Neural Network

Implementing CIM (Computer Integrated Manufacturing) often requires models of manufacturing processes to be devised for determining optimal process parameters and designing adaptive control systems. Despite the progress made in analytical (mechanistic) modeling, however, empirical models derived from experimental data are more frequently used in practice. This paper describes the development of a neural network model for injection molding process. The model uses CAE (Computer Aided Engineering) analysis data based on Taguchi method which ensures the effectiveness of the model and the efficient learning by the network. In view of the robust process design, only those input parameters that are not overly sensitive to external disturbances but sensitive enough to injection performance are identified using analysis of variance. The model is compared with the traditional polynomial regression model. Nomenclature j A , j = 1, ..., 0 N CAE analysis outputs df Degree of freedom DSI RMS deviation of SI DWP ave ave LS WP − LS Linear shrinkage ) ( ave ave WP LS Average LS (WP)