A NUMERICAL VIRTUAL PROCESS MODELER BASED ON COMPUTER AIDED ENGINEERING SOFTWARE FOR INJECTION MOLDING

摘要

A numerical process modeler based on back-propagation neural networks has been employed for emulating the process of injection molding. Contrast to the Computer Aided Engineering simulations, the processing parameters, which are bounded by the process window, are used by the modeler to calculate the selected properties of interests, namely the outputs, by making use of a set of Radial Basis Functions. A systematic procedure for constructing the model parameters has been illustrated based on standard procedures derived from Design of Experiment to obtain the basic training data from the CAE simulations. After going through the iterative training, the process modeler was established for emulating the molding process. Verifications on the process modeler have been shown by randomly choosing the processing parameters in the process window.