Development of a Fuzzy-Neuro System for Parameter Resetting of Injection Molding

Abstract

An intelligent system has been used for injection molding. Five molded part defects, two mold parameters and the part weight are used as system inputs which are described by fuzzy terms. Twenty process parameter adjusters on an injection molding machine are used as the outputs. A neural network has been trained using the data obtained from test-runs of injection molding. The intelligent system can predict the amount to be adjusted for each parameter towards reducing or eliminating the observed defects. Using this system for the parameter resetting, production time and efforts can be saved drastically. Feasibility studies showed that this intelligent system is capable of reducing the test run time by at least 80 percent.