Process parameters optimization in plastic injection molding based on support vector machine and genetic algorithm

摘要

Shorter cycle times and thinner or more complex parts increase continuously the need for precise quality control in plastic injection molding (PIM). Molding optimization is further complicated by operations that move the products directly from the molding machine to assembly stations. Therefore, effective process planning and quality control is essential to maintain the benefits of modern process technology. Many studies show the approach using Back-Propagation Network (BPN) trained based on input/outputs data which were taken from simulation works carried out through a CAE system, such as MoldFlow. They reduced the time required for planning and optimization of process settings. However, the structure of BPN can not be determined easily and it needs more samples. In this study, a predictive model for part warpage is created by support vector machine (SVM) exploiting CAE results. Mold temperature, melt temperature, packing pressure, packing time and cooling time are regarded as process parameters. SVM model is validated for predictive capability and then interfaced with Genetic Algorithm (GA) to solve the problem of optimization for process parameters. The SVM is trained by fewer samples to be a precise predictive model, and the warpage of the initial part model is reduced by about 28.6% through optimizing process parameters by GA.