Process parameters optimization using a novel classification model for plastic injection molding

Intelligent technology is widely used to optimize process parameters for injection molding. Traditional process parameter optimization methods for esthetic defects suffer from convergence and stability problems. This paper proposes a novel optimization method, utilizing the fact that the feasible parameter domain is usually sandwiched between two opposite defects when a parameter increases from a low level to a high level. By maximizing the margin between the opposite defects of the samples, optimized parameters are obtained by choosing the parameter combination that is furthest away from both types of defects. Background data is introduced for the initialization of the model. Two practical product experiments are conducted to verify the proposed method, and comparisons are made with the fuzzy reasoning method. The results show that the proposed optimization method has more stable convergence performance and does not suffer from the oscillation problem compared with the fuzzy reasoning method. The injection process under the optimized injection parameters obtained from the proposed method provides a much more stable product quality than traditional methods, with only half the standard deviation and a process capability index eight times higher. This method can also be used for other industry applications that share similar solution distribution characteristics.