Parameter optimization method of injection molding process based on genetic algorithm and support vector machine regression

Abstract In response to the prevalent reliance on engineer experience for the adjustment of injection molding process parameters, this study presents an innovative approach which combines Genetic Algorithms (GA) and Support Vector Machine Regression (SVR) to address this challenge. The input variables consist of the process parameters, while the output variable pertains to defects in the molded product. SVR is employed to construct predictive models for these defects, subsequently serving as the fitness function within the GA framework. GA, in turn, is utilized to iteratively determine the optimal injection molding process parameters. To alleviate issues related to overfitting and underfitting, the SVR prediction model was optimized using PSO in combination with five-fold cross-validation. Through this iterative optimization, the SVR model is improved, thereby improving its prediction accuracy. A case study involving automobile mesh confirms the method's effectiveness in reducing defects, showcasing its industrial applicability.