Multi-objective optimization of injection molding process parameters for thin-walled shell parts based on a RIME-RF-MOGWO framework

Abstract

Thin-walled components are widely used in automotive interior parts. During injection molding, thin-walled shell structures are highly sensitive to process parameters, which often lead to warpage and volumetric shrinkage, thereby reducing dimensional accuracy and product consistency. To address these coupled quality issues, a multi-objective optimization method for injection molding process parameters of thin-walled plastic parts is proposed based on an integrated RIME-RF-MOGWO framework. Simulation-generated samples are used as the research data, with volumetric shrinkage and warpage deformation selected as the optimization objectives. In the proposed framework, the Synthetic Minority Over-sampling Technique (SMOTE) is adopted to improve data distribution in sparse regions, random forest (RF) is used to establish the nonlinear relationship between process parameters and quality responses, the RIME optimization algorithm (RIME) is employed to optimize RF hyperparameters, and the multi-objective grey wolf optimizer (MOGWO) is used to obtain Pareto-optimal process parameter combinations. Compared with the average quality responses of the original Moldflow simulation samples, the optimized process parameter combination reduced volumetric shrinkage from 15.45% to 12.51% and warpage deformation from 0.6000 mm to 0.2962 mm, corresponding to reductions of 19.02% and 50.63%, respectively. The optimized shrinkage satisfies the target range of 12%-13%, and the optimized warpage is below the 0.3 mm threshold. These results indicate that the proposed framework can improve the molding quality of thin-walled plastic parts under the investigated simulation conditions and provide a practical data-driven approach for injection molding process optimization.