Quality monitoring of injection molding based on TSO-SVM and MOSSA

Abstract

Based on the tuna swarm optimization-based support vector machine (TSO-SVM) and the multi-objective sparrow search algorithm (MOSSA), this paper proposes a multi-objective optimization approach for injection molding of thin-walled plastic components, addressing the issues of warpage deformation and volume shrinkage that compromise molding quality. Firstly, data samples are obtained based on the Box–Behnken experimental design and computer-aided engineering (CAE) simulation. Subsequently, SVM is employed to build a predictive model between the experimental factors and quality objectives. Additionally, the TSO is applied to optimize the hyperparameters of SVM, aiming to enhance its regression performance and prediction accuracy. Finally, the MOSSA is employed for multi-objective optimization, combined with the CRITIC scoring method for decision-making, to obtain the optimal combination of process parameters. The obtained parameters are then validated through simulation in Moldflow software. After optimization, the warpage deformation is reduced to 0.5085 mm, and the volume shrinkage rate is decreased to 7.573 %, representing a significant reduction of 40.9 % and 18.1 %, respectively, compared to the pre-optimized results. The remarkable improvement demonstrates the effectiveness of the method based on TSO-SVM and MOSSA for the efficient monitoring of the injection molding process.