Interpreting injection molding quality defect using explainable artificial intelligence and analysis of variance

Abstract

Injection molding is a widely employed manufacturing process known for its efficiency, precision, and scalability in producing complex plastic components. However, persistent quality defects, particularly shrinkage, remain a significant challenge, directly impacting product reliability and performance. Traditional studies have relied heavily on Analysis of Variance (ANOVA) to identify and analyze the parameters influencing such defects. While ANOVA is effective in determining the significance of individual factors, it often falls short in capturing the nonlinear interactions and complex dependencies characteristic of injection molding processes. Our study addresses this limitation by adopting a hybrid methodology that integrates ANOVA with explainable artificial intelligence (XAI) techniques, specifically SHAP (Shapley Additive Explanations) and LIME (Local Interpretable Model-Agnostic Explanations). This innovative approach offers a deeper and more interpretable analysis of the factors affecting shrinkage. We investigate five key parameters: mold temperature, melt temperature, injection time, packing time, and packing pressure. The results reveal that the hybrid approach not only identifies significant parameters but also uncovers complex interactions overlooked by ANOVA alone. This nuanced understanding of process dynamics facilitates improved defect prediction and control, advancing the quality of injection-molded products. By bridging the gap in conventional methodologies, our research blends statistical precision with XAI's interpretability, providing a novel framework for optimizing injection molding processes and enhancing product reliability.