Application of artificial intelligence techniques to select the objectives in the multi-objective optimization of injection molding

Abstract Injection molding is a highly intricate manufacturing process where the design of the mold plays a pivotal role in mitigating defects such as incomplete parts, flash, sink marks, weld lines, air bubbles, warping, and shrinkage. Effective mold design, particularly during the cooling phase, is paramount for achieving high-quality parts and minimizing cycle time. However, optimizing the cooling phase remains challenging due to the extensive number of variables and their interdependent effects. This study addresses these challenges by focusing on the optimization of conformal cooling channels (CCCs), an advanced cooling technology engineered to significantly improve thermal efficiency and temperature uniformity compared to conventional cooling methods. The research utilizes advanced numerical simulation tools, such as Moldex3D, to assess the performance of CCCs and explore their potential to enhance injection molding outcomes. A key obstacle in optimizing CCCs is the multi-objective nature of the design problem, which initially involves 34 performance criteria, such as temperature gradients, cycle time, and defect rates. To manage this complexity, the primary aim of this study was to reduce the number of objectives while preserving the core attributes of the problem. Principal Component Analysis (PCA) was employed to condense these 34 objectives into four principal metrics, facilitating a streamlined and computationally efficient optimization process. To solve the reduced optimization problem, a suite of Artificial Intelligence (AI) methodologies, including data mining, Artificial Neural Networks (ANNs), and Multi-Objective Evolutionary Algorithms (MOEAs), was deployed. A case study involving a coffee-cup part validated the approach, demonstrating that CCCs achieved substantial improvements in temperature uniformity, defect reduction, and cycle time minimization. The findings underscore the efficacy of integrating AI-driven optimization with numerical modeling, highlighting the transformative potential of CCCs as a state-of-the-art solution for advancing injection molding processes.