Intelligent Design of Squeeze Casting Gating System Based on Enhanced Conditional Generative Adversarial Networks and Hypernetworks

Squeeze casting, a hybrid metal forming process that combines the benefits of forging and casting, is widely used in automotive, aerospace, and related industries. Traditional design methods for squeeze casting molds rely heavily on engineers' experience, making it difficult to meet the growing demand for high efficiency and quality. To address these challenges, this study develops an AI-driven design platform for squeeze casting molds, focusing on the gating system and leveraging Siemens NX. A grid projection method is introduced to efficiently characterize part features with minimal data, while parametric gating system modeling is achieved through guide curve techniques. By integrating an enhanced conditional generative adversarial network (ECGAN) with a hypernetwork-based variable structure strategy, sample similarity is improved from 44.3% to 68.4%. The framework combines sample library construction, training parameter optimization, and NX secondary development, unifying grid projection, intelligent algorithms, and guide curve methods. Experimental validation using a new energy vehicle bracket and runner plate confirms the system's effectiveness. It not only generates high-quality solutions for simple parts but also provides valuable design guidance for complex geometries.