Generative AI in Engineering Design: Creation and Optimization of Engineering Models in Computer Graphics

Abstract

Generative artificial intelligence has emerged as a transformative approach in modern engineering design and computer graphics, enabling automated creation and optimization of complex structural models. This study proposes a generative AI-based framework that integrates deep learning models, including generative adversarial networks (GANs) and diffusion models, with computer-aided design (CAD) systems to enhance engineering design efficiency and innovation. The system learns geometric and structural patterns from large-scale datasets and generates multiple feasible design alternatives under given physical and functional constraints. A hybrid evaluation mechanism combining physics-based simulation and optimization algorithms is employed to ensure structural validity and performance efficiency. Experimental results demonstrate a 62% reduction in design time, a 91.4% feasibility rate, and significant improvements in material efficiency and structural optimization. The proposed framework also enhances design exploration capability and supports intelligent decision-making in engineering workflows. These findings confirm that generative AI can significantly advance automated engineering design, making it more efficient, adaptive, and innovation-driven.