Integratie van Prestatie en Productieprocessen voor Ontwerpoptimalisatie: Een Benadering met Surrogaatmodellen

In today's competitive industrial landscape, the demand for rapid product development and customized mass production calls for advanced approaches to design optimization. Traditional methods, which typically address design aspects sequentially, often fall short when applied to complex, multidisciplinary projects. This thesis proposes a framework that integrates diverse design considerations—including product performance, manufacturing and assembly processes, and production costs—into a unified optimization process. Leveraging parametric CAD modeling, finite element simulations, and Design for X principles, the framework optimizes multiple design metrics simultaneously, enabling a more comprehensive and efficient assessment of design solutions. Through a Python-based API, it also facilitates seamless software integration, accommodating both commercial and open-source tools for evaluating quantitative metrics across various design dimensions. The core innovation of this work lies in integrating advanced surrogate models, particularly Gaussian Processes and Artificial Neural Networks, within the framework, along with the development of a novel adaptive sampling algorithm. Together, these techniques enable high-accuracy approximations of complex simulations, significantly reducing computational costs. The adaptive sampling algorithm enhances model accuracy by iteratively selecting sample points around potential optima and constraint boundaries, making it especially effective for complex constrained design optimization problems. This combination allows for efficient, high-quality optimization across the design space, addressing the challenges of integrated design more effectively than traditional approaches. Applications of this framework to engineering test cases demonstrate its ability to yield high-quality, cost-effective designs under constrained computational budgets, making it broadly applicable to real-world industrial challenges. This work advances multidisciplinary design optimization by offering a flexible, efficient approach that integrates multiple facets of product design and optimizes their combined influence on the final product outcome.