Generative AI-driven design and optimization of building structures 

Special lecture at the 60th anniversary of the Department of Architecture at Korea University
The existing design methods of building structures are inefficient and rely heavily on engineers' experience. Therefore, this study proposes a structural design and optimization method driven by generative AI. The AI design algorithm has acquired the ability to generate building structural design schemes by learning from many existing design drawings and comprehensively considering the constraints of architectural layout, design conditions, mechanical principles, and empirical rules. Through deep learning-enabled computational models and evaluation functions, the intelligent optimization algorithm can rapidly optimize and continuously improve structural performance. The design algorithm, integrating intelligent generation and optimization, can efficiently enhance the safety and economy of the generated schemes and overcome the bottlenecks of limited and low-quality training data. Case studies indicate that the structural schemes generated by intelligent design are comparable to those of human experts and generally meet the requirements at the design stage of building structures, significantly improving design efficiency. The developed intelligent design platform has been used by over a hundred design and research institutions for thousands of engineering designs.