AI-DRIVEN OPTIMIZATION OF STEEL STRUCTURAL DESIGN FOR HIGH-RISE BUILDINGS: A SYSTEMATIC LITERATURE REVIEW

Abstract

The design of steel structures for high-rise buildings involves complex trade-offs among safety, cost, material efficiency, and dynamic performance under wind and seismic loads. Recent advances in artificial intelligence offer transformative potential for addressing these challenges through optimization, prediction, and generative design. In this systematic literature review, we aimed to synthesize and critically evaluate the state of research on AI-driven optimization of steel structural design for high-rise buildings. Our methodology followed the PRISMA guidelines to ensure transparency and replicability; we searched across major academic databases using a structured set of keywords, then screened and classified the retrieved studies according to eight predefined dimensions, including AI-driven optimization algorithms, machine learning for response prediction, generative design, vibration control, topology and shape optimization, integration with BIM and digital twins, and structural health monitoring. The review revealed that evolutionary algorithms and reinforcement learning are widely applied to optimize member sizing and topology, while deep neural networks increasingly serve as surrogate models to accelerate seismic and wind response simulations. Generative adversarial networks and variational autoencoders show promise for producing novel structural layouts, often with improved material efficiency. However, we found that most studies remain limited to simplified benchmark problems or low-rise structures, with few validated against full-scale high-rise building cases. Furthermore, integration with real-time construction management and digital twin frameworks is still nascent. We conclude that the field holds substantial promise but requires more rigorous validation, standardized performance metrics, and closer collaboration between AI researchers and practicing structural engineers to bridge the gap between algorithmic innovation and practical deployment.