AI-Driven Optimization of Wind-Resistant Shear Wall Layouts in High-Rise Buildings

A.G. Khedr; Tim K.T. Tse; Jize Zhang

Abstract:

The rapid growth of the global urban population has increased the demand for high-rise buildings, which address spatial limitations in densely populated areas. However, designing these structures poses significant challenges, including ensuring structural stability, managing lateral loads, and balancing safety with cost-effectiveness. Traditional design processes often rely on trial-and-error methods and the expertise of engineers, which are time-consuming and may not yield economically efficient solutions. This study proposes an automated framework that integrates artificial intelligence (AI) with optimization algorithms to achieve optimal structural layouts for shear wall buildings under wind loads while adhering to architectural and structural constraints. The principal innovation of this research is the development of a computationally efficient framework that synergizes convolutional neural networks (CNNs) as a surrogate model with a genetic algorithm (GA) for structural optimization. The CNN model, trained using a finite element method (FEM) dataset generated with OpenSees, predicts structural responses to vertical and lateral loads. The GA determines the optimal number and location of shear walls. Results demonstrate the GA's effectiveness in solving this optimization problem and the CNN model's accuracy in predicting structural behavior. This approach highlights the potential of AI-driven design to enhance economic efficiency and structural performance in high-rise construction.