Expert Experience–Embedded Evaluation and Decision-Making Method for Intelligent Design of Shear Wall Structures

Abstract

Rapid progress in intelligent design technology for shear wall structures has significantly advanced the field. However, efficient evaluation and decision-making of various intelligent design outcomes remain challenging, particularly in the automated modeling and analysis of artificial intelligence (AI)-generated designs and the rational selection of evaluation indicators. To address this challenge, this study proposes an expert experience–embedded evaluation and decision-making method for the intelligent design of shear wall structures. Specifically, an adaptive multilevel fuzzy comprehensive evaluation method is developed by integrating expert experience into a multiobjective scheme selection process, meeting the multiobjective needs of structural scheme evaluation. In addition, data extraction and automated parametric modeling analysis methods are developed based on the application programming interface of the structural analysis and design software, enhancing the modeling efficiency of structural schemes. Under this approach, effective and automatic evaluation and decision-making across a variety of schemes can be conducted while also uncovering the fundamental principles of a multi-indicator evaluation, offering a foundation for the decision-making of intelligent design schemes. Further, this study also conducts modeling and evaluation work on multiple cases, analyzing the relationship between data and evaluation results to prove the interpretability of the evaluation outcomes. The results demonstrate that the multilevel fuzzy comprehensive evaluation approach effectively meets the requirements of a multi-indicator quantitative evaluation. In addition, the structural design outcomes of generative adversarial networks and diffusion models currently perform well, whereas designs from graph neural networks require further improvement.