An intelligent framework for deriving formulas of aerodynamic forces between high-rise buildings under interference effects using symbolic regression algorithms

Created by W.Langdon from gp-bibliography.bib Revision:1.8414

@Article{Wang:2025:jobe,

• author = "Kun Wang2 and Tianhao Shen and Jingyu Wei and Jinlong Liu and Weicheng Hu",
• title = "An intelligent framework for deriving formulas of aerodynamic forces between high-rise buildings under interference effects using symbolic regression algorithms",
• journal = "Journal of Building Engineering",
• year = "2025",
• volume = "99",
• pages = "111614",
• keywords = "genetic algorithms, genetic programming, High-rise buildings, Interference effects, Symbolic regression, Machine learning, Sensitivity analysis",
• ISSN = "2352-7102",
• URL = "https://www.sciencedirect.com/science/article/pii/S2352710224031826",
• DOI = "doi:10.1016/j.jobe.2024.111614",
• abstract = "Numerous high-rise buildings in megacities create complex interference effects, significantly impacting aerodynamic forces and leading to severe wind-induced disasters. While current machine learning applications predominantly use black-box models with SHapley Additive exPlanations (SHAP) to study these effects, they fall short in providing explicit formulas for practical engineering use. This study pioneers an intelligent framework designed to derive explicit formulas for evaluating interference effects on high-rise buildings. The framework uses multiple symbolic regression algorithms based on genetic programming to generate these formulas, which are then evaluated for accuracy and complexity. Sensitivity and physical trend analyses were performed on the most accurate expressions. In addition, a combination of Extreme Gradient Boosting (XGBoost) and SHAP was used to verify the consistency of the results. The study found that the Offspring Selection Genetic Programming (OS-GP) symbolic regression model excelled in both accuracy and complexity. Sensitivity analysis confirmed the influence of contributing factors on aerodynamic forces, consistent with the results from XGBoost and SHAP, thus further validating the accuracy and interpretability of the OS-GP model. Physical trend analysis revealed that formulas derived from OS-GP align more closely with wind tunnel results compared to those obtained from XGBoost. Overall, the proposed symbolic regression expressions offer significant advantages for engineering applications due to their simplicity and high accuracy, providing valuable guidance for wind-resistant design and urban planning",

}

Genetic Programming entries for Kun Wang2 Tianhao Shen Jingyu Wei Jinlong Liu Weicheng Hu

Citations