Scholarlink

<p>This paper presents a novel ensemble learning-based framework for accurately predicting the ultimate axial compressive load-carrying capacity of S/RCFST columns, contributing to the development of resilient and sustainable structural solutions. A comprehensive dataset of 932 experimental samples was compiled, encompassing key design parameters such as column dimensions, material properties, and load capacities. Five ensemble learning algorithms, Decision Tree, Adaptive Boosting, Extreme Gradient Boosting, Categorical Gradient Boosting (CatBoost), and Boosted Regression Tree, were systematically evaluated with extensive hyperparameter tuning to enhance predictive accuracy. The CatBoost model exhibited superior performance, achieving exceptional precision and robustness during the training and testing phases. Trend Consistency Analysis and Monte Carlo Simulations were integrated to validate model reliability, ensuring practical applicability. A graphical user interface was developed to facilitate seamless adoption by engineers and researchers, enabling real-time applications in building information Modeling environments. The findings underscore the potential of AI-driven predictive models in smart infrastructure development, promoting efficiency, accuracy, and sustainability in structural engineering practice.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

AI-powered smart prediction of axial load in CFST columns: a sustainable and resilient structural engineering approach

  • 期刊论文

  • 发表日期：

    2025年4月15日

查看全文
  • Megha Gupta,
  • Satya Prakash,
  • Sufyan Ghani

摘要

This paper presents a novel ensemble learning-based framework for accurately predicting the ultimate axial compressive load-carrying capacity of S/RCFST columns, contributing to the development of resilient and sustainable structural solutions. A comprehensive dataset of 932 experimental samples was compiled, encompassing key design parameters such as column dimensions, material properties, and load capacities. Five ensemble learning algorithms, Decision Tree, Adaptive Boosting, Extreme Gradient Boosting, Categorical Gradient Boosting (CatBoost), and Boosted Regression Tree, were systematically evaluated with extensive hyperparameter tuning to enhance predictive accuracy. The CatBoost model exhibited superior performance, achieving exceptional precision and robustness during the training and testing phases. Trend Consistency Analysis and Monte Carlo Simulations were integrated to validate model reliability, ensuring practical applicability. A graphical user interface was developed to facilitate seamless adoption by engineers and researchers, enabling real-time applications in building information Modeling environments. The findings underscore the potential of AI-driven predictive models in smart infrastructure development, promoting efficiency, accuracy, and sustainability in structural engineering practice.