Scholarlink

<p>The axial performance of Cold-Formed Steel (CFS) back-to-back built-up columns was investigated through a combination of experimental testing, finite element (FE) modelling, and machine learning (ML) techniques. Six column specimens with varying web depths and flange widths were tested under axial compression. The results were validated using FE simulations developed in ANSYS. To enhance the dataset for ML modelling, a parametric study involving 60 column configurations was conducted. Predictive models, including linear regression and artificial neural networks (ANN), were employed to estimate axial deformation. The linear regression model produced a predictive equation of y = 0.9803x + 0.0115, with a high coefficient of determination (R² = 0.9907), indicating excellent predictive accuracy. Correlation analysis identified column length and yield strength as the most influential parameters. Evaluation metrics, including Mean Squared Error (MSE), Mean Absolute Error (MAE), Mean Absolute Relative Error (MARE), and Mean Squared Relative Error (MSRE), yielded average values of 0.000058, 0.005779, 0.012620, and 0.000275, respectively. The integrated framework—combining physical testing, validated FE modelling, and data-driven ML prediction—offers a robust and efficient approach for the structural assessment and design optimisation of CFS built-up columns in lightweight steel construction.</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

Axial deformation prediction in back-to-back CFS built-up columns using machine learning

  • 期刊论文

  • 发表日期：

    2025年8月21日

查看全文
  • Papitha Palaian Premalalitha,
  • P. Sangeetha

摘要

The axial performance of Cold-Formed Steel (CFS) back-to-back built-up columns was investigated through a combination of experimental testing, finite element (FE) modelling, and machine learning (ML) techniques. Six column specimens with varying web depths and flange widths were tested under axial compression. The results were validated using FE simulations developed in ANSYS. To enhance the dataset for ML modelling, a parametric study involving 60 column configurations was conducted. Predictive models, including linear regression and artificial neural networks (ANN), were employed to estimate axial deformation. The linear regression model produced a predictive equation of y = 0.9803x + 0.0115, with a high coefficient of determination (R² = 0.9907), indicating excellent predictive accuracy. Correlation analysis identified column length and yield strength as the most influential parameters. Evaluation metrics, including Mean Squared Error (MSE), Mean Absolute Error (MAE), Mean Absolute Relative Error (MARE), and Mean Squared Relative Error (MSRE), yielded average values of 0.000058, 0.005779, 0.012620, and 0.000275, respectively. The integrated framework—combining physical testing, validated FE modelling, and data-driven ML prediction—offers a robust and efficient approach for the structural assessment and design optimisation of CFS built-up columns in lightweight steel construction.