Scholarlink

The axial compressive strength and flexural resistance of reinforced concrete (RC) columns strengthened with Fibre Reinforced Polymer (FRP) composite materials have been extensively studied, aiming at (i) identifying potential advancements in strengthening techniques and (ii) improving design methods. From a technical perspective, accurately predicting the stress-strain behavior of carbon FRP (CFRP)-strengthened RC columns is relevant for engineering design. In this context, this paper investigates the use of machine learning methods to predict the behavior of RC columns strengthened with CFRP systems. In particular, an Artificial Neural Network (ANN) model is employed to estimate the ultimate axial stress and strain of CFRP-confined RC columns. The prediction model considers various influencing factors, such as geometry, unconfined concrete compressive strength, steel and CFRP properties, and load application eccentricity. The network is trained using 264 experimental data points from square-section columns reported in the literature, considering 14 input variables and 2 output variables. The performance of the network is statistically analyzed using regression metrics (R2, MSE, RMSE, MAE, MAPE), scatter plots, and comparisons with international design standards (ACI 440.2R-17, fib Bulletin 90, EN 1992-1-1 Annex J, CNR-DT 200 R1/2013). The results confirm the ANN's accuracy in reproducing the load capacity of CFRP-strengthened RC columns, demonstrating and quantifying the influence of the various input variables.

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

Artificial Neural Network Model for Predicting the Behavior of Confined Reinforced Concrete Columns

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Matheus de Santana Santos,
  • Francisco Evangelista Júnior,
  • João Ramôa Correia

摘要

The axial compressive strength and flexural resistance of reinforced concrete (RC) columns strengthened with Fibre Reinforced Polymer (FRP) composite materials have been extensively studied, aiming at (i) identifying potential advancements in strengthening techniques and (ii) improving design methods. From a technical perspective, accurately predicting the stress-strain behavior of carbon FRP (CFRP)-strengthened RC columns is relevant for engineering design. In this context, this paper investigates the use of machine learning methods to predict the behavior of RC columns strengthened with CFRP systems. In particular, an Artificial Neural Network (ANN) model is employed to estimate the ultimate axial stress and strain of CFRP-confined RC columns. The prediction model considers various influencing factors, such as geometry, unconfined concrete compressive strength, steel and CFRP properties, and load application eccentricity. The network is trained using 264 experimental data points from square-section columns reported in the literature, considering 14 input variables and 2 output variables. The performance of the network is statistically analyzed using regression metrics (R2, MSE, RMSE, MAE, MAPE), scatter plots, and comparisons with international design standards (ACI 440.2R-17, fib Bulletin 90, EN 1992-1-1 Annex J, CNR-DT 200 R1/2013). The results confirm the ANN's accuracy in reproducing the load capacity of CFRP-strengthened RC columns, demonstrating and quantifying the influence of the various input variables.