Scholarlink

<p>Reinforced concrete corbels (RCCs) are widely used in precast structures to support beams and crane loads. However, existing design standards, such as ACI 318-19, adopt conservative and semi-empirical approaches for their design. These standards limit their application to shear span-to-depth ratios (a/d) of 2 or less and are only experimentally verified for a/d ≤ 1. To address these limitations, this study presents an artificial neural network (ANN) model to predict corbel strength. The model’s results are then compared with four commonly used design codes. A dataset of 330 experimentally tested corbel specimens sourced from the literature was used to train, test, and validate the model. The best-performing ANN architecture (13-10-9-1) showed high predictive accuracy with a Pearson correlation coefficient of 0.990. In contrast, the standard methods showed broad inconsistencies, with correlation coefficients ranging from 0.901 to 0.927 and r<sup>2</sup> values between 0.812 and 0.859. Moreover, a MATLAB programming code was developed to identify the optimal model with the highest overall Pearson value. For practical use, a novel web-based application with a graphical user interface (GUI) was created to facilitate the use of the ANN model for corbel strength prediction. This tool offers a more reliable and efficient alternative to conventional design methods. </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

Experimental data-driven neural network model for predicting RC corbel shear strength via a web-based interface

  • 期刊论文

  • 发表日期：

    2025年6月23日

查看全文
  • Maher K. Abbas,
  • Ammar Yasir Ali

摘要

Reinforced concrete corbels (RCCs) are widely used in precast structures to support beams and crane loads. However, existing design standards, such as ACI 318-19, adopt conservative and semi-empirical approaches for their design. These standards limit their application to shear span-to-depth ratios (a/d) of 2 or less and are only experimentally verified for a/d ≤ 1. To address these limitations, this study presents an artificial neural network (ANN) model to predict corbel strength. The model’s results are then compared with four commonly used design codes. A dataset of 330 experimentally tested corbel specimens sourced from the literature was used to train, test, and validate the model. The best-performing ANN architecture (13-10-9-1) showed high predictive accuracy with a Pearson correlation coefficient of 0.990. In contrast, the standard methods showed broad inconsistencies, with correlation coefficients ranging from 0.901 to 0.927 and r2 values between 0.812 and 0.859. Moreover, a MATLAB programming code was developed to identify the optimal model with the highest overall Pearson value. For practical use, a novel web-based application with a graphical user interface (GUI) was created to facilitate the use of the ANN model for corbel strength prediction. This tool offers a more reliable and efficient alternative to conventional design methods.