Scholarlink

<p>Concrete has been a cornerstone of construction for centuries, but its environmental impact necessitates the development of sustainable alternatives. Ceramic-modified concrete (CMC) offers a promising solution by incorporating waste ceramics, yet accurately predicting its mechanical properties remains challenging due to the material's heterogeneous composition. Traditional methods, such as linear regression and finite element modeling (FEM), often fail to capture the complex interactions between ceramic content and concrete performance, resulting in suboptimal prediction accuracy and limited generalizability. To address these limitations, this study proposed the CMC-MRGCN-TTAO framework, which leveraged an optimized Multidimensional Refinement Graph Convolutional Network (MRGCN) combined with the Triangulation Topology Aggregation Optimizer (TTAO). Experimental results demonstrate that incorporating 20% ceramic waste as a partial replacement yielded the optimal mechanical performance, achieving a 28.4% improvement in compressive strength (from 39 to 50&#xa0;MPa) and a 34.7% increase in tensile strength (from 4.5&#xa0;MPa to 6.1&#xa0;MPa) after 28&#xa0;days of curing compared to conventional concrete. The proposed model outperformed traditional machine learning and deep learning methods, with an R<sup>2</sup> of 0.99 and low RMSE values, making it a reliable tool for sustainable construction material optimization.</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

Predicting mechanical properties and behavioral performance of ceramic-modified concrete using optimized multidimensional refinement graph convolutional networks

  • 期刊论文

  • 发表日期：

    2025年7月23日

查看全文
  • B. Shuriya,
  • V. Balajishanmugam,
  • S. S. Sivaraju,
  • S. Mythili

摘要

Concrete has been a cornerstone of construction for centuries, but its environmental impact necessitates the development of sustainable alternatives. Ceramic-modified concrete (CMC) offers a promising solution by incorporating waste ceramics, yet accurately predicting its mechanical properties remains challenging due to the material's heterogeneous composition. Traditional methods, such as linear regression and finite element modeling (FEM), often fail to capture the complex interactions between ceramic content and concrete performance, resulting in suboptimal prediction accuracy and limited generalizability. To address these limitations, this study proposed the CMC-MRGCN-TTAO framework, which leveraged an optimized Multidimensional Refinement Graph Convolutional Network (MRGCN) combined with the Triangulation Topology Aggregation Optimizer (TTAO). Experimental results demonstrate that incorporating 20% ceramic waste as a partial replacement yielded the optimal mechanical performance, achieving a 28.4% improvement in compressive strength (from 39 to 50 MPa) and a 34.7% increase in tensile strength (from 4.5 MPa to 6.1 MPa) after 28 days of curing compared to conventional concrete. The proposed model outperformed traditional machine learning and deep learning methods, with an R2 of 0.99 and low RMSE values, making it a reliable tool for sustainable construction material optimization.