Scholarlink

<p>Strength measurements of concrete in the laboratory are fraught with high resource, time, and effort constraints. Destructive testing methods, compressive and flexural strength testing, for example, are used conventionally, and require considerable sample preparation and long curing time, in many cases more than 28&#xa0;days; also, environmental conditions, mix proportions, and material inconsistencies introduce high variance and uncertainty, and hinder accurate and reproducible results. The reliance on empirical models neglecting the complex nonlinear relationship between material properties and performance metrics leads to suboptimal selection of materials. This study attempts to mitigate the above-mentioned issues by incorporating machine learning techniques, particularly Random Forest Regression and Support Vector Machines (SVM), to increase prediction accuracy at significantly less computational complexity. The model had a very high R<sup>2</sup> of 0.92 and a low Mean Absolute Error of 1.5&#xa0;MPa in the compressive strength predictions. This is supplemented by using Support Vector Machines (SVM) with the Radial Basis Function kernel in classifying miscellaneous composite material formulations, according to their contribution to the performance of concrete, which obtained a classification accuracy of 95%. The hybrid model inherits the feature selection capability of random forest with the classification power of SVM. This model reduces by 60% the features considered, while improving the accuracy of classification to 97%. This hybrid method thus forms a robust alternative to appraise durability or strength by facilitating rapid decision-making concerning material selection for high-performance concrete applications. Results indicate that this new hybrid model offers a strong and effective platform for concrete formulation optimization, significantly enhancing the field of high-performance concrete materials. The findings are of tremendous significance in the development of more durable and stronger concrete structures, especially for hostile environmental conditions.</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

Hybrid Machine Learning Based Strength and Durability Predictions of Polypropylene Fiber-Reinforced Graphene Oxide Based High-Performance Concrete

  • 期刊论文

  • 发表日期：

    2025年4月19日

查看全文
  • Monica N. Kalbande,
  • Tejaswini G. Panse,
  • Yashika A. Gaidhani,
  • Rupali S. Balpande,
  • Tejas R. Patil,
  • Niteen T. Kakade,
  • Sruthi Nair,
  • Nilesh Shelke,
  • Haytham F. Isleem,
  • Vikrant S. Vairagade

摘要

Strength measurements of concrete in the laboratory are fraught with high resource, time, and effort constraints. Destructive testing methods, compressive and flexural strength testing, for example, are used conventionally, and require considerable sample preparation and long curing time, in many cases more than 28 days; also, environmental conditions, mix proportions, and material inconsistencies introduce high variance and uncertainty, and hinder accurate and reproducible results. The reliance on empirical models neglecting the complex nonlinear relationship between material properties and performance metrics leads to suboptimal selection of materials. This study attempts to mitigate the above-mentioned issues by incorporating machine learning techniques, particularly Random Forest Regression and Support Vector Machines (SVM), to increase prediction accuracy at significantly less computational complexity. The model had a very high R2 of 0.92 and a low Mean Absolute Error of 1.5 MPa in the compressive strength predictions. This is supplemented by using Support Vector Machines (SVM) with the Radial Basis Function kernel in classifying miscellaneous composite material formulations, according to their contribution to the performance of concrete, which obtained a classification accuracy of 95%. The hybrid model inherits the feature selection capability of random forest with the classification power of SVM. This model reduces by 60% the features considered, while improving the accuracy of classification to 97%. This hybrid method thus forms a robust alternative to appraise durability or strength by facilitating rapid decision-making concerning material selection for high-performance concrete applications. Results indicate that this new hybrid model offers a strong and effective platform for concrete formulation optimization, significantly enhancing the field of high-performance concrete materials. The findings are of tremendous significance in the development of more durable and stronger concrete structures, especially for hostile environmental conditions.