Scholarlink

<p>Alccofine-modified mortar can reduce cement demand while improving matrix strength; however, the interaction of Alccofine with graphene oxide (GO) and nano-titanium dioxide (nano-TiO<sub>2</sub>) has not been sufficiently quantified under controlled and source-aware conditions. This study evaluates the strength evolution of Alccofine mortar incorporating GO and nano-TiO<sub>2</sub> using experimental testing supported by compact machine learning (ML). The experimental matrix used Alccofine replacement levels of 0–15% and nanomaterial dosages of 0-0.11% by binder mass. After literature harmonization, the final dataset contained 566 records, including experimental and literature records. Four ML models, namely support vector regression with radial basis function kernel, random forest, gradient boosting, and XGBoost, were trained using physically justified features. The nano-free system showed the best 28-day Alccofine response at 10% replacement, whereas 15% Alccofine + 0.09% GO provided the highest composite strength, reaching 69.07&#xa0;MPa at 28 days and 84.27&#xa0;MPa at 180 days. Gradient Boosting produced the best holdout performance, with R<sup>2</sup> values of 0.993, 0.746, and 0.864 for experimental, literature, and combined datasets, respectively. Variance Inflation Factor (VIF) values below 1.54, residual analysis, repeated splitting, and Y-scrambling supported model robustness. Domain-transfer tests showed poor direct transfer between experimental and literature domains, indicating source-shift sensitivity. The model is therefore interpreted as a harmonized-domain interpolation tool rather than a universal predictor.</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

Strength Evolution of Alccofine Mortar Incorporating GO and Nano-TiO₂: A Source-Aware Experimental–Machine Learning Study

  • 期刊论文

  • 发表日期：

    2026年7月20日

查看全文
  • Venkatesh Wadki,
  • Bhavana B

摘要

Alccofine-modified mortar can reduce cement demand while improving matrix strength; however, the interaction of Alccofine with graphene oxide (GO) and nano-titanium dioxide (nano-TiO2) has not been sufficiently quantified under controlled and source-aware conditions. This study evaluates the strength evolution of Alccofine mortar incorporating GO and nano-TiO2 using experimental testing supported by compact machine learning (ML). The experimental matrix used Alccofine replacement levels of 0–15% and nanomaterial dosages of 0-0.11% by binder mass. After literature harmonization, the final dataset contained 566 records, including experimental and literature records. Four ML models, namely support vector regression with radial basis function kernel, random forest, gradient boosting, and XGBoost, were trained using physically justified features. The nano-free system showed the best 28-day Alccofine response at 10% replacement, whereas 15% Alccofine + 0.09% GO provided the highest composite strength, reaching 69.07 MPa at 28 days and 84.27 MPa at 180 days. Gradient Boosting produced the best holdout performance, with R2 values of 0.993, 0.746, and 0.864 for experimental, literature, and combined datasets, respectively. Variance Inflation Factor (VIF) values below 1.54, residual analysis, repeated splitting, and Y-scrambling supported model robustness. Domain-transfer tests showed poor direct transfer between experimental and literature domains, indicating source-shift sensitivity. The model is therefore interpreted as a harmonized-domain interpolation tool rather than a universal predictor.