Scholarlink

<p>The increasing demand for sustainable and high-performance construction materials has created a need for reliable predictive modeling tools for geopolymer concrete systems. Conventional experimental methods for determining compressive strength are expensive, time-consuming, and often insufficient for capturing the complex nonlinear interactions among fiber reinforcement, precursor materials, and alkaline activator chemistry. This study introduces a novel explainable ensemble machine learning framework for predicting the compressive strength of fiber-reinforced geopolymer concrete incorporating ground granulated blast furnace slag (GGBS). The novelty of this work lies in the integration of multi-model boosting algorithms with interpretability techniques, including SHapley Additive exPlanations (SHAP) and Partial Dependence Plot (PDP) analysis, to simultaneously achieve high predictive accuracy and physical insight into geopolymerization behavior. A comprehensive database of 370 geopolymer mix designs was used to develop a robust predictive framework for compressive strength estimation. Seven machine learning models, including XGBoost, LightGBM, Gradient Boosting, AdaBoost, CatBoost, Decision Tree, and K-Nearest Neighbors, were evaluated using multiple performance metrices. The LightGBM model achieved the best performance with an R<sup>2</sup> of about 0.79 and RMSE of 9.74&#xa0;MPa, indicating reliable prediction capability for heterogeneous fiber-reinforced geopolymer systems. Parametric analysis showed that the water–solid ratio, activator Ca/Na ratio, and additive–binder ratio were the most influential parameters, contributing the majority of strength variation. The proposed framework offers a data-driven approach to reduce experimental effort while supporting sustainable geopolymer mix 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

Evaluate the compressive strength of fiber-reinforced geopolymer concrete incorporating ground granulated blast furnace slag with machine learning approach

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月8日

查看全文
  • Md. Habibur Rahman Sobuz,
  • Md. Kawsarul Islam Kabbo,
  • Samir Mazumder,
  • S. M. Fahim Azmal Bin Hoque,
  • Md. Kawsar Akon,
  • Mohammad Alameri,
  • Mohammed Jameel,
  • Sani Aliyu Abubakar

摘要

The increasing demand for sustainable and high-performance construction materials has created a need for reliable predictive modeling tools for geopolymer concrete systems. Conventional experimental methods for determining compressive strength are expensive, time-consuming, and often insufficient for capturing the complex nonlinear interactions among fiber reinforcement, precursor materials, and alkaline activator chemistry. This study introduces a novel explainable ensemble machine learning framework for predicting the compressive strength of fiber-reinforced geopolymer concrete incorporating ground granulated blast furnace slag (GGBS). The novelty of this work lies in the integration of multi-model boosting algorithms with interpretability techniques, including SHapley Additive exPlanations (SHAP) and Partial Dependence Plot (PDP) analysis, to simultaneously achieve high predictive accuracy and physical insight into geopolymerization behavior. A comprehensive database of 370 geopolymer mix designs was used to develop a robust predictive framework for compressive strength estimation. Seven machine learning models, including XGBoost, LightGBM, Gradient Boosting, AdaBoost, CatBoost, Decision Tree, and K-Nearest Neighbors, were evaluated using multiple performance metrices. The LightGBM model achieved the best performance with an R2 of about 0.79 and RMSE of 9.74 MPa, indicating reliable prediction capability for heterogeneous fiber-reinforced geopolymer systems. Parametric analysis showed that the water–solid ratio, activator Ca/Na ratio, and additive–binder ratio were the most influential parameters, contributing the majority of strength variation. The proposed framework offers a data-driven approach to reduce experimental effort while supporting sustainable geopolymer mix optimization.