Scholarlink

<p>To minimize the environmental and economic hurdles, waste materials, such as industrial and demolition wastes, are being utilized in concrete production. But there is a great challenge to design a concrete mix with different waste materials in the laboratory. In this study, machine learning (ML) tools are applied with the help of Python interface using Google Colab to predict the compressive strength (CS) of waste material-based concrete. The various ML models have been evaluated by employing the metrics like mean square error (MSE), root mean square error (RMSE), mean absolute error (MAE) and R-squared (R<sup>2</sup>). This involves the development of a stack regression (SR) model that combines 14 base learner models into CatBoost (CB) as meta-model (M-M). By Leveraging 526 datasets from the literature, the base models have undergone training on an 80/20 train-test split, with CB employed as the M-M for SR. With the lowest RMSE of 5.65 and the highest R<sup>2</sup> of 0.87 among the models evaluated, the optimized SR model with CB as an M-M (SR_CB) demonstrates superior performance in forecasting the concrete CS produced using waste materials. Further, the experimental CS values of additional 10 concrete mixes with waste materials from the literature are well compared with the predicted ones from the SR_CB model having error from 0.1 to 9.8% indicating as the robust model. The most significant factors influencing CS have been identified using the SHapley Additive exPlanations (SHAP) analysis. This study streamlines waste material-based concrete mix design, reducing reliance on traditional laboratory testing.</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

Use of machine learning models for prediction of compressive strength of concrete produced with waste materials

  • 期刊论文

  • 发表日期：

    2025年9月12日

查看全文
  • Kshitish Parida,
  • Laren Satpathy,
  • Amar N. Nayaka,
  • Manoj K. Amat

摘要

To minimize the environmental and economic hurdles, waste materials, such as industrial and demolition wastes, are being utilized in concrete production. But there is a great challenge to design a concrete mix with different waste materials in the laboratory. In this study, machine learning (ML) tools are applied with the help of Python interface using Google Colab to predict the compressive strength (CS) of waste material-based concrete. The various ML models have been evaluated by employing the metrics like mean square error (MSE), root mean square error (RMSE), mean absolute error (MAE) and R-squared (R2). This involves the development of a stack regression (SR) model that combines 14 base learner models into CatBoost (CB) as meta-model (M-M). By Leveraging 526 datasets from the literature, the base models have undergone training on an 80/20 train-test split, with CB employed as the M-M for SR. With the lowest RMSE of 5.65 and the highest R2 of 0.87 among the models evaluated, the optimized SR model with CB as an M-M (SR_CB) demonstrates superior performance in forecasting the concrete CS produced using waste materials. Further, the experimental CS values of additional 10 concrete mixes with waste materials from the literature are well compared with the predicted ones from the SR_CB model having error from 0.1 to 9.8% indicating as the robust model. The most significant factors influencing CS have been identified using the SHapley Additive exPlanations (SHAP) analysis. This study streamlines waste material-based concrete mix design, reducing reliance on traditional laboratory testing.