Scholarlink

<p>The construction industry, which is one of the greatest contributors of greenhouse gas emission, is under tremendous pressure due to the growing concern about global climate change and its adverse effects on communities. As a sustainable construction material, geopolymer concrete (GPC) has gained popularity due to the environmental issues that come with the production of cement. This study experimentally investigated the impact on compressive strength of partially replacing ground granulated blast-furnace slag (GGBS) with fly ash (FA) and metakaolin (MK) by 0–30% with 5% increments in the GPC activated by sodium hydroxide (NaOH) solution with varying concentrations (4, 6, and 8&#xa0;M) and sodium silicate (water glass) solution. According to the results, adding MK and FA to the concrete increased its compressive strength, whereas increasing the concentration of NaOH decreased its strength. Additionally, an ensemble machine learning technique was used to predict the compressive strength of pozzolanic GPC based on GGBS at 7, 28, and 90&#xa0;days. Experimental data, including compressive strength measurements from 117 concrete specimens prepared from 39 different mixtures, were used to develop the predictive model<b>.</b> An ensemble machine learning model was develop using the samples age, the concentration of NaOH, and the contents of MK, FA, and GGBS as input variables to predict the compressive strength of sustainable GPC modified MK and FA. Findings show that, Gradient Boosting outperformed others, achieving the lower error rates and higher predictive accuracy. Evaluation metrics, including mean squared error, mean absolute error, root mean square error, and correlation coefficient (R<sup>2</sup>), were conducted to prove the accuracy of the model. It was found that the ensemble models' performance levels varied.</p> Graphical abstract <p></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

Machine learning-based prediction of the compressive strength of sustainable geopolymer concrete with metakaolin and fly ash using ensemble techniques

  • 期刊论文

  • 发表日期：

    2025年12月24日

查看全文
  • Adamu Abubakar Sani,
  • Mohamed Mubarak Abdul Wahab,
  • Shuaibu Bello Abubakar,
  • Kamaluddeen Usman Danyaro,
  • Nuruddeen Usman,
  • Shehu Ahmadu Bustani

摘要

The construction industry, which is one of the greatest contributors of greenhouse gas emission, is under tremendous pressure due to the growing concern about global climate change and its adverse effects on communities. As a sustainable construction material, geopolymer concrete (GPC) has gained popularity due to the environmental issues that come with the production of cement. This study experimentally investigated the impact on compressive strength of partially replacing ground granulated blast-furnace slag (GGBS) with fly ash (FA) and metakaolin (MK) by 0–30% with 5% increments in the GPC activated by sodium hydroxide (NaOH) solution with varying concentrations (4, 6, and 8 M) and sodium silicate (water glass) solution. According to the results, adding MK and FA to the concrete increased its compressive strength, whereas increasing the concentration of NaOH decreased its strength. Additionally, an ensemble machine learning technique was used to predict the compressive strength of pozzolanic GPC based on GGBS at 7, 28, and 90 days. Experimental data, including compressive strength measurements from 117 concrete specimens prepared from 39 different mixtures, were used to develop the predictive model. An ensemble machine learning model was develop using the samples age, the concentration of NaOH, and the contents of MK, FA, and GGBS as input variables to predict the compressive strength of sustainable GPC modified MK and FA. Findings show that, Gradient Boosting outperformed others, achieving the lower error rates and higher predictive accuracy. Evaluation metrics, including mean squared error, mean absolute error, root mean square error, and correlation coefficient (R2), were conducted to prove the accuracy of the model. It was found that the ensemble models' performance levels varied.

Graphical abstract