Scholarlink

<p>The environmental hazards caused by the cement industry, including carbon dioxide emissions and the production of cement kiln dust, have motivated the development of geopolymer technology as a safer alternative. Due to the high cost and the setting time for geopolymer samples, this article collects data from previous experimental work that is in the normal strength limit and uses it to propose a statistical model to predict the compressive strength of geopolymer mortar modified with palm oil fuel ash (POFA), which has not been investigated in studies to date. The model proposes reducing the time and effort required for preparing trial mixes. This innovation reduces cement usage by replacing it with pozzolanic materials. This article explores using cost-effective pozzolanic materials from waste sources, specifically POFA, in geopolymer mortar. A review of previous experimental work was conducted to collect data on varying proportions of POFA as a partial binder in geopolymer mortar. Statistical models including linear regression, nonlinear regression, and artificial neural networks (ANN) were employed to predict the compressive strength of geopolymer mortar, eliminating the need for trial batches and saving time. The performance of the models was assessed using statistical parameters, including the coefficient of determination (<i>R</i><sup>2</sup>), mean absolute error, root mean square error (RMSE), and scatter index. A comparative analysis showed that the ANN method outperformed the other models regarding efficiency and robustness. Notably, the <i>R</i><sup>2</sup> value for the ANN model was 0.99, which is 310% higher than that for the linear model and 296% higher than the <i>R</i><sup>2</sup> value for the nonlinear model. Also, the RMSE for the testing data was 0.66 MPa, which is lower than the RMSE for both the nonlinear regression (13.36 MPa) and linear regression (13.04 MPa) models. The scatter index for the ANN method ranged from 0 to 0.1, demonstrating the excellent predictive capability of the ANN model compared to the others.</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 Model to Predict the Compressive Strength of Geopolymer Mortar Modified with Palm Oil Fuel Ash, Incorporating Linear, Nonlinear, and Artificial Neural Network (ANN) Approaches

  • 期刊论文

  • 发表日期：

    2025年7月30日

查看全文
  • Hersh F. Mahmood,
  • Rozhnw Omer Mustafa,
  • Bilal Kamal Mohammed,
  • Sozan Sabir Hayder Ali,
  • Soran Abdrahman Ahmad

摘要

The environmental hazards caused by the cement industry, including carbon dioxide emissions and the production of cement kiln dust, have motivated the development of geopolymer technology as a safer alternative. Due to the high cost and the setting time for geopolymer samples, this article collects data from previous experimental work that is in the normal strength limit and uses it to propose a statistical model to predict the compressive strength of geopolymer mortar modified with palm oil fuel ash (POFA), which has not been investigated in studies to date. The model proposes reducing the time and effort required for preparing trial mixes. This innovation reduces cement usage by replacing it with pozzolanic materials. This article explores using cost-effective pozzolanic materials from waste sources, specifically POFA, in geopolymer mortar. A review of previous experimental work was conducted to collect data on varying proportions of POFA as a partial binder in geopolymer mortar. Statistical models including linear regression, nonlinear regression, and artificial neural networks (ANN) were employed to predict the compressive strength of geopolymer mortar, eliminating the need for trial batches and saving time. The performance of the models was assessed using statistical parameters, including the coefficient of determination (R2), mean absolute error, root mean square error (RMSE), and scatter index. A comparative analysis showed that the ANN method outperformed the other models regarding efficiency and robustness. Notably, the R2 value for the ANN model was 0.99, which is 310% higher than that for the linear model and 296% higher than the R2 value for the nonlinear model. Also, the RMSE for the testing data was 0.66 MPa, which is lower than the RMSE for both the nonlinear regression (13.36 MPa) and linear regression (13.04 MPa) models. The scatter index for the ANN method ranged from 0 to 0.1, demonstrating the excellent predictive capability of the ANN model compared to the others.