Compressive strength prediction of metakaolin based high-performance concrete with machine learning
Amgoth Rajender, Amiya K. Samanta
Abstract
The demands of the building sector have significantly intensified the search for the development of high-strength and high-performance concretes due to its conspicuous advantage of high strength and performance, which increases the service life of reinforced concrete structures . The requirement for cement in the construction industry is progressively increasing with the recent trend. The cement industry is one of the world's leading producers of carbon dioxide (CO 2 ). To improve the mechanical properties of the concrete and reduce the liberation of CO 2 during manufacturing, cement is partially replaced with various cementitious materials . In the present work, the cement is partially replaced with metakaolin in high-performance concrete, while the tool of machine learning application has also been utilized. The present work focuses on developing ML-based models to assess the characteristic compressive strength of metakaolin-based high-performance concrete (M60) based on the mix proportions. Four Machine learning algorithms Backpropagation Neural Networks (BPNN), Linear Regression (LR), Random Forest (RF), and Support Vector Machine (SVM) are employed using this dataset. Cement, metakaolin , water content, admixtures, fine aggregate, and coarse aggregate quantities are used as input variables, and the characteristic compressive strength achieved at 28 days of curing is used as the output variable. Employed ‘Coefficient of Correlation’ (R 2 ), ‘Mean Absolute Error (MAE)’, and Root Mean Square Error (RMSE) to validate the experimental data. A value of R 2 equal to or more than 0.9 is considered a good correlation among the input variables. The value of coefficient correlation R2 value achieved about 0.8944 For the Random Forest (RF) algorithm-based model, which shows the higher efficiency among the proposed machine learning algorithms for an optimum 10% percentage of metakaolin as a partial replacement for cement in high-performance concrete.