Evaluating the impact of waste marble on the compressive strength of traditional concrete using machine learning DOI Creative Commons
Kennedy C. Onyelowe, Viroon Kamchoom‬, Ahmed M. Ebid

et al.

Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: April 18, 2025

Waste marble, an industrial byproduct generated from marble cutting and polishing processes, can be effectively utilized as a partial replacement in concrete mixtures. Incorporating waste not only addresses environmental concerns related to disposal but also contributes the sustainability of construction materials. Using machine learning (ML) predict impact on compressive strength traditional offers several advantages over repeated laboratory experiments. ML powerful alternative costly time-consuming experiments, enabling faster more sustainable exploration potential improving concrete's strength. This research has focused evaluating using (ML). Advanced techniques such Group Methods Data Handling Neural Network (GMDH-NN), Support Vector Regression (SVR), K-Nearest Neighbors (kNN) Adaptive Boosting (AdaBoost) have been applied this work. The GMDH-NN model was created GMDH Shell 3.0 software, while AdaBoost, SVR kNN models were "Orange Mining" software version 3.36. Error indices sum squared error (SSE), mean absolute (MAE), (MSE), root (RMSE), (%), performance metrics Accuracy % R2 between predicted calculated parameters used evaluate overall behavior models. Finally, Hoffman sensitivity analysis procedure determine individual relative input variables output. At end total 1135 entries collected containing constituents cement density (C), (WM), fine aggregate (FAg), coarse (CAg), water (W), superplasticizer (PL) curing age (Age) model. records divided into training set (900 = 80%) validation (235 20%) following standard partitioning pattern reported literature. with SSE 1408.5 MPa2 1397 respectively tie 95.5% 0.985 showed best suggesting excellent worst. Conversely, RF balances accuracy complexity, making it practical AdaBoost. And lastly, Age, Coarse Aggregates, Water, Plasticizer play most significant roles determining strength, Cement, Marble, Fine Aggregates comparatively smaller impacts. However, considering proportion required for powder replace cement, remarkable influence thus recommended its cement.

Language: Английский

Evaluating the impact of waste marble on the compressive strength of traditional concrete using machine learning DOI Creative Commons
Kennedy C. Onyelowe, Viroon Kamchoom‬, Ahmed M. Ebid

et al.

Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: April 18, 2025

Waste marble, an industrial byproduct generated from marble cutting and polishing processes, can be effectively utilized as a partial replacement in concrete mixtures. Incorporating waste not only addresses environmental concerns related to disposal but also contributes the sustainability of construction materials. Using machine learning (ML) predict impact on compressive strength traditional offers several advantages over repeated laboratory experiments. ML powerful alternative costly time-consuming experiments, enabling faster more sustainable exploration potential improving concrete's strength. This research has focused evaluating using (ML). Advanced techniques such Group Methods Data Handling Neural Network (GMDH-NN), Support Vector Regression (SVR), K-Nearest Neighbors (kNN) Adaptive Boosting (AdaBoost) have been applied this work. The GMDH-NN model was created GMDH Shell 3.0 software, while AdaBoost, SVR kNN models were "Orange Mining" software version 3.36. Error indices sum squared error (SSE), mean absolute (MAE), (MSE), root (RMSE), (%), performance metrics Accuracy % R2 between predicted calculated parameters used evaluate overall behavior models. Finally, Hoffman sensitivity analysis procedure determine individual relative input variables output. At end total 1135 entries collected containing constituents cement density (C), (WM), fine aggregate (FAg), coarse (CAg), water (W), superplasticizer (PL) curing age (Age) model. records divided into training set (900 = 80%) validation (235 20%) following standard partitioning pattern reported literature. with SSE 1408.5 MPa2 1397 respectively tie 95.5% 0.985 showed best suggesting excellent worst. Conversely, RF balances accuracy complexity, making it practical AdaBoost. And lastly, Age, Coarse Aggregates, Water, Plasticizer play most significant roles determining strength, Cement, Marble, Fine Aggregates comparatively smaller impacts. However, considering proportion required for powder replace cement, remarkable influence thus recommended its cement.

Language: Английский

Citations

0