Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Development of a hybrid stacked machine learning model for predicting compressive strength of high-performance concrete
https://orcid.org/https://orcid.org/0000-0001-8476-8364
https://orcid.org/https://orcid.org/0000-0002-2134-2937
https://orcid.org/https://orcid.org/0000-0001-8600-6160
This paper presents a state-of-the-art hybrid stacked machine learning (ML) model for predicting the compressive strength of high-performance concrete (HPC). The proposed model combines the strengths of five individual ML models, namely support vector machine (SVR), decision tree (DT), random forest (RF), gradient boosting (GB), and extreme gradient boost (XGBoost), and uses particle swarm optimization (PSO) meta-heuristic algorithm to optimize the hyperparameters. Linear regression is used to ensemble the predictions of all base learners to improve overall prediction accuracy. A k-fold cross-validation technique with tenfold is employed to prevent overfitting on the train set of the final stacked model. In addition, several interaction features are created to improve the accuracy of ML models. All the data samples are scaled before training models. The model was trained and evaluated on an experimental dataset from the literatures study and achieved an accuracy of 97.1% in terms of the determination coefficient (R2) and a root mean squared error (RMSE) of 3.8 MPa. The feature importance analysis showed that the water–cement ratio, age, and superplasticizer are the most significant factors affecting the compressive strength of HPC. The proposed model provides a more accurate and efficient way of predicting the compressive strength of HPC, thereby improving its mix design, and enhancing its mechanical properties. It reduces the dependency on laboratory destructive and uneconomical tests, which can save time, resources, and costs in the construction industry. The model can be used by engineers and construction professionals to make informed decisions on the appropriate combination of concrete constituents to achieve the desired strength and durability of HPC.
Development of a hybrid stacked machine learning model for predicting compressive strength of high-performance concrete
https://orcid.org/https://orcid.org/0000-0001-8476-8364
https://orcid.org/https://orcid.org/0000-0002-2134-2937
https://orcid.org/https://orcid.org/0000-0001-8600-6160
This paper presents a state-of-the-art hybrid stacked machine learning (ML) model for predicting the compressive strength of high-performance concrete (HPC). The proposed model combines the strengths of five individual ML models, namely support vector machine (SVR), decision tree (DT), random forest (RF), gradient boosting (GB), and extreme gradient boost (XGBoost), and uses particle swarm optimization (PSO) meta-heuristic algorithm to optimize the hyperparameters. Linear regression is used to ensemble the predictions of all base learners to improve overall prediction accuracy. A k-fold cross-validation technique with tenfold is employed to prevent overfitting on the train set of the final stacked model. In addition, several interaction features are created to improve the accuracy of ML models. All the data samples are scaled before training models. The model was trained and evaluated on an experimental dataset from the literatures study and achieved an accuracy of 97.1% in terms of the determination coefficient (R2) and a root mean squared error (RMSE) of 3.8 MPa. The feature importance analysis showed that the water–cement ratio, age, and superplasticizer are the most significant factors affecting the compressive strength of HPC. The proposed model provides a more accurate and efficient way of predicting the compressive strength of HPC, thereby improving its mix design, and enhancing its mechanical properties. It reduces the dependency on laboratory destructive and uneconomical tests, which can save time, resources, and costs in the construction industry. The model can be used by engineers and construction professionals to make informed decisions on the appropriate combination of concrete constituents to achieve the desired strength and durability of HPC.
Development of a hybrid stacked machine learning model for predicting compressive strength of high-performance concrete
https://orcid.org/https://orcid.org/0000-0001-8476-8364
https://orcid.org/https://orcid.org/0000-0002-2134-2937
https://orcid.org/https://orcid.org/0000-0001-8600-6160
This paper presents a state-of-the-art hybrid stacked machine learning (ML) model for predicting the compressive strength of high-performance concrete (HPC). The proposed model combines the strengths of five individual ML models, namely support vector machine (SVR), decision tree (DT), random forest (RF), gradient boosting (GB), and extreme gradient boost (XGBoost), and uses particle swarm optimization (PSO) meta-heuristic algorithm to optimize the hyperparameters. Linear regression is used to ensemble the predictions of all base learners to improve overall prediction accuracy. A k-fold cross-validation technique with tenfold is employed to prevent overfitting on the train set of the final stacked model. In addition, several interaction features are created to improve the accuracy of ML models. All the data samples are scaled before training models. The model was trained and evaluated on an experimental dataset from the literatures study and achieved an accuracy of 97.1% in terms of the determination coefficient (R2) and a root mean squared error (RMSE) of 3.8 MPa. The feature importance analysis showed that the water–cement ratio, age, and superplasticizer are the most significant factors affecting the compressive strength of HPC. The proposed model provides a more accurate and efficient way of predicting the compressive strength of HPC, thereby improving its mix design, and enhancing its mechanical properties. It reduces the dependency on laboratory destructive and uneconomical tests, which can save time, resources, and costs in the construction industry. The model can be used by engineers and construction professionals to make informed decisions on the appropriate combination of concrete constituents to achieve the desired strength and durability of HPC.
Development of a hybrid stacked machine learning model for predicting compressive strength of high-performance concrete
Asian J Civ Eng
Tipu, Rupesh Kumar (Autor:in) / Suman (Autor:in) / Batra, Vandna (Autor:in)
Asian Journal of Civil Engineering ; 24 ; 2985-3000
01.12.2023
16 pages
Aufsatz (Zeitschrift)
Elektronische Ressource
Englisch
| Springer Verlag | 2024
| Springer Verlag | 2024
Advanced Machine Learning Techniques for Predicting Concrete Compressive Strength
| DOAJ | 2025
Predicting Confined Compressive Strength of Concrete Using Machine Learning Approach
| Springer Verlag | 2023