A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Enhanced prediction of compressive strength in high-strength concrete using a hybrid adaptive boosting - particle swarm optimization
This article accurately predicts the compressive strength of high-strength concrete (HSC) using the proposed hybrid Adaptive Boosting - Particle Swarm Optimization (AB-PSO) model. A dataset consisting of 413 experimentally tested data points, collected from published studies, was used to train and test the hybrid AB-PSO model. The input variables considered were cement (C), fly ash (F), water (W), fine aggregate (S), coarse aggregate (CO), and superplasticizer (SP), with compressive strength as the output prediction. The performance of the hybrid AB-PSO model was evaluated using various statistical coefficients, including R² (coefficient of determination), MSE (mean squared error), MAE (mean absolute error), and RMSE (root mean squared error). A 10-fold cross-validation method was also employed to assess its accuracy. The results demonstrated that the hybrid AB-PSO model achieved high accuracy, with R² values exceeding 0.88 during training and 0.91 during testing. The hybrid AB-PSO model outperformed the default AB paradigm for predicting HSC compressive strength, improving the R² value by 1.03 times. Furthermore, Shapley Additive Explanations (SHAP) and two-way partial dependence plots (PDP-2D) were used to explore the key factors influencing HSC compressive strength. It was found that cement and superplasticizer significantly affected the compressive strength predictions. Finally, an optimal design strategy for achieving the best compressive strength of HSC was analyzed and discussed.
Enhanced prediction of compressive strength in high-strength concrete using a hybrid adaptive boosting - particle swarm optimization
This article accurately predicts the compressive strength of high-strength concrete (HSC) using the proposed hybrid Adaptive Boosting - Particle Swarm Optimization (AB-PSO) model. A dataset consisting of 413 experimentally tested data points, collected from published studies, was used to train and test the hybrid AB-PSO model. The input variables considered were cement (C), fly ash (F), water (W), fine aggregate (S), coarse aggregate (CO), and superplasticizer (SP), with compressive strength as the output prediction. The performance of the hybrid AB-PSO model was evaluated using various statistical coefficients, including R² (coefficient of determination), MSE (mean squared error), MAE (mean absolute error), and RMSE (root mean squared error). A 10-fold cross-validation method was also employed to assess its accuracy. The results demonstrated that the hybrid AB-PSO model achieved high accuracy, with R² values exceeding 0.88 during training and 0.91 during testing. The hybrid AB-PSO model outperformed the default AB paradigm for predicting HSC compressive strength, improving the R² value by 1.03 times. Furthermore, Shapley Additive Explanations (SHAP) and two-way partial dependence plots (PDP-2D) were used to explore the key factors influencing HSC compressive strength. It was found that cement and superplasticizer significantly affected the compressive strength predictions. Finally, an optimal design strategy for achieving the best compressive strength of HSC was analyzed and discussed.
Enhanced prediction of compressive strength in high-strength concrete using a hybrid adaptive boosting - particle swarm optimization
Asian J Civ Eng
Nguyen, Duy-Liem (author) / Phan, Tan-Duy (author)
Asian Journal of Civil Engineering ; 26 ; 1059-1076
2025-03-01
18 pages
Article (Journal)
Electronic Resource
English