Advances in Civil Engineering (Jan 2024)

Enhancing Concrete Workability Prediction Through Ensemble Learning Models: Emphasis on Slump and Material Factors

  • Jiangsong Jiang,
  • Chunhong Xin,
  • Sifei Wu,
  • Wenbing Chen,
  • Hui Li,
  • Zhaolun Ran

DOI
https://doi.org/10.1155/2024/4616609
Journal volume & issue
Vol. 2024

Abstract

Read online

This study advances concrete workability prediction by integrating ensemble learning models like Random Forest (RF), Extreme Gradient Boosting (XGBoost), adaptive boosting (AdaBoost), and gradient boosted regression trees (GBRTs), and XGBoost showing superior accuracy. Using Shapley additive explanations (SHAPs), it identifies key factors such as the S/log(W) ratio and aggregate components affecting concrete performance. Theoretically, this enhances the predictive analytics framework in material science, while practically, it improves construction efficiency by optimizing material use and reducing physical testing. The development of a user-friendly graphical user interface (GUI) makes these models accessible for industry application. Addressing gaps in model accuracy and interpretability, the research contributes to sustainable construction practices and underscores the need for future expansions in data diversity and computational optimization in the construction industry.