EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

28-day cement strength prediction via transformer-based feature extraction and XGBoost

Dianyuan Ju, Xiaoyu Ma, Rongfeng Zhang iD, Zhao Liu, Xiaohong Wang, Bing Huang

PLOS One · 24.03.2026 · 21 (3) · e0345378

DOI: 10.1371/journal.pone.0345378 ↗

Kurz gefasst

Fachartikel in der Zeitschrift PLOS One, online erschienen am 24.03.2026. Verfasst von Dianyuan Ju und 5 weiteren Autor:innen. Beteiligte Einrichtung: Shandong Provincial Key Laboratory of Preparation and Measurement of Building Materials, University of Jinan, Jinan, China. Hauptthemen laut MEDLINE: Compressive Strength, Construction Materials, Materials Testing. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 30 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Key Research and Development Program of China.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The 28-day compressive strength of cement is a key indicator for assessing cement quality. To overcome the time delays inherent in manual testing, this paper proposed a 28-day cement strength fusion prediction method based on a Transformer feature extractor and an XGBoost meta-learner. This method first encoded the physicochemical multi-source strength variables through the Transformer embedding layer, then calculated the attention scores using the multi-head attention mechanism to allocate weights dynamically. Next, XGBoost’s gradient boosting tree structure and regularization techniques were employed to enhance the robustness of the cement strength prediction model in small-sample scenarios. Finally, the method was validated using real-world 28-day strength testing data from cement plants. The results indicated that, compared to the model without feature extraction, the regression model’s R 2 increased by 5.62%, and its RMSE decreased by 22.33% after applying Transformer feature extraction. Furthermore, when compared with other small-sample models, XGBoost achieved the highest average R 2 of 0.93 in 5-fold cross-validation (CV). Its training efficiency, robustness to noise, and ability to handle feature missingness outperformed other meta-learners. Compared to other methods, TF-XGBoost achieved the highest average R 2 of 0.94 in 25 Monte Carlo (MC) CVs, providing the best fit. The method proposed in this paper demonstrates higher accuracy, better generalization, and greater stability, offering a new approach for the prediction of cement 28-day strength with small sample sizes.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Compressive StrengthConstruction MaterialsMaterials TestingBoosting Machine Learning AlgorithmsPrediction Algorithms

Autor:innen und Institutionen

  • Shandong Provincial Key Laboratory of Preparation and Measurement of Building Materials, University of Jinan, Jinan, China.
  • School of Chemistry and Chemical Engineering, University of Jinan, Jinan, Shandong, China.
  • School of Electrical Engineering, University of Jinan, Jinan, China.

ORCID-Profile: Rongfeng Zhang

Förderung

  • National Key Research and Development Program of China

Zitieren

APA

Ju, D., Ma, X., Zhang, R., Liu, Z., Wang, X., & Huang, B. (2026). 28-day cement strength prediction via transformer-based feature extraction and XGBoost. PLOS One, 21(3), e0345378. https://doi.org/10.1371/journal.pone.0345378

Vancouver

Ju D, Ma X, Zhang R, Liu Z, Wang X, Huang B. 28-day cement strength prediction via transformer-based feature extraction and XGBoost. PLOS One. 2026;21(3):e0345378. doi:10.1371/journal.pone.0345378

BibTeX
@article{Ju2026,
  author = {Ju, Dianyuan and Ma, Xiaoyu and Zhang, Rongfeng and Liu, Zhao and Wang, Xiaohong and Huang, Bing},
  title = {28-day cement strength prediction via transformer-based feature extraction and XGBoost},
  journal = {PLOS One},
  year = {2026},
  volume = {21},
  number = {3},
  pages = {e0345378},
  doi = {10.1371/journal.pone.0345378}
}
Kostenlos starten →