考虑强度分类的融合机器学习混凝土徐变建模

doi:10.11896/cldb.24100121

Abstract
Figure/Table
References
Related Citation (15)

Download:

Full Text ( PDF ) (7872KB)
Export: BibTeX | EndNote (RIS)

Abstract This study establishes a high-accuracy prediction model for concrete creep through strength classification, employing a fused machine learning approach (Stacking) to address the limited adaptability of existing models across varying strength grades. The Stacking framework, composed of base learners and a meta-learner, integrates predictions from multiple base models through meta-learning to enhance predictive performance. First, outlier detection and feature selection were performed on the creep database. The outlier detection process identified and removed 1 080 outliers in creep compliance measurements from the NU-ITI database. After feature selection, 13 parameters were chosen as input variables for the machine learning model. Second, the NU-ITI database was divided into normal-strength concrete (NSC, f_c＜60 MPa) and high-strength concrete (HSC, f_c ≥60 MPa) datasets to conduct comparative analysis of four base learners’ predictive performance. Results indicated XGBoost’s superior accuracy for NSC and CatBoost’s optimal performance for HSC. Thus, these two algorithms were adopted as the base learners in the Stacking model. Subsequently, both the linear regression (LR) and ridge regression (RR) were tested as meta-learners in the Stacking model. The stacking model with LR and RR as meta-learners achieves comparable accuracy (R²=0.982 6, MAE=3.29 με/MPa, RMSE=5.18 με/MPa), while outperforming all four base learners and existing creep prediction models in the literature. Finally, the SHapley Additive exPlanation (SHAP) method was used to identify key parameters influencing concrete creep in NSC and HSC. Compressive strength, temperature, aggregate-to-cement ratio, and ambient relative humidity are the primary parameters governing the differences in creep mechanisms between two strength grades. Additionally, further validation using the MC2010 and B4 models confirmed that integrating SHAP-based feature importance significantly improves the accuracy of concrete creep predictions.

Key words： concrete creep fused machine learning algorithm strength classification SHAP feature importance

Published: Online: 2026-04-16

CLC number:

TU17

	Service
	E-mail this article
	Add to my bookshelf
	Add to citation manager
	E-mail Alert
	RSS
	Articles by authors
	MEI Shengqi
	LIU Xiaodong
	WANG Xingju
	LI Xufeng
	NIE Liangtao
	KANG Xuejian

Cite this article:

MEI Shengqi,LIU Xiaodong,WANG Xingju, et al. Fused Machine Learning Modeling of Concrete Creep with Strength Classification[J]. Materials Reports, 2026, 40(7): 24100121-8.

URL:

https://www.mater-rep.com/EN/10.11896/cldb.24100121 OR https://www.mater-rep.com/EN/Y2026/V40/I7/24100121

1 Bažant Z P, Kim J K. Materials and Structures, 1991, 24, 409.
2 Bažant Z P, Hubler M H, Yu Q. ACI Structural Journal, 2011, 108(6), 766.
3 Gedam B A, Bhandari N M, Upadhyay A. Journal of Materials in Civil Engineering, 2016, 28(4), 04015173.
4 Le Roy R, Le Maou F, Torrenti J M. Materials and Structures, 2017, 50(85), 1.
5 Hong S H, Choi J S, Yuan T F, et al. Journal of Materials Research and Technology, 2023, 22, 230.
6 ACI Committee 209. ACI 209R-92, Prediction of creep, shrinkage, and temperature effects in concrete structures. American Concrete Institute, 2008.
7 CEB-FIP. fib Model Code for Concrete Structures 2010. Ernst & Sohn, Germany. 2013.
8 Gardner N J, Lockman M J. Materials Journal, 2001, 98(2), 159.
9 Bažant Z P, Murphy W P. Materials and Structures, 1995, 28(180), 357.
10 Bažant Z P, Jirasek Mi, Hubler M H, et al. Materials and Structures, 2015, 48(4), 753.
11 Mei S Q, Liu X D, Wang X J, et al. Journal of Jilin University (Engineering and Technology Edition), DOI,10.13229/j.cnki.jdxbgxb.20230814 (in Chinese).
梅生启, 刘晓东, 王兴举, 等. 吉林大学学报(工学版), DOI,10.13229/j.cnki.jdxbgxb.20230814.
12 Huo X S, Al-Omaishi N, Tadros M K. ACI Materials Journal, 2001, 98(6), 440.
13 Mazloom M, Ramezanianpour A A, Brooks J J. Cement and Concrete Composites, 2004, 26(4), 347.
14 Mazloom M. Cement and Concrete Composites, 2008, 30(4), 316.
15 Xu X H, Hu Z L, Liu J P, et al. Materials Reports, 2023, 41(2), 1. (in Chinese).
徐潇航, 胡张莉, 刘加平, 等. 材料导报, 2023, 41(2), 1.
16 Wang S R, Hu P, Chen S B, et al. Journal of Building Materials, 2023, 26(7), 705. (in Chinese).
汪声瑞, 胡畔, 陈思宝, 等. 建筑材料学报, 2023, 26(7), 705.
17 Ma G, Liu K. Journal of Hunan University, Natural Sciences, 2021, 48(9), 88. (in Chinese).
马高, 刘康. 湖南大学学报:自然科学版, 2021, 48(9), 88.
18 Kumar R, Kumar S, Rai B, et al. Structures, 2024, 66, 106850.
19 Hubler M H, Wendner R, Bažant Z P. ACI Materials Journal, 2015, 112(4). 547.
20 Bal L, Buyle-Bodin F. Neural Computing and Applications, 2014, 25(6), 1359.
21 Zhu J, Wang Y. Construction and Building Materials, 2021, 306, 124868.
22 Li K, Long Y, Wang H, et al. Journal of Materials in Civil Engineering, 2021, 33(8), 04021206.
23 Liang M, Chang Z, Wan Z, et al. Cement and Concrete Composites, 2022, 125, 104295.
24 Hu Y C, Liang M, Xie C R, et al. Bulletin of the Chinese Ceramic Society, 2023, 42(11), 3914. (in Chinese).
胡以婵, 梁铭, 谢灿荣, 等. 硅酸盐通报, 2023, 42(11), 3914.
25 Mai H-V T, Nguyen M H, Trinh S H, et al. Construction and Building Materials, 2023, 369, 130613.
26 Hosseinzadeh M, Dehestani M, Hosseinzadeh A. Journal of Building Engineering, 2023, 76, 107006.
27 Long W, Cheng B, Luo S, et al. Construction and Building Materials, 2023, 393, 132101.
28 Kaltenbach H M. A concise guide to statistics. Springer Berlin, Germany, 2011.
29 Mesfin W M, Kim H K. Engineering Applications of Artificial Intelligence, 2024, 136, 108888.
30 Balasooriya Arachchilage C, Huang G, Fan C, et al. Construction and Building Materials, 2023, 409, 134083.
31 Chen T, Guestrin C. Xgboost, A scalable tree boosting system. In:Proceedings of the 22nd acm sigkdd international conference on knowledge discovery and data mining. US, 2016, pp. 785.
32 Wolpert D H. Stacked generalization. Neural Networks, 1992, 5(2), 241.
33 Li Q, Song Z. Journal of Cleaner Production, 2023, 382, 135279.
34 Gao X, Yang J, Zhu H, et al. Construction and Building Materials, 2023, 371, 130778.
35 Gandomi A H, Sajedi S, Kiani B, et al. Automation in Construction, 2016, 70, 89.
36 Zheng Y, Zhang Y T. China Civil Engineeing Journal, 2013, 46(12), 59 (in Chinese).
郑怡, 张耀庭. 土木工程学报, 2013, 46(12), 59.
37 Jiang Y Z, Luo S R, Huang H, et al. Journal of Fuzhou University( Natural Science Edition), 2025, 53(2), 185. (in Chinese).
江宇舟, 罗素蓉, 黄欢, 等. 福州大学学报(自然科学版), 2025, 53(2), 185.
38 Li K. Study on the time-dependent mechanical property development of fly ash concrete based on machine learning. Ph. D. Thesis, Beijing Jiaotong University, China, 2023(in Chinese).
李凯. 基于机器学习的粉煤灰混凝土力学性能时变规律研究. 博士学位论文, 北京交通大学, 2023.
39 Breiman L. Statistical modeling: Statistical Science, 2001, 16(3), 199.
40 Lundberg S M, Lee S I. Advances in Neural Information Processing Systems, 2017, 30, 1.
41 Bažant Z P, Cusatis G, Cedolin L. Journal of Engineering Mechanics, 2004, 130(6), 691.
42 Tu Y, Yu H, Ma H, et al. Construction and Building Materials, 2022, 352, 128990.
43 Wang X Y, Park K B. Cement and Concrete Research, 2017, 102, 1.
44 Ma G, Xie Y, Long G, et al. Construction and Building Materials, 2022, 342, 127957.
45 De Schutter G, Taerwe L. Materials and Structures, 2000, 33(6), 370.
46 Theiner Y, Drexel M, Neuner M, et al. Strain, 2017, 53(2), e12223.
47 Rossi P, Tailhan J L, Le Maou F. Cement and Concrete Research, 2013, 51, 78.
48 Huang Y, Xie T, Ding Y, et al. Construction and Building Materials, 2021, 286, 122763.
49 Hwang E, Kim G, Koo K, et al. Materials, 2021, 14(17), 5026.