基于神经网络的硬化水泥浆体等效强度预测

宋敏; 杨予舒; 祝华杰; 王志勇

doi:10.11858/gywlxb.20251024

基于神经网络的硬化水泥浆体等效强度预测

doi: 10.11858/gywlxb.20251024

宋敏^{1, 2,},
杨予舒^{2, 3},
祝华杰¹,
王志勇^{2, 3,*, ,}

1.
山西省交通科技研发有限公司桥梁与隧道工程研究院, 山西太原　030000
2.
太原理工大学航空航天学院, 山西太原　030024
3.
太原理工大学应用力学研究所, 山西太原　030024

基金项目: 国家自然科学基金（12272257）；山西省基础研究计划青年项目（202303021222387，202403021222519）

详细信息

作者简介:
宋　敏（1991－），男，博士，工程师，主要从事桥梁工程结构优化设计与应用研究. E-mail：songmin595@163.com

通讯作者:
王志勇（1982－），男，博士，教授，主要从事冲击动力学与断裂力学研究. E-mail：wangzhiyong@tyut.edu.cn

中图分类号: O346; O521.9
计量
- 文章访问数: 86
- HTML全文浏览量: 20
- PDF下载量: 7
出版历程
- 收稿日期: 2025-02-14
- 修回日期: 2025-04-15
- 网络出版日期: 2025-04-20

Prediction of Equivalent Strength of Hydrated Cement Paste Based on Neural Networks

SONG Min^{1, 2
,},
YANG Yushu^{2, 3},
ZHU Huajie¹,
WANG Zhiyong^{2, 3,*
, ,}

1.
Bridge and Tunnel Engineering Research Institute, Shanxi Transportation Technology R&D Co., Ltd., Taiyuan 030000, Shanxi, China
2.
College of Aeronautics and Astronautics, Taiyuan University of Technology, Taiyuan 030024, Shanxi, China
3.
Institute of Applied Mechanics, Taiyuan University of Technology, Taiyuan 030024, Shanxi, China

摘要

摘要: 为实现材料性能优化并保障工程结构安全，需要研究具有复杂结构的水泥水化模型的力学性能。为此，考察了水灰比及各相体积分数对水泥浆体等效力学性能的影响，提出了一种基于数据驱动的模型，用于预测水化水泥结构的力学性能。通过HYMOSTRUC 3D软件生成波特兰硬化水泥浆体三维结构切片，基于Python编写的批处理程序，将切片批量转换为ABAQUS模型。通过拉伸仿真模拟，得到结构的等效弹性性能和等效强度，运用数据驱动方法建立反向传播预测模型。模型的超参数优化采用K折交叉验证方法，以提高模型的泛化能力。最终训练得到的神经网络模型能够准确预测水泥水化结构的力学性能，显著降低传统分析方法在材料微观尺度研究中的复杂性。研究结果为水泥基材料的性能预测提供了一种高效且可靠的解决方案。
- 硬化水泥浆体 /
- 有限元方法 /
- 神经网络 /
- 数据驱动方法 /
- 单轴拉伸
Abstract: To optimize material performance and ensure the safety of engineering structures, it is essential to investigate the mechanical properties of hydration models of cement, which possess complex structures. This study aims to investigate the influence of the water-to-cement ratio and phase volume fractions on the equivalent mechanical properties of cement paste, particularly focusing on how these parameters influence the behavior of the material. A data-driven model is proposed to predict the mechanical performance of hydrated cement structures. Three-dimensional structural slices of Portland hydrated cement paste were created by utilizing the HYMOSTRUC 3D software. Subsequently, an automated batch-processing script, which was coded in Python, was applied to transform these slices into ABAQUS models. Tensile simulations were performed to determine the equivalent elastic modulus and equivalent strength of the structures. Based on the simulation results, a backpropagation prediction model was developed using a data-driven approach. Hyperparameter optimization of the model was performed using K-fold cross-validation to improve its generalization capability. Consequently, the trained neural network model demonstrates high accuracy in predicting the mechanical properties of hydrated cement structures. This approach not only ensures reliable predictions but also significantly reduces the complexity associated with traditional microscale material analysis methods. Overall, this study offers an efficient and robust solution for performance prediction of cement-based materials.
- hydrated cement paste /
- finite element method /
- neural network /
- data-driven approach /
- uniaxial tension

HTML全文

图 1 边界条件示意图

Figure 1. Schematic diagram of boundary condition

下载: 全尺寸图片幻灯片

图 2 不同水灰比条件下部分试件的应力-应变曲线

Figure 2. Stress-strain curves of specimens at different water-to-cement ratios

下载: 全尺寸图片幻灯片

图 3 不同水灰比试件单元的损伤分布

Figure 3. Damage distribution in specimen units at different water-to-cement ratios

下载: 全尺寸图片幻灯片

图 4 神经网络示意图

Figure 4. Schematic diagram of neural network

下载: 全尺寸图片幻灯片

图 5 4折交叉验证示意图

Figure 5. 4-fold cross-validation diagram

下载: 全尺寸图片幻灯片

图 6 模型训练进程

Figure 6. Model training process

下载: 全尺寸图片幻灯片

图 7 测试集预测回归曲线

Figure 7. Prediction regression plots for the test set

下载: 全尺寸图片幻灯片

图 8 SHAP蜂群图

Figure 8. SHAP swarm plots

下载: 全尺寸图片幻灯片

图 9 SHAP平均值条形图

Figure 9. Bar plots of the average of SHAP

下载: 全尺寸图片幻灯片

表 1 水泥的化学组分（质量分数）

Table 1. Each chemical component in cement (mass fraction) %

CaO	SiO₂	Al₂O₃	Fe₂O₃	SO₃	MgO	Others
65.24	21.12	5.34	4.63	1.58	1.25	0.84

下载: 导出CSV

表 2 不同水灰比HCP中各组分的平均体积分数

Table 2. Average volume fraction of each component for HCP with different water-to-cement ratios

r	Volume fraction/%
r	LDP	HDP	Pore	Anhydrous
0.375	49.8439	24.4324	10.9256	14.7981
0.400	48.6137	23.9870	12.9135	14.4858
0.425	48.0042	23.9356	14.7209	13.3585
0.450	46.8344	23.4919	17.1533	12.5203
0.475	45.5721	23.3070	19.8090	11.3119
0.500	44.9101	23.1363	21.1294	10.8241

下载: 导出CSV

表 3 HCP各成分的力学参数

Table 3. Mechanical parameters of each component in HCP

Component	Young’s modulus/GPa	Tensile strength/MPa	Poisson’s ratio
HDP	31.6	92	0.2
LDP	25.2	66	0.2
Anhydrous	99.2	683	0.2
Pore	1×10⁻⁶		0

下载: 导出CSV

参考文献(25)

[1]	WU X G, ZHENG S Y, FENG Z B, et al. Prediction of the frost resistance of high-performance concrete based on RF-REF: a hybrid prediction approach [J]. Construction and Building Materials, 2022, 333: 127132. doi: 10.1016/j.conbuildmat.2022.127132
[2]	GUO F Q, ZHANG Z H, YANG Z J. A continuous hydration model for cement paste with realistic CT image-based particles and simulation of microstructural evolution [J]. Cement and Concrete Research, 2024, 184: 107607. doi: 10.1016/j.cemconres.2024.107607
[3]	NGUYEN-TUAN L, KLEINER F, RÖBLER C, et al. Numerical simulation of model cement hydration using level set based method [J]. Cement and Concrete Research, 2024, 186: 107674. doi: 10.1016/j.cemconres.2024.107674
[4]	BENTZ D P. Three-dimensional computer simulation of Portland cement hydration and microstructure development [J]. Journal of the American Ceramic Society, 1997, 80(1): 3–21. doi: 10.1111/j.1151-2916.1997.tb02785.x
[5]	VAN BREUGEL K. Numerical simulation of hydration and microstructural development in hardening cement-based materials (Ⅰ) theory [J]. Cement and Concrete Research, 1995, 25(2): 319–331. doi: 10.1016/0008-8846(95)00017-8
[6]	曹秀丽, 刘元寿, 张红飞, 等. 水泥水化计算机模拟研究进展 [J]. 材料导报, 2013, 27(1): 157–160. doi: 10.3969/j.issn.1005-023X.2013.01.032 CAO X L, LIU Y S, ZHANG H F, et al. A review on computer simulation of cement hydration models [J]. Materials Review, 2013, 27(1): 157–160. doi: 10.3969/j.issn.1005-023X.2013.01.032
[7]	QIAN Z W. Multiscale modeling of fracture processes in cementitious materials [D]. Delft: Delft University of Technology, 2012.
[8]	YU P, REN Z Y, CHEN Z, et al. A multiscale finite element model for prediction of tensile strength of concrete [J]. Finite Elements in Analysis and Design, 2023, 215: 103877. doi: 10.1016/j.finel.2022.103877
[9]	SHERZER G, GAO P, SCHLANGEN E, et al. Upscaling cement paste microstructure to obtain the fracture, shear, and elastic concrete mechanical LDPM parameters [J]. Materials, 2017, 10(3): 242. doi: 10.3390/ma10030242
[10]	QIAN Z W, SCHLANGEN E, YE G, et al. Modeling framework for fracture in multiscale cement-based material structures [J]. Materials, 2017, 10(6): 587. doi: 10.3390/ma10060587
[11]	KAMALI-BERNARD S, BERNARD F. Effect of tensile cracking on diffusivity of mortar: 3D numerical modelling [J]. Computational Materials Science, 2009, 47(1): 178–185. doi: 10.1016/j.commatsci.2009.07.005
[12]	HLOBIL M, KUMPOVÁ I, HLOBILOVÁ A. Surface area and size distribution of cement particles in hydrating paste as indicators for the conceptualization of a cement paste representative volume element [J]. Cement and Concrete Composites, 2022, 134: 104798. doi: 10.1016/j.cemconcomp.2022.104798
[13]	PUTTBACH C, PRINZ G S, MURRAY C D. Estimation of cement paste stiffness and UHPC elastic modulus through measured phase-property upscaling [J]. Cement, 2024, 17: 100110. doi: 10.1016/j.cement.2024.100110
[14]	LE V T, BUI H H, NGUYEN G D, et al. Meso to macro connections to capture fatigue damage in cemented materials [J]. International Journal of Fatigue, 2023, 176: 107890. doi: 10.1016/j.ijfatigue.2023.107890
[15]	巨凯萱, 赵婷婷, 刘嘉英, 等. 级配对颗粒材料力学特性影响的卷积神经网络分析 [J]. 中国科学: 技术科学, 2025, 55(1): 115–132. doi: 10.1360/SST-2024-0187 JU K X, ZHAO T T, LIU J Y, et al. Convolution neural network analysis of the effect of gradation on the mechanical properties of granular materials [J]. Scientia Sinica Technologica, 2025, 55(1): 115–132. doi: 10.1360/SST-2024-0187
[16]	MURR J, ALAM S Y, GRONDIN F. Micromechanics and microstructure based machine learning approach: unveiling the role of porosity and hydrated phases on the tensile behaviour of cement pastes [J]. Engineering Fracture Mechanics, 2024, 312: 110613. doi: 10.1016/j.engfracmech.2024.110613
[17]	KIM J, LEE D, UBYSZ A. Comparative analysis of cement grade and cement strength as input features for machine learning-based concrete strength prediction [J]. Case Studies in Construction Materials, 2024, 21: e03557. doi: 10.1016/j.cscm.2024.e03557
[18]	中华人民共和国住房和城乡建设部. 混凝土结构设计规范: GB/T 5001—2010 [S]. 北京: 中国建筑工业出版社, 2011. Ministry of Housing and Urban-Rural Development of the People’s Republic of China. Code for design of concrete structures: GB/T 5001—2010 [S]. Beijing: China Architecture & Building Press, 2011.
[19]	ZHANG H Z, ŠAVIJA B, FIGUEIREDO S C, et al. Microscale testing and modelling of cement paste as basis for multi-scale modelling [J]. Materials, 2016, 9(11): 907. doi: 10.3390/ma9110907
[20]	卢振宇, 李文彬, 姚文进, 等. 不同应变率下两种岩石的压缩破碎特征试验研究 [J]. 高压物理学报, 2021, 35(1): 014101. doi: 10.11858/gywlxb.20200605 LU Z Y, LI W B, YAO W J, et al. Experimental study on compression and fracture characteristics of two kinds of rocks under different strain rates [J]. Chinese Journal of High Pressure Physics, 2021, 35(1): 014101. doi: 10.11858/gywlxb.20200605
[21]	马昊, 陈美多, 袁良柱, 等. 中等应变率下纸蜂窝结构的力学性能研究 [J]. 高压物理学报, 2024, 38(4): 044104. doi: 10.11858/gywlxb.20240701 MA H, CHEN M D, YUAN L Z, et al. Study on mechanical properties of paper honeycomb structure at medium strain rates [J]. Chinese Journal of High Pressure Physics, 2024, 38(4): 044104. doi: 10.11858/gywlxb.20240701
[22]	洛绒邓珠, 刘潇如, 杨佳, 等. 不同应变率下高强钢的拉伸行为及力学性能分析 [J]. 高压物理学报, 2024, 38(3): 030104. doi: 10.11858/gywlxb.20240702 LUORONG D Z, LIU X R, YANG J, et al. Tensile behavior and mechanical performance analysis of high-strength steels at varying strain rates [J]. Chinese Journal of High Pressure Physics, 2024, 38(3): 030104. doi: 10.11858/gywlxb.20240702
[23]	张春春, 王艳超, 黄争鸣. 横观各向同性基体复合材料的等效弹性常数 [J]. 应用数学和力学, 2018, 39(7): 750–765. doi: 10.21656/1000-0887.380267 ZHANG C C, WANG Y C, HUANG Z M. Effective elastic properties of transversely isotropic matrix based composites [J]. Applied Mathematics and Mechanics, 2018, 39(7): 750–765. doi: 10.21656/1000-0887.380267
[24]	朱俊, 桂林, 李果, 等. 基于结构参数的平纹机织复合材料等效弹性性能预测 [J]. 复合材料学报, 2023, 40(2): 804–813. doi: 10.13801/j.cnki.fhclxb.20220424.004 ZHU J, GUI L, LI G, et al. Prediction of the effective elastic properties for plain woven fabric composite based on the structural parameters [J]. Acta Materiae Compositae Sinica, 2023, 40(2): 804–813. doi: 10.13801/j.cnki.fhclxb.20220424.004
[25]	顾春苗, 刘冠琳, 周风华, 等. 人造结石的静动态巴西劈裂试验研究 [J]. 高压物理学报, 2024, 38(5): 054105. doi: 10.11858/gywlxb.20240738 GU C M, LIU G L, ZHOU F H, et al. Study on static and dynamic Brazilian splitting test of artificial stones [J]. Chinese Journal of High Pressure Physics, 2024, 38(5): 054105. doi: 10.11858/gywlxb.20240738