硅酸盐学报

基于物理驱动与数据修正协同的复合胶凝材料水化热预测模型

PDF下载

元强 ¹^,² , 夏瑞 ¹^,² , 刘易 ¹^,² , 马嘉璐 ¹^,² , 伍军 ³

硅酸盐学报 | 第二十八届中国科协年会智能建造关键无机非金属材料–AI驱动水泥基材料创新专题-研究论文 2026,54(10): 3328-3344

收起

硅酸盐学报 |第二十八届中国科协年会智能建造关键无机非金属材料–AI驱动水泥基材料创新专题-研究论文 2026 , 54 (10) : 3328 -3344

基于物理驱动与数据修正协同的复合胶凝材料水化热预测模型

全屏

元强¹^,², 夏瑞¹^,², 刘易¹^,², 马嘉璐¹^,², 伍军³

作者信息

^1.中南大学土木工程学院，长沙　410075

^2.高速铁路建造技术国家工程研究中心，长沙　410075

^3.中国中铁股份有限公司，北京　100039

作者简介:

元强（1981-），男，博士，教授。

Physics-Driven and Data-Correction Synergistic Model for Hydration Heat Prediction of Composite Cementitious Materials

Qiang YUAN¹^,², Rui XIA¹^,², Yi LIU¹^,², Jialu MA¹^,², Jun WU³

Affiliations

^1.School of Civil Engineering, Central South University, Changsha 410075, China

^2.National Engineering Research Center for High-speed Railway Construction Technology, Changsha 410075, China

^3.China Railway Group Limited, Beijing 100039, China

YUAN Qiang (1981-), male, Ph.D., Professor. E-mail: yuanqiang@csu.edu.cn

出版时间: 2026-07-29 doi: 10.14062/j.issn.0454-5648.20260307

文章导航

摘要

收起

复合胶凝材料是现代混凝土的基本特征，其水化热的精准预测对混凝土抗裂及强度发展调控具有重要意义。本工作提出一种物理驱动与数据修正协同的水化热预测模型，将S-F方程、Krstulović-Dabić模型等水化动力学方程作为物理先验嵌入神经网络前向传播路径，通过物理参数预测分支与残差修正分支协同实现“物理驱动+数据修正”的水化热曲线端到端建模。结果表明，所提出的SFQmaxDecompNet模型在预测精度、泛化能力和稳健性方面均优于纯数据驱动模型，将水化动力学方程作为归纳偏置嵌入神经网络，是实现水化热预测精度与可解释性协同提升的有效途径。

关键词

复合胶凝材料 / 水化热 / 物理嵌入神经网络 / 水化动力学方程 / 残差修正 / 可解释性

Abstract

收起

Introduction

Hydration heat prediction is important for mixture design, thermal control, and early-age cracking risk assessment for cement-based materials. The increasing use of supplementary cementitious materials (SCMs) and chemical admixtures in low-carbon concrete has made hydration behavior more complex, while conventional empirical or simplified analytical models have difficulty describing multicomponent heat release. Therefore, this study develops a general model for predicting cumulative hydration heat directly from mixture composition.

Methods

A database of 219 cumulative hydration heat curves and 13 048 data points was compiled from 33 published isothermal calorimetry studies, covering ordinary Portland cement and blends with fly ash, ground granulated blast-furnace slag, and silica fume. The input features included water-to-binder ratio, component fractions, oxide compositions, specific surface area, and loss on ignition. Hydration kinetic equations, including the Schindler-Folliard, Krstulovic-Dabic, and Knudsen equations, were embedded into neural networks as physical priors. The framework consisted of a physical-parameter prediction branch, a physical-equation layer, and a residual-correction branch, enabling end-to-end physics-driven prediction with data correction. Ten model variants were trained and compared under identical experimental settings.

Results and discussion

SFQmaxDecompNet achieved the best overall performance, with a test R² of 0.962, a generalization gap of 0.018, and 94% of test curves reaching R²≥0.9. Its component-wise decomposition of Q_max better represented the different heat-release potentials of cement, FA, GGBS, and SF. KD-B ranked second (R² = 0.937), benefiting from its two-peak representation of early cement hydration and delayed SCM reactions. In contrast, KD-C had seven physical parameters but reached only R² = 0.868, showing that model performance depended more on the consistency between physical assumptions and data characteristics than on the number of parameters alone. PureNN reached R² = 0.874, lower than several physics-embedded models.

Data-efficiency tests showed that physical priors reduced the dependence on training data. When only 30% of the training curves (45 curves) were used, SFQmaxDecompNet still achieved R²=0.571, whereas PureNN degraded nearly completely. The physics branch dominated under limited data, while the residual branch provided local corrections when data were sufficient. Interpretability analyses further showed that Q_max represented the total heat potential, τ reflected the hydration time scale, and β described the concentration of heat release. Permutation importance and ALE analyses identified cement content as the dominant output feature, followed by FA, GGBS, FA-CaO, W/B, and SF. The temporal ALE trends of Q and dQ/dt were consistent with the delayed reactions of FA and GGBS and the early promoting effect of SF.

Conclusions

The proposed physics-driven and data-correction synergistic neural network provides accurate and interpretable hydration heat prediction of composite cementitious materials. Embedding hydration kinetic equations as inductive biases improves generalization, suppresses non-physical oscillations, and enhances robustness under small-sample conditions. This framework provides a useful route for mixture design, thermal analysis, and early-age crack-risk control in low-carbon cement-based materials.

Key words

composite cementitious materials / hydration heat / physics-embedded neural network / hydration kinetics / residual correction / interpretability

引用本文

元强, 夏瑞, 刘易, 马嘉璐, 伍军. 基于物理驱动与数据修正协同的复合胶凝材料水化热预测模型. 硅酸盐学报, 2026 , 54 (10) : 3328 -3344 . DOI: 10.14062/j.issn.0454-5648.20260307

Qiang YUAN, Rui XIA, Yi LIU, Jialu MA, Jun WU. Physics-Driven and Data-Correction Synergistic Model for Hydration Heat Prediction of Composite Cementitious Materials[J]. Journal of the Chinese Ceramic Society, 2026 , 54 (10) : 3328 -3344 . DOI: 10.14062/j.issn.0454-5648.20260307

基金

收起

国家自然科学基金面上项目(52378278)
中国中铁股份有限公司科技研究开发计划项目(2023-重大-08)

参考文献

收起

文献

收起

参考文献引证文献

排序方式：

[1]

MA X S, HU H B, LUO Y, et al. A carbon footprint assessment for usage of recycled aggregate and supplementary cementitious materials for sustainable concrete: A life-cycle perspective in China[J]. J Clean Prod, 2025, 490: 144772.

[2]

SCRIVENER K L, MATSCHEI T, GEORGET F, et al. Advances in hydration and thermodynamics of cementitious systems[J]. Cem Concr Res, 2023, 174: 107332.

[3]

JOHN E, LOTHENBACH B. Cement hydration mechanisms through time-a review[J]. J Mater Sci, 2023, 58(24): 9805-9833.

[4]

BULLARD J W, JENNINGS H M, LIVINGSTON R A, et al. Mechanisms of cement hydration[J]. Cem Concr Res, 2011, 41(12): 1208-1223.

[5]

YANG J L, YUAN Q, ZHANG K, et al. Early-age thermal cracking behavior of high-speed railway bridge piers in plateau regions: Formwork removal recommendations[J]. J Cent South Univ, 2025, 32(10): 4055-4072.

[6]

BIERNACKI J J, BULLARD J W, SANT G, et al. Cements in the 21^stcentury: Challenges, perspectives, and opportunities[J]. J Am Ceram Soc, 2017, 100(7): 2746-2773.

[7]

THOMAS J J, BIERNACKI J J, BULLARD J W, et al. Modeling and simulation of cement hydration kinetics and microstructure development[J]. Cem Concr Res, 2011, 41(12): 1257-1278.

[8]

李琴飞, 顾彬彬, 侯鹏坤, 等. 超细矿物掺合料在水泥混凝土中的应用现状[J]. 硅酸盐学报, 2025, 53(8): 2374-2387.

LI Qinfei, GU Binbin, HOU Pengkun, et al. J Chin Ceram Soc, 2025, 53(8): 2374-2387.

[9]

蔡景顺, 李华, 舒鑫, 等. 提升结构混凝土耐久性的外加剂研究进展[J]. 硅酸盐学报, 2025, 53(3): 553-573.

CAI Jingshun, LI Hua, SHU Xin, et al. J Chin Ceram Soc, 2025, 53(3): 553-573.

[10]

YUAN Q, ZHOU D J, LI B Y, et al. Effect of mineral admixtures on the structural build-up of cement paste[J]. Constr Build Mater, 2018, 160: 117-126.

[11]

LOTHENBACH B, SCRIVENER K, HOOTON R D. Supplementary cementitious materials[J]. Cem Concr Res, 2011, 41(12): 1244-1256.

[12]

IEA/WBCSD. Technology roadmap: low-carbon transition in the cement industry[R]. Paris: International Energy Agency, 2018: 66.

[13]

BERODIER E, SCRIVENER K. Understanding the filler effect on the nucleation and growth of C-S-H[J]. J Am Ceram Soc, 2014, 97(12): 3764-3773.

[14]

JUENGER M C G, SNELLINGS R, BERNAL S A. Supplementary cementitious materials: New sources, characterization, and performance insights[J]. Cem Concr Res, 2019, 122: 257-273.

[15]

SCHINDLER A K, FOLLIARD K J. Heat of hydration models for cementitious materials[J]. Mater J, 102(1): 24-33.

[16]

KRSTULOVIĆ R, DABIĆ P. A conceptual model of the cement hydration process[J]. Cem Concr Res, 2000, 30(5): 693-698.

[17]

KNUDSEN T. The dispersion model for hydration of Portland cement I. General concepts[J]. Cem Concr Res, 1984, 14(5): 622-630.

[18]

姜占, 滕乐, 童昕阳, 等. 人工智能驱动可持续混凝土: 方法、机遇、挑战[J]. 硅酸盐学报, 2026, 54(3): 1016-1031.

JIANG Zhan, TENG Le, TONG Xinyang, et al. J Chin Ceram Soc, 2026, 54(3): 1016-1031.

[19]

张文生, 曹傅荔, 郅晓, 等. 机器学习方法用于水泥基材料的研究进展[J]. 硅酸盐学报, 2024, 52(11): 3617-3630.

ZHANG Wensheng, CAO Fuli, ZHI Xiao, et al. J Chin Ceram Soc, 2024, 52(11): 3617-3630.

[20]

BEN CHAABENE W, FLAH M, NEHDI M L. Machine learning prediction of mechanical properties of concrete: Critical review[J]. Constr Build Mater, 2020, 260: 119889.

[21]

元强, 夏瑞, 刘易, 等. 基于多目标寻优的低碳混凝土智能设计[J]. 硅酸盐学报, 2026, 54(3): 878-893.

YUAN Qiang, XIA Rui, LIU Yi, et al. J Chin Ceram Soc, 2026, 54(3): 878-893.

[22]

LIU Y, ZHANG J L, ZHANG S H, et al. Machine learning-guided optimization of coarse aggregate mix proportion based on CO₂ intensity index[J]. J CO₂ Util, 2024, 85: 102862.

[23]

HAN T H, BHAT R, PONDURU S A, et al. Deep learning to predict the hydration and performance of fly ash-containing cementitious binders[J]. Cem Concr Res, 2023, 165: 107093.

[24]

WILLARD J, JIA X W, XU S M, et al. Integrating scientific knowledge with machine learning for engineering and environmental systems[J]. ACM Comput Surv, 2023, 55(4): 1-37.

[25]

VON RUEDEN L, MAYER S, BECKH K, et al. Informed machine learning-A taxonomy and survey of integrating prior knowledge into learning systems[J]. IEEE Trans Knowl Data Eng, 2023, 35(1): 614-633.

[26]

KARPATNE A, ATLURI G, FAGHMOUS J H, et al. Theory-guided data science: A new paradigm for scientific discovery from data[J]. IEEE Trans Knowl Data Eng, 2017, 29(10): 2318-2331.

[27]

REICHSTEIN M, CAMPS-VALLS G, STEVENS B, et al. Deep learning and process understanding for data-driven Earth system science[J]. Nature, 2019, 566(7743): 195-204.

[28]

杨正伟, 刘子图, 张王颖, 等. 材料领域知识嵌入的分而治之机器学习[J]. 硅酸盐学报, 2026, 54(1): 14-24.

YANG Zhengwei, LIU Zitu, ZHANG Wangying, et al. J Chin Ceram Soc, 2026, 54(1): 14-24.

[29]

KARNIADAKIS G E, KEVREKIDIS I G, LU L, et al. Physics-informed machine learning[J]. Nat Rev Phys, 2021, 3(6): 422-440.

[30]

HE K M, ZHANG X Y, REN S Q, et al. Deep residual learning for image recognition[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Las Vegas, NV, USA. IEEE, 2016: 770-778.

[31]

LI S R, FENG X L. Dynamic weight strategy of physics-informed neural networks for the 2D navier-stokes equations[J]. Entropy, 2022, 24(9): 1254.

[32]

SKIBSTED J, SNELLINGS R. Reactivity of supplementary cementitious materials (SCMs) in cement blends[J]. Cem Concr Res, 2019, 124: 105799.

[33]

阎培渝, 徐志全. 水胶比和组成对补偿收缩胶凝材料的水化反应的影响[J]. 硅酸盐学报, 2003, 31(8): 790-794.

YAN Peiyu, XU Zhiquan. J Chin Ceram Soc, 2003, 31(8): 790-794.

[34]

APLEY D W, ZHU J Y. Visualizing the effects of predictor variables in black box supervised learning models[J]. J R Stat Soc Ser B Stat Methodol, 2020, 82(4): 1059-1086.

2026年第54卷第10期

PDF下载

引用本文

BibTeX

文章信息

doi: 10.14062/j.issn.0454-5648.20260307

接收时间：2026-05-09
首发时间：2026-10-22
出版时间：2026-07-29

补充材料

相关文章

文章信息

作者

出版历史

收稿日期：2026-05-09
修回日期：2026-06-04

基金

国家自然科学基金面上项目(52378278)

中国中铁股份有限公司科技研究开发计划项目(2023-重大-08)

作者信息

^1.中南大学土木工程学院，长沙　410075

^2.高速铁路建造技术国家工程研究中心，长沙　410075

^3.中国中铁股份有限公司，北京　100039

参考文献

分享链接

https://castjournals.cast.org.cn/joweb/gsyxb/CN/10.14062/j.issn.0454-5648.20260307

分享至

全文二维码

扫描看全文

引用本文

BibTeX

本文的引用情况

2种不同金属材料的力学参数

科 Family	属数 Number of genus	种数 Number of species	占总种数比例 Percentage of total species (%)	属 Genus	种数 Number of species	占总种数比例 Percentage of total species (%)
鹅膏菌科Amanitaceae	2	11	5.26	鹅膏菌属 Amanita	10	4.78
小菇科 Mycenaceae	2	12	5.74	丝盖伞属 Inocybe	5	2.39
多孔菌科 Polyporaceae	8	14	6.70	蜡蘑属 Laccaria	5	2.39
红菇科 Russulaceae	3	23	11.00	小皮伞属 Marasmius	6	2.87
				小菇属 Mycena	11	5.26
				光柄菇属 Pluteus	5	2.39
				红菇属 Russula	17	8.13
				栓菌属 Trametes	5	2.39

关闭全屏

BibTeX
EndNote
RefWorks
TxT