|Table of Contents|

[1] QIN Wenbo, LUO Hanbin, LI Yanjin, YU Qunzhou, et al. Intelligent prediction of long-term tunnel service performance in complex strata via a deep surrogate model [J]. Journal of Southeast University (English Edition), 2026, 42 (2): 156-164. [doi:10.3969/j.issn.1003-7985.2026.02.002]
Copy

Intelligent prediction of long-term tunnel service performance in complex strata via a deep surrogate model()
基于深度代理模型的复杂地层隧道长期服役性能智能预测

Journal of Southeast University (English Edition)[ISSN:1003-7985/CN:32-1325/N]

Volumn:
42
Issue:
2026 2
Page:
156-164
Research Field:
Publishing date:
2026-05-29

Info

Title:
Intelligent prediction of long-term tunnel service performance in complex strata via a deep surrogate model
基于深度代理模型的复杂地层隧道长期服役性能智能预测
Author(s):
QIN Wenbo1,2, LUO Hanbin1,2, LI Yanjin3, YU Qunzhou1,2, ZHOU Cheng1,2
1.School of Civil and Hydraulic Engineering, Huazhong University of Science and Technology, Wuhan 430074, China
2.National Center of Technology Innovation for Digital Construction, Wuhan 430074, China
3.China Railway Siyuan Survey and Design Group Co., Ltd., Wuhan 430063, China
覃文波1,2, 骆汉宾1,2, 李彦锦3, 余群舟1,2, 周诚1,2
1.华中科技大学土木与水利工程学院, 武汉 430074
2.国家数字建造技术创新中心, 武汉 430074
3.中铁第四勘察设计院集团有限公司, 武汉 430063
Keywords:
tunnel service performance;  tunnel convergence;  deep learning;  surrogate model;  intelligent prediction;  complex strata
隧道服役性能;  隧道收敛;  深度学习;  代理模型;  智能预测;  复杂地层
PACS:
TU17
DOI:
10.3969/j.issn.1003-7985.2026.02.002
Abstract:
Tunnels often traverse variable and complex strata, thereby posing significant challenges in analyzing long-term tunnel service performance. This study proposes the tunnel service performance deep surrogate model (TSP-DSM), an intelligent prediction framework designed for predicting long-term tunnel service performance under complex geological conditions. The TSP-DSM framework employs DenseNet as a deep surrogate model, extracting multilevel feature representations from tunnel geological cross-sectional diagrams and horizontal-convergence monitoring data, and adaptively learning the potential nonlinear mapping relationship between geological conditions and service performance. To increase the prediction accuracy, the model’s hyperparameters are optimized via random grid search. Experimental results demonstrate that the TSP-DSM achieves a training accuracy of 91.29%, outperforming GoogleNet by 12.79% and ResNet by 5.72%; the prediction performance achieved using grayscale images is comparable to that attained using RGB images. On test samples, which were sourced from complex strata, the TSP-DSM achieves a prediction accuracy of 78.19%, demonstrating strong generalization across various strata. A tunnel service-performance visualization map, constructed based on the intelligent prediction results, provides an intuitive basis for tunnel condition assessment and risk prediction.
复杂多变的地层环境给隧道长期服役性能分析带来了显著挑战。为此,本研究提出一种名为隧道服役性能深度代理模型(TSP-DSM)的智能预测框架,用于在复杂地质条件下预测隧道长期服役性能。TSP-DSM采用DenseNet作为深度代理模型,能够从隧道地质剖面图与水平收敛监测数据中提取多层特征表示,自适应学习地质条件与服役性能之间潜在的非线性映射关系。为提升预测精度,模型超参数采用随机网格搜索进行优化。实验结果表明:TSP-DSM在训练集上的预测准确率达到91.29%,较GoogleNet与ResNet分别提升12.79%和5.72%;灰度图像与RGB图像在预测精度上表现相近。TSP-DSM在测试集中预测准确率为78.19%,表现出在复杂地层条件下较强的泛化能力。基于智能预测结果构建的隧道服役性能可视化分布图,能够为隧道状态评估与风险预判提供直观依据。

References:

[1]ZHANG W G, SOMERVILLE I, PANEIRO G, et al. Design and construction of tunnels and tunnelling: Understanding the importance of geological conditions, landslide susceptibility and risk assessment[J]. Geological Journal, 2024, 59(9): 2365-2370.
[2]FU K, QIU D H, XUE Y G, et al. TBM tunneling strata automatic identification and working conditions decision support[J]. Automation in Construction, 2024, 163: 105425.
[3]XU Z W, ZHANG C, ZHOU S H, et al. GFNS: An OpenGL-based tool for shield tunneling simulation in 3D complex stratum[J]. Computers and Geotechnics, 2024, 167: 106111.
[4]WANG K, XU S S, ZHONG Y J, et al. Deformation failure characteristics of weathered sandstone strata tunnel: A case study[J]. Engineering Failure Analysis, 2021, 127: 105565.
[5]ZHANG S L, GAO J L, LIU C, et al. Model test on the collapse mechanism of subway tunnels in the soil-sand- rock composite strata[J]. Engineering Failure Analysis, 2024, 162: 108356.
[6]YU T S, ZHANG Y X, YAN Z G. A new generation method of tunnel progressive defect status random field (TPDSRF) for subway tunnel structure[J]. Tunnelling and Underground Space Technology, 2023, 141: 105340.
[7]LI Z Z, WANG H, CHANG X Y, et al. Prediction of surrounding rock convergence deformation of high-speed railway tunnel based on combined model[J]. Journal of Southeast University (Natural Science Edition), 2021, 51(5): 851-858. (in Chinese)
[8]ZHANG Z G, MAO M D, ZHANG C P, et al. Analysis on viscoelastic soil consolidation settlement of fractional order induced by tunnel leakage[J]. Journal of Southeast University (Natural Science Edition), 2022, 52(3): 530-537. (in Chinese)
[9]BELLINI MACHADO L, MASSAO FUTAI M. Tunnel performance prediction through degradation inspection and digital twin construction[J]. Tunnelling and Underground Space Technology, 2024, 144: 105544.
[10]SHI C, WANG Y. Development of subsurface geological cross-section from limited site-specific boreholes and prior geological knowledge using iterative convolution XGBoost[J]. Journal of Geotechnical and Geoenvironmental Engineering, 2021, 147(9): 04021082.
[11]GAN B L, ZHANG D M, HUANG Z K, et al. Ontology-driven knowledge graph for decision-making in resilience enhancement of underground structures: Framework and application[J]. Tunnelling and Underground Space Technology, 2025, 163: 106739.
[12]CARTER T G, BARNETT W P. Improving reliability of structural domaining for engineering projects[J]. Rock Mechanics and Rock Engineering, 2022, 55(5): 2523-2549.
[13]GUO Z Z, QIU D H, YU Y H, et al. Analysis and prediction of nonuniform deformation in composite strata during tunnel excavation[J]. Computers and Geotechnics, 2023, 157: 105338.
[14]QIU J T, ZHOU X J, SHEN Y S, et al. Failure mechanism of the deep-buried metro tunnel in mixed strata: Physical model test and numerical investigation[J]. Tunnelling and Underground Space Technology, 2023, 139: 105224.
[15]LIU Z H, FANG Q, SHEN Y, et al. Two-stage surrogate modeling strategy for predicting foundation pit excavation-induced strata and tunnel deformation[J]. Tunnelling and Underground Space Technology, 2024, 151: 105845.
[16]WANG F, ZENG J P, FANG Y B, et al. Research on the mechanism of deformation of tunnel lining structure exposed to fire based on transient thermal strain[J]. Journal of Southeast University (Natural Science Edition), 2025, 55(3): 856-863. (in Chinese)
[17]POTTS D M. Numerical analysis: A virtual dream or practical reality？[J]. Géotechnique, 2003, 53(6): 535- 573.
[18]ZHANG J Z, PHOON K K, ZHANG D M, et al. Deep learning-based evaluation of factor of safety with confidence interval for tunnel deformation in spatially variable soil[J]. Journal of Rock Mechanics and Geotechnical Engineering, 2021, 13(6): 1358-1367.
[19]ZHANG J Z, PHOON K K, ZHANG D M, et al. Novel approach to estimate vertical scale of fluctuation based on CPT data using convolutional neural networks[J]. Engineering Geology, 2021, 294: 106342.
[20]ZHANG H L, LUO F, GENG W J, et al. An efficient method for reliability analysis of high-speed railway tunnel convergence in spatially variable soil based on a deep convolutional neural network[J]. International Journal of Geomechanics, 2023, 23(11): 04023210.
[21]QIN W B, CHEN E J, WANG F, et al. Data-driven models in reliability analysis for tunnel structure: A systematic review[J]. Tunnelling and Underground Space Technology, 2024, 152: 105928.
[22]ALIZADEH R, ALLEN J K, MISTREE F. Managing computational complexity using surrogate models: A critical review[J]. Research in Engineering Design, 2020, 31(3): 275-298.
[23]HUANG G, LIU Z, VAN DER MAATEN L, et al. Densely connected convolutional networks[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Honolulu, HI, USA, 2017: 2261- 2269.
[24]MAO A Q, MOHRI M, ZHONG Y T. Cross-entropy loss functions: Theoretical analysis and applications[C]// International Conference on Machine Learning. Honolulu, HI, USA, 2023, 202: 23803-23828.
[25]LOSHCHILOV I, HUTTER F. SGDR: Stochastic gradient descent with warm restarts[EB/OL]. (2016-08-13)[2025-08-10]. https: //doi. org/10. 48550/arXiv. 1608. 03983.
[26]ZHOU C, QIN W B, LUO H B, et al. Digital twin for smart metro service platform: Evaluating long-term tunnel structural performance[J]. Automation in Construction, 2024, 167: 105713.
[27]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, 2016: 770-778.
[28]SZEGEDY C, LIU W, JIA Y Q, et al. Going deeper with convolutions[C]//2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Boston, MA, USA, 2015: 1-9.

Memo

Memo:
Received: 2025-08-14; Revised: 2025-11-13.
Biographies: QIN Wenbo (1993—), male, Ph.D.candidate; ZHOU Cheng (corresponding author), male, doctor, professor, chengzhou@hust.edu.cn.
Foundation item: The National Natural Science Foundation of China (No.52192664).
Citation: QIN Wenbo, LUO Hanbin, LI Yanjin, et al. Intelligent prediction of long-term tunnel service performance in complex strata via a deep surrogate model[J]. Journal of Southeast University (English Edition), 2026, 42(2): 156-164. DOI: 10. 3969/j. issn. 1003-7985. 2026. 02. 002.
Last Update: 2026-06-20