1 Underground Building and Engineering Department, School of Civil Engineering, Tongji University, Shanghai 200092, China
2 Key Laboratory of Disaster Prevention and Mitigation in Civil Engineering, Tongji University, Shanghai 200092, China
3 School of Civil and Transportation Engineering, Shenzhen University, Shenzhen 518061, China
Show less
文章历史+
Received
Published
Revised
2025-04-28
2026-04-28
2025-11-28
Online Date
2026-05-29
PDF (15801K)
摘要
为准确预测运营期隧道的长期沉降,基于数值模拟和机器学习的双向嵌合模型(bidirectional and nested model,BNM),提出一种数据和物理双驱动的盾构隧道长期沉降预测方法。不同于由机器学习方法直接承载物理知识的物理信息神经网络(physics-informed neural network,PINN)方法,BNM方法将机器学习和解析/数值模拟方法嵌合,再由内嵌的解析/数值模拟方法承载物理知识,物理耦合能力更强。以上海地铁2号线南京东路站—陆家嘴站区间的长期沉降问题为背景,构建了基于BNM的盾构隧道长期沉降预测模型,并将其与纯数据驱动的机器学习方法进行了对比。结果表明:单一的机器学习方法不足以应对复杂工况下的隧道变形预测问题,在不同的验证集上表现出较大的性能差异,工程参考价值低。基于BNM的隧道长期沉降预测模型适用于复杂工况下的隧道变形预测问题,在不同的验证集上都有较高的精度,对于复杂影响条件下的长期沉降预测具有良好的适用性。该方法同时具备机器学习和解析/数值计算的优势,适用于复杂工况下的隧道长期沉降预测与工程决策支持。
Abstract
A data and physics driven method based on the bidirectional and nested model (BNM) was proposed to predict the long-term settlement of shield tunnels during operation. Unlike the physics-informed neural network (PINN) method, where physical knowledge was directly embedded into the neural networks, the BNM integrated machine learning with analytical and numerical simulations, allowing physical laws to be internally represented by the nested analysis or numerical submodels, thereby enhancing physical coupling capability. Taking the Nanjing East Road—Lujiazui section of Shanghai Metro Line 2 as a case study, a BNM-based long-term settlement prediction model was developed and compared with purely data-driven approaches. The results indicate that single machine learning models are sensitive to data partitioning and perform inconsistently across validation sets, showing limited engineering applicability. In contrast, the proposed BNM model achieves stable and high prediction accuracy under multiple and complex influencing conditions. Combining the advantages of data-driven method and physics-based method, the BNM is suitable for long-term settlement prediction and engineering decision support for shield tunnels under complex operational environments.
为解决上述难题,提出一种基于双向嵌合模型(bidirectional and nested model, BNM)的长期沉降预测方法。该方法将解析/数值计算模型与机器学习模型双向嵌合,统一训练。相比PINN,双向嵌合模型采用与神经网络结构嵌合的解析/数值方法承载物理机理,实现了数据和物理的深层协同,可用于多机理影响下的复杂工程问题。此外,与PINN方法不同,该方法不仅在计算过程中具备物理课解释性,其部分神经网络隐层节点同样具有物理意义。隧道长期沉降预测案例分析发现,相比单一的数据驱动模型,该模型具有极高的泛化性,在与训练集差异较大的验证集中也能表现出极佳的预测效果。此外,该模型除沉降预测值外,还输出了沉降诱变载荷等关键物理参数,可用于机理分析,揭示隧道在长期沉降发展过程中的内在规律。该模型不仅提高了长期沉降预测的可靠性,还为关键物理参数的反演与机理认知提供了新的技术路径,对隧道等基础设施的长期安全评估与运维决策具有重要意义。
HuangDawei, LiBinxin, ShiYufeng, et al. Experimental study on the effect of groundwater level variation on settlement of subway tunnel in sandy stratum[J]. China Railway Science, 2024, 45(1): 122-130.
WangYanning, ZhouHuanzhu, YuJin. Long-term settlement model for immersed tube tunnels under back silting loads during operation and maintenance periods[J]. Chinese Journal of Geotechnical Engineering, 2023, 45(2): 292-300.
[5]
SangB, YanC, WangC, et al. Study on dynamic response and long-term settlement of silty soil around Shanghai metro tunnel[J]. Scientific Reports, 2024, 14(1): 9172.
Editorial DepartmentofChina Journal of Highway and Transport. Review of academic research on China's traffic tunnel engineering research·2022[J]. China Journal of Highway and Transport, 2022, 35(4): 1-40.
HanShengzhang, HeYunqing, LiuYing, et al. Deformation control of shield tunnel with small angle diagonal intersection underneath box tunnel: a case study of Tianjin Metro Line 4 Project[J]. Science Technology and Engineering, 2025, 25(26): 11294-11305.
[10]
WangL, YangS, HuH. Stochastic dynamic response and long-term settlement performance of superstructure-underground tunnel-soil systems subjected to subway-traffic excitation[J]. Buildings, 2023, 13(3): 621.
MuYongjiang, YuanDajun, WangJiang, et al. Calculation model of the long-term settlement in the weak watery stratum considering leakage of shield tunnel[J]. China Civil Engineering Journal, 2020, 53(S1): 7-12.
YangBingming. Research on the law and prediction of long-term settlement of shield tunnel in soft soil stratum[J]. Railway Engineering Journal, 2015, 32(11): 87-92.
ZhuQixuan, ZhouWeixiang, XuKefeng, et al. Laws of long-term operation accumulation settlement of deep drainage pipelines in soft soil areas[J]. Chinese Journal of Geotechnical Engineering, 2024, 46(S2): 231-235.
HuangQiang, WanLing, LiuGanbin, et al. Field measurement and numerical analysis of train-induced vibration effect on metro tunnel in soft ground[J]. Journal of Southeast University (Natural Science Edition), 2021, 51(3): 435-441.
ZhouSijian, ZhangDi, ZhouJian, et al. Deformation control of shield tunnel operation based on tunnel jet system method[J]. Journal of Zhejiang University (Engineering Science), 2024, 58(7): 1427-1435.
ZhangWengang, TangXuecheng, LiuHanlong, et al. Calculation method of long-term settlement and optimal tunnel depth of shield tunnel in under-consolidated soft soil under static and dynamic load[J]. Tunnel Construction, 2024, 44(9): 1829-1840.
[23]
WuH, ShenS, ChenR, et al. Three-dimensional numerical modelling on localised leakage in segmental lining of shield tunnels[J]. Computers and Geotechnics, 2020, 122: 103549.
LeeH, KimS, KangB, et al. Long-term settlement prediction of ground reinforcement foundation using a deep cement mixing method in reclaimed land[J]. Buildings, 2022, 12(8): 1279.
ZhangJichao, ZengChuigang, CuiLejian, et al. Ground settlement prediction of shallow-buried super-large-diameter shield tunnel based on Loganathan & Poulos and Clough modified formulas[J]. Tunnel Construction, 2022, 42(S1): 274-280.
[28]
ChenX, YangJ, HeG, et al. Development of an LSTM-based model for predicting the long-term settlement of land reclamation and a GUI-based tool[J]. Acta Geotechnica, 2023, 53: 1-14.
[29]
ShenY, ZhangD, WangR, et al. SBD-K-medoids-based long-term settlement analysis of shield tunnel[J]. Transportation Geotechnics, 2023, 42: 101053.
LuDechun, XuBing, KongFanchao, et al. Intelligent prediction method for settlement curves of shield tunnel based on machine learning algorithms[J]. Journal of Beijing University of Technology, 2024, 50(11): 1285-1300.
LiLuobin, GongXiaonan, GanXiaolu, et al. Prediction of maximum ground settlement induced by shield tunneling based on recurrent neural network[J]. China Civil Engineering Journal, 2020, 53(S1): 13-19.
ZhangYanqin, DiHonggui, XuYonggang. Machine learning based prediction model for long-term settlement in a metro-shield tunnel[J]. Urban Rapid Rail Transit, 2022, 35(3): 89-94.
ChenCheng, ShiPeixin, JiaPengjiao, et al. Correlation analysis of shield driving parameters and structural deformation prediction based on MK-LSTM algorithm[J]. Journal of Jilin University(Engineering and Technology Edition), 2024, 54(6): 1624-1633.
LinZhi, WuYifei, YangYing, et al. Intelligent classification prediction model of tunnel surrounding rock considering drilling parameters[J]. Science Technology and Engineering, 2025, 25(15): 6510-6519.
PanQiujing, WuHongtao, ZhangZilong, et al. Prediction of tunneling-induced ground surface settlement within composite strata using multi-physics-informed neural network[J]. Rock and Soil Mechanics, 2024, 45(2): 539-551.
ZhangDongming, ZhouYelu, HuangHongwei, et al. A physics and information dual-driven intelligent diagnosis method for longitudinal mechanical behavior of long-distance shield tunnels[J]. Rock and Soil Mechanics, 2023, 44(10): 2997-3010.
[44]
ZhangM, GuoT, ZhangG, et al. Physics-informed deep learning for structural vibration identification and its application on a benchmark structure[J]. Philosophical Transactions of the Royal Society A, 2024, 382: 2264.
XieSenlin, HuAnfeng, XiaoZhirong, et al. Improved PINNs algorithm and its application in rheological consolidation analysis of soft Soil around tunnels[J]. China Journal of Highway and Transport, 2025, 38(8): 356-369.
[49]
GuoX Y, ZhangD M, ChenX S. Bidirectional and nested model for numerical simulation and machine learning: a case study of long-term settlement prediction of tunnel[J]. IOP Conference Series: Earth and Environmental Science, 2024, 1333(1):012055.
[50]
KarimS A A. Theoretical, modelling and numerical simulations toward industry 4.0[M]. Berlin: Springer, 2021.
YuHongdan, ChenWeizhong, LuChen, et al. Experimental and theoretical study of the time-dependent deformation characteristics of clayey rock[J]. Rock and Soil Mechanics, 2022, 43(2): 317-326.
[53]
ShenS, WuH, CuiY, et al. Long-term settlement behaviour of metro tunnels in the soft deposits of Shanghai[J]. Tunnelling and Underground Space Technology, 2014, 40: 309-323.
FanHu, ZhuangYan, SongXiangwei. Study on mechanical analysis model and settlement law of overburden layer of shallow buried highway tunnel arch during operation period[J]. Journal of Central South University (Natural Science Edition), 2024, 55(1): 230-241.
WangXueqing, LiuShuang, LiQiuyan, et al. Classification and discrimination of surrounding rock of tunnel based on SVM of K-fold cross validation[J]. Mining and Metallurgical Engineering, 2021, 41(6): 126-128.
[60]
WongT, YehP. Reliable accuracy estimates from K-fold cross validation[J]. IEEE Transactions on Knowledge and Data Engineering, 2019, 32(8): 1586-1594.
HeJiazhi, FengXianda, LiuTianqi. Sand liquefaction prediction by using CatBoost algorithm combined with Optuna framework[J]. Journal of University of Jinan (Science and Technology), 2024, 38(4): 496-502.