Physics-guided gradient boosting regression tree for predicting tunnel deformation induced by foundation pit excavation

Abstract

The upward movement of an existing tunnel caused by excavation of a nearby foundation pit has emerged as a pressing issue in urban rail transit safety. Conventional data-driven prediction methods often lack physical interpretability, and their performance tends to deteriorate when only a small number of training samples are available. To fill this gap, the present study proposes a physics-guided gradient boosting regression tree (PG-GBRT) framework. This architecture integrates domain knowledge from geotechnical engineering into a machine learning model to achieve reliable tunnel deformation predictions. The framework consists of two hierarchical layers. In the first layer, a physics-based empirical formula captures the fundamental geometric relationships. It draws on three theoretical foundations: Gaussian settlement trough theory, the link between excavation depth and volume loss, and three-dimensional geometric effects. The second layer is a residual machine learning model designed to handle the complex nonlinear interactions that the physical component cannot adequately represent. The empirical formula adopts a multiplicative structure. Four key geometric parameters enter this formulation: excavation depth, vertical clear distance, crossing length, and intersection angle. The gradient boosting regression tree then learns systematic deviations from what the physical formula predicts. To validate the framework, we compile 154 engineering cases from major cities across China. On the training set, PG-GBRT achieves an R² of 0.9023, an RMSE of 0.6123 mm, an MAE of 0.4567 mm, and a MAPE of 8.23%. These results compare favorably with three baseline models: standalone GBRT, XGBoost, and LightGBM. Relative to the best performing baseline, PG-GBRT improves R² by 7.1% and reduces RMSE by 37.9%. Physical interpretability analysis reveals that the physical prediction component dominates feature importance. Residual learning compresses the prediction variance effectively, as reflected by a reduction in residual RMSE from 4.078 mm to 0.477 mm. Parameter sensitivity analysis further shows distinct sensitivity patterns across different physical parameters, with each pattern mirroring the parameter's specific role in the mathematical structure of the proposed formula. Overall, the PG-GBRT framework reconciles physical interpretability with the flexibility of machine learning, offering a practical prediction tool for excavation–tunnel interaction problems, particularly in data‑scarce scenarios.

Keywords

Foundation pit excavation / Tunnel deformation / Physics-guided machine learning / Gradient boosting regression tree / Physical interpretability

References

[1] Chen H, Shen Q, Feng Z, Liu Y. Vulnerability assessment in urban metro systems based on an improved cloud model and a Bayesian network. Sustainable Cities and Society., 2023, 98(2210–6707 104823

[2] Clough, G.W., O’Rourke, T.D. (1990). Construction induced movements of in situ walls. In: O’Rourke T D, Clough G W, editors. Design and Performance of Earth Retaining Structures. ASCE Geotechnical Special Publication No. 25. New York: ASCE; p. 439–470.

[3] Feng T, Wang C, Zhang J, Wang B, Jin YF. An improved artificial bee colony-random forest (IABC-RF) model for predicting the tunnel deformation due to an adjacent foundation pit excavation. Underground Space., 2022, 74): 514-527

[4] Gao Z, Fu Z, Wen M, Guo Y, Zhang Y. Physical informed neural network for thermo-hydral analysis of fire-loaded concrete. Engineering Analysis with Boundary Elements., 2024, 158: 252-261

[5] Ge S, Gao W, Cui S, Chen X, Wang S. Safety prediction of shield tunnel construction using deep belief network and whale optimization algorithm. Automation in Construction., 2022, 142(104488 104488

[6] Huang J, Feng C, Wang X, Zhang Y. Continuous–discontinuous element method for simulating three-dimensional reinforced concrete structures. Structural Concrete., 2024, 265): 5846-5861

[7] Kim D, Kwon K, Pham K, Oh JY, Choi H. Surface settlement prediction for urban tunneling using machine learning algorithms with Bayesian optimization. Automation in Construction., 2022, 140(104331 104331

[8] Kim D, Pham K, Oh JY, Lee S, Choi H. Classification of surface settlement levels induced by TBM driving in urban areas using random forest with data-driven feature selection. Automation in Construction., 2022, 135104109): 104109-104109

[9] Li Y, Feng C, Zhang Y. Numerical analysis of directional rock blasting with continuous-discontinuous element method. KSCE Journal of Civil Engineering., 2023, 27(8): 3591-3598

[10] Lin XT, Chen RD, Chen X, Zeng H. Hybrid physics and data-driven method for predicting existing tunnel lining deformation in twin tunnels construction. Computers and Geotechnics., 2025, 179(07 107019

[11] Liu B, Wu W, Lu H, Chen S, Zhang D. Effect and control of foundation pit excavation on existing tunnels: A state-of-the-art review. Tunnelling and Underground Space Technology., 2024, 147105704): 105704-105704

[12] Liu J, Xue B, Wang H, Zhang X, Zhang Y. Numerical study on the behavior of an existing tunnel during excavating adjacent deep foundation pit. Sustainability., 2023, 15(12): 9740-9740

[13] Long M. Database for retaining wall and ground movements due to deep excavations. Journal of Geotechnical and Geoenvironmental Engineering., 2001, 127(3): 203-224

[14] Ma J, Ma C, Li T, Yan W, Dai K. Real-time classification model for tunnel surrounding rocks based on high-resolution neural network and structure–optimizer hyperparameter optimization. Computers and Geotechnics, 2024

[15] Mair, R.J., Taylor, R.N. (1997). Bored tunnelling in the urban environment. In: Proceedings of the 14th International Conference on Soil Mechanics and Foundation Engineering. Hamburg; Vol. 4, p. 2353–2385.

[16] Mair RJ, Taylor RN, Bracegirdle A. Subsurface settlement profiles above tunnels in clays. Géotechnique., 1993, 43(2): 315-320

[17] Meng F, Chen R, Wu H, Xie S, Liu Y. Observed behaviors of a long and deep excavation and collinear underlying tunnels in Shenzhen granite residual soil. Tunnelling and Underground Space Technology., 2020, 103(103504): 103504-103504

[18] Nie W, Wang J, Feng C, Zhang Y. Continuous-discontinuous element method for three-dimensional thermal cracking of rocks. Journal of Rock Mechanics and Geotechnical Engineering., 2023, 15(11): 2917-2929

[19] O’Reilly, M.P., New, B.M. (1982). Settlements above tunnels in the United Kingdom – their magnitude and prediction. In: Tunnelling ’82. London: Institution of Mining and Metallurgy; p. 173–181.

[20] Pan Y, Zhang L. Mitigating tunnel-induced damages using deep neural networks. Automation in Construction., 2022, 138(104219 104219

[21] Peck, R.B. (1969). Deep excavations and tunneling in soft ground. In: Proceedings of the 7th International Conference on Soil Mechanics and Foundation Engineering, State-of-the-Art Volume. Mexico City; p. 225–290.

[22] Phoon KK, Zhang W. Future of machine learning in geotechnics. Georisk: Assessment and Management of Risk for Engineered Systems and Geohazards, 2022, 171): 1-16

[23] Potts DM, Addenbrooke TI. A structure’s influence on tunnelling-induced ground movements. Géotechnique., 1997, 47(4): 991-994

[24] Qian Q, Lin P. Safety risk management of underground engineering in China: Progress, challenges and strategies. Journal of Rock Mechanics and Geotechnical Engineering., 2016, 8(4): 423-442

[25] Qiu J, Jiang J, Zhou X, Zhang Y, Pan Y. Analytical solution for evaluating deformation response of existing metro tunnel due to excavation of adjacent foundation pit. Journal of Central South University., 2021, 28(6): 1888-1900

[26] Sun H, Chen Y, Zhang J, Kuang T-S. Analytical investigation of tunnel deformation caused by circular foundation pit excavation. Computers and Geotechnics, 2019, 106: 193-198

[27] Sun Z, Zhang Y, Yuan Y, Mang HA. Stability analysis of a fire-loaded shallow tunnel by means of a thermo-hydro-chemo-mechanical model and discontinuity layout optimization. International Journal for Numerical and Analytical Methods in Geomechanics., 2019, 43(16): 2551-2564

[28] Wang Y, Fan Q, Dai F, Wang R, Ding B. A physics-data-driven method for predicting surface and building settlement induced by tunnel construction. Computers and Geotechnics., 2024, 179 107020

[29] Wu X, Feng Z, Yang S, Qin Y, Chen H, Liu Y. Safety risk perception and control of water inrush during tunnel excavation in karst areas: An improved uncertain information fusion method. Automation in Construction., 2024, 163(0926–5805 105421

[30] Zhang P, Wu HN, Chen RP, Chan TH. Hybrid meta-heuristic and machine learning algorithms for tunneling-induced settlement prediction: A comparative study. Tunnelling and Underground Space Technology, 2020, 99: 103383

[31] Zhang Y, Zeiml M, Pichler C, Lackner R. Model-based risk assessment of concrete spalling in tunnel linings under fire loading. Engineering Structures., 2014, 77: 207-215

[32] Zhou Z, Chen S, Tu P, Zhang H. An analytic study on the deflection of subway tunnel due to adjacent excavation of foundation pit. Journal of Modern Transportation., 2015, 234): 287-297

[33] Zhuang Y, Cui X, Hu S. Numerical simulation and simplified analytical method to evaluate the displacement of adjacent tunnels caused by excavation. Tunnelling and Underground Space Technology., 2023, 132104879 104879

Funding

the National Key R&D Program of China(Grant No. 2024YFB2604900)

RIGHTS & PERMISSIONS

The Author(s)