Hybrid physics–data prediction approach for interpretable deformation prediction in pre-support tunnel construction

Abstract

Deformation prediction during pre-support tunnel construction remains a major challenge: conventional methods lack adaptability, while deep learning models often sacrifice physical interpretability for accuracy. This study addresses the research question of how to achieve predictions that are both accurate and physically consistent under dynamic tunnel construction. We propose a hybrid physics–data prediction approach (HPDPA) that integrates a physical–mathematical model grounded in Winkler elastic foundation beam theory with a physics-informed attention long short-term memory (PI-ALSTM) network. Through a bidirectional collaboration mechanism, the PI-ALSTM supplies data-driven variables to refine the physical–mathematical model, while the latter informs the learning process of the data-driven model, ensuring consistency with physical mechanics and adaptability to the updated site-specific conditions. When validated on real-world pipe-roof tunneling data, HPDPA achieves high predictive precision (MAE = 1.11 mm, RMSE = 0.97 mm, MRE = 0.07) and markedly outperforms purely data-driven baselines. These results demonstrate the framework’s capability to provide physically trustworthy, high-resolution, and real-time guidance for tunnel construction, thereby contributing to the digitalization process of risk management in pre-support tunnel projects.

Keywords

Winkler elastic foundation beam model / Deep learning / Hybrid physics–data / Pre-support tunnel construction / Structural deformation

References

[1] Bazmara, M., Silani, M., Mianroodi, M., & Sheibanian, M . (2023). Physics-informed neural networks for nonlinear bending of 3D functionally graded beam. Structures, 49, 152-162.

[2] Chang, J. Q., Huang, H. W., Thewes, M., Zhang, D. M., & Wu, H. M. (2024). Data-based postural prediction of shield tunneling via machine learning with physical information. Computers and Geotechnics, 174, 106584.

[3] Hao, Z. H., Zhang, H., Zhang, G., Xiong, W., & Wang, L. (2022). The prediction of ground settlement of a box culvert jacked under the action of an ultra-shallow buried pipe curtain. Arabian Journal for Science and Engineering, 47, 12423-12438.

[4] Li, J. L., Zhang, S., & Wang, X. F. (2025). Physics-informed neural network with fuzzy partial differential equation for pavement performance prediction. Automation in Construction, 171, 105983.

[5] Li, X. Y., Pan, Y., Zhang, L. M., & Chen, J. J. (2023). Dynamic and explainable deep learning-based risk prediction on adjacent building induced by deep excavation. Tunnelling and Underground Space Technology, 140, 105243.

[6] Liang, L. J., Xu, C. J., Zhu, B. T., & Deng, J. L. (2020). Theoretical method for an elastic infinite beam resting on a deformable foundation with a local subsidence. Computers and Geotechnics, 127, 103740.

[7] Lin, Y. H., Li, M. Y., Zhang, Q. R., & Yang, J. H. (2024). Application and effect evaluation of elastic foundation beam method in deformation analysis of underground passage underneath excavated tunnel. Applied Sciences, 14(22), 10435.

[8] Lu, B., Dong, J. C., Zhao, W., Du, X., Cheng, C., Bai, Q., Wang, Z. G., Zhao, M. C., & Han, J. Y. (2022). Novel pipe-roof method for a super shallow buried and large-span metro underground station. Underground Space, 7(1), 134-150.

[9] Pan, Y., He, W., & Chen, J. J. (2025a). Spatiotemporal deep learning for multi-attribute prediction of excavation-induced risk. Automation in Construction, 171, 105964.

[10] Pan, Y., Li, X. Y., Qin, J. J., Chen, J. J., & Gardoni, P. (2025b). Towards trustworthy excavation-induced risk warning for adjacent building: A Bayesian reasoning based probabilistic deep learning method. Underground Space, 25, 156-175.

[11] Phoon, K. K., Ching, J. Y., & Cao, Z. J. (2022). Unpacking data-centric geotechnics. Underground Space, 7(6), 967-989.

[12] Sato, T., Nakahara, F., Sakai, K., Lin, W. R., & Kishida, K . (2025). Practical use of initial displacement monitoring for predicting support behavior with asymmetric deformation in deep tunnel. Underground Space, 23, 43-67.

[13] Shi, H., Ji, W. W., Zhang, Y. Q., Lyu, Y. X., Yong, L., & You, L. F. (2023). Research on flexural performance of new pipe-roof pre-construction structures. Structures, 57, 105279.

[14] Sun, J. F., Fang, Y., Luo, H., Yao, Z. G., Xiang, L., Wang, J. F., Wang, Y. B., & Jiang, Y. F. (2025). Hybrid deep learning approach for rock tunnel deformation prediction based on spatio-temporal patterns. Underground Space, 20, 100-118.

[15] Wang, X., Pan, Y., & Chen, J. J. (2025). Digital twin with uncertainty-informed deep learning for prompt quantitative risk assessment of deep excavation. Computer-Aided Civil and Infrastructure Engineering, 40(25), 4226-4252.

[16] Xie, X. Y., Zhao, M. R., & Shahrour, I. (2019). Face stability model for rectangular large excavations reinforced by pipe roofing. Tunnelling and Underground Space Technology, 94, 103132.

[17] Yan, H., Gou, H. Y., Hu, F., Ni, Y. Q., Wang, Y. W., Wu, D. C., & Bao, Y. (2024). Smart control of bridge support forces using adaptive bearings based on physics-informed neural network (PINN). Automation in Construction, 168, 105790.

[18] Yang, S. S., Zhang, D. W., Wang, M., Xu, J. M., Shen, C., & Zhang, C. Z. (2024). Ground settlement caused by pipe-roof pre-construction method: Effect of the sequence of jacking pipe groups. Journal of Central South University, 31, 576-588.

[19] Zhang, X. Y., Chen, L., Sheng, Y., Huang, L., Ming, F., & Peng, C. Y. (2024). Modeling mechanical behavior of buried pipe suffering from frost-heaving force based on the Winkler Elastic Foundation Beam Theory. Transportation Geotechnics, 49, 101425.

[20] Zhang, C. Y., Pan, Y., & Chen, J. J. (2025). Image-based prediction for enclosure structure deformation in pipe-roof tunnel construction using a physical-guided and generative deep learning method. Automation in Construction, 171, 106002.

[21] Zhang, L. M., Wang, J. Q., Xia, Z., & Song, X. Q. (2025). Physics-guided deep learning for generative design of large-diameter tunnels under existing metro lines. Automation in Construction, 170, 105901.

[22] Zhang, C. Y., Hou, Y. M., Xia, X. H., Chen, J. J., & Pan, Y. (2026). Multimodal feature fusion deep learning for spatiotemporal prediction of deformation and environmental impacts in pipe-roof tunnel construction. Advanced Engineering Informatics, 69, 104022.

[23] Zhou, X. J., Pan, Y., Qin, J. J., Chen, J. J., & Gardoni, P. (2024). Spatio-temporal prediction of deep excavation-induced ground settlement: A hybrid graphical network approach considering causality. Tunnelling and Underground Space Technology, 146, 105605.

[24] Zhou, Y., Lai, H. P., Chen, Y. P., & Li, Z. L. (2023). A novel model for mechanics and safety calculation of pipe roof based on the load transfer mechanism. KSCE Journal of Civil Engineering, 27(12), 5362-5376.

[25] Zhou, Z. L., Zhao, J. P., Tan, Z. S., & Zhou, X. P. (2021). Mechanical responses in the construction process of super-large cross-section tunnel: A case study of Gongbei tunnel. Tunnelling and Underground Space Technology, 115, 104044.