Abstract:Ground settlement prediction during the process of mechanized tunneling is of paramount importance and remains a challenging research topic. Typically, two paradigms are existing: a physics-driven approach utilizing process-oriented computational simulation models for the tunnel-soil interaction and the settlement prediction, and a data-driven approach employing machine learning techniques to establish mappings between influencing factors and the ground settlement. To integrate the advantages of both approaches and to assimilate the data from different sources, we propose a multi-fidelity deep operator network (DeepONet) framework, leveraging the recently developed operator learning methods. The presented framework comprises of two components: a low-fidelity subnet that captures the fundamental ground settlement patterns obtained from finite element simulations, and a high-fidelity subnet that learns the nonlinear correlation between numerical models and real engineering monitoring data. A pre-processing strategy for causality is adopted to consider the spatio-temporal characteristics of the settlement during tunnel excavation. Transfer learning is utilized to reduce the training cost for the low-fidelity subnet. The results show that the proposed method can effectively capture the physical information provided by the numerical simulations and accurately fit measured data as well. Remarkably, even with very limited noisy monitoring data, the proposed model can achieve rapid, accurate, and robust predictions of the full-field ground settlement in real-time during mechanized tunnel excavation.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to predict ground settlement in real - time during mechanized tunnel construction. Specifically, the paper focuses on how to use the multi - fidelity Deep Operator Network (DeepONet) to fuse simulation data and monitoring data in order to achieve real - time prediction of ground settlement during tunnel construction. This problem is very important in engineering practice because the surface settlement caused by the excavation of Tunnel Boring Machines (TBM) may threaten the stability of existing buildings on the ground. Therefore, it is necessary to predict the expected ground settlement in real - time during the construction process so as to adjust construction parameters in a timely manner and ensure construction safety. ### Background and Problem Description of the Paper 1. **Background**: - With the acceleration of the urbanization process, the demand for the construction of underground transportation systems is increasing day by day, and mechanized tunnel construction techniques are widely used in underground engineering projects under various geological and hydrological conditions. - During mechanized tunnel construction, the excavation of Tunnel Boring Machines (TBM) will cause surface settlement, which may pose a threat to buildings on the ground. - Real - time prediction of ground settlement in front of the tunnel is crucial for controlling the face pressure and grouting pressure during the construction process and keeping the ground movement within a predetermined range. 2. **Limitations of Existing Methods**: - Physics - driven methods: Use process - based computational simulation models to simulate tunnel - soil interaction and settlement prediction, but this method depends on complex physical models and has high computational costs. - Data - driven methods: Use machine learning techniques to establish the mapping relationship between influencing factors and ground settlement, but this method is highly dependent on the quantity and quality of data, and the monitoring data available in actual engineering are usually very limited and noisy. 3. **The Method Proposed in the Paper**: - In order to combine the advantages of the two methods and fuse data from different sources, the paper proposes a multi - fidelity Deep Operator Network (DeepONet) framework. - This framework consists of two parts: a low - fidelity sub - network and a high - fidelity sub - network. - **Low - fidelity sub - network**: Obtain the basic ground settlement pattern through finite element simulation. - **High - fidelity sub - network**: Learn the nonlinear correlation between numerical models and actual engineering monitoring data. - Adopt a causal pre - processing strategy to consider the spatial and temporal characteristics of settlement, and use transfer learning to reduce the training cost of the low - fidelity sub - network. ### Main Contributions - **Fast**: Combining two recent advances in DeepONet, the computational cost is reduced through transfer learning in the online stage, enabling settlement reconstruction to be completed within one minute. - **Flexible**: It can predict the settlement at any point on the surface, even if the location information of these points is not included in the training process. - **Accurate**: In the online learning stage, the R² score of the model on the settlement reconstruction task is approximately 0.9. - **Robust**: The model has strong anti - interference ability against the common noise in the measurement process. Even when the data are very limited and noisy, the R² score can still be maintained above 0.8. ### Conclusion The multi - fidelity DeepONet framework proposed in the paper performs excellently in predicting ground settlement in real - time during mechanized tunnel construction. It can effectively fuse simulation data and monitoring data and provide fast, accurate and robust prediction results. This method is expected to be widely used in actual engineering and improve the safety and efficiency of tunnel construction.

A multi-fidelity deep operator network (DeepONet) for fusing simulation and monitoring data: Application to real-time settlement prediction during tunnel construction

Spatial-temporal Fusion Network for Maximum Ground Surface Settlement Prediction During Tunnel Excavation

ANN-Based Dynamic Prediction of Daily Ground Settlement of Foundation Pit Considering Time-Dependent Influence Factors

Influence of adjacent shield tunneling construction on existing tunnel settlement: field monitoring and intelligent prediction

Deep Learning Neural Network Model for Tunnel Ground Surface Settlement Prediction Based on Sensor Data

Data- and experience-driven neural networks for long-term settlement prediction of tunnel

An intelligent construction model in whole process for shield tunnelling

Tunnel Boring Machines (Tbm) Performance Prediction: A Case Study Using Big Data And Deep Learning

Deep-learning surrogate models for the stability of a wide rectangular tunnel

U-DeepONet: U-Net enhanced deep operator network for geologic carbon sequestration

Prediction of surface settlement around subway foundation pits based on spatiotemporal characteristics and deep learning models

Spatio-temporal prediction of deep excavation-induced ground settlement: A hybrid graphical network approach considering causality

Study on intelligent method of prediction by small samples for ground settlement in shield tunnelling

Prediction of Tunneling-Induced Ground Surface Settlement Within Composite Strata Using Multi-Physics-informed Neural Network

Machine learning-based forecasting of soil settlement induced by shield tunneling construction

Fully Convolutional Network-Enhanced DeepONet-Based Surrogate of Predicting the Travel-Time Fields

Point-DeepONet: A Deep Operator Network Integrating PointNet for Nonlinear Analysis of Non-Parametric 3D Geometries and Load Conditions

Measurement and prediction of tunnelling-induced ground settlement in karst region by using expanding deep learning method

DATCN: Deep Attention fused Temporal Convolution Network for the prediction of monitoring indicators in the tunnel

Geom-DeepONet: A Point-cloud-based Deep Operator Network for Field Predictions on 3D Parameterized Geometries

Interfacing finite elements with deep neural operators for fast multiscale modeling of mechanics problems