Abstract:Accurate penetration rate prediction enhances rock-breaking efficiency and reduces disc cutter damage in tunnel boring machine (TBM) construction. However, this process faces significant challenges such as the high uncertainty of ground conditions and the complexity of maintaining optimal TBM operation in long and large tunnels. To address these challenges, we propose TCN-SENet++, a novel hybrid multistep real-time penetration rate prediction model that combines a temporal convolutional network (TCN) and a squeeze-and-excitation (SENet) block for aided tunneling. This study aims to demonstrate the application of TCN-SENet++, as well as other models such as RNN, LSTM, GRU, and TCN, for TBM penetration rate prediction. The model was developed using actual datasets collected from the Yin-Song diversion project. We employ a 30-s time step to predict the future time steps of the penetration rate (1st, 3rd, 5th, 7th, and 9th). The features that influence the penetration rate, such as the cutterhead torque, thrust, and cutterhead power, were considered. A comparative analysis using the mean absolute error and mean squared error revealed that the TCN-SENet++ model outperformed the other models, including RNN, LSTM, GRU, TCN, and TCN-SENet+. In comparison, TCN-SENet++ achieved average MSE reductions of 18%, 6%, 3%, 1%, and 2%, respectively. The TCN-SENet++ model demonstrated fewer errors in the new project, validating its effectiveness and suitability for real-time penetration rate prediction in TBM construction.

What problem does this paper attempt to address?

This paper aims to solve the problem of how to improve the tunneling efficiency of tunnel boring machines (TBMs) and reduce cutterhead damage during construction in hard rock. Specifically, the paper focuses on how to accurately predict the penetration rate of TBMs, as this directly affects the progress and cost of the project. ### Background and Challenges - **High Uncertainty**: The high uncertainty of underground conditions poses significant challenges to TBM construction. - **Complex Operations**: Maintaining the optimal operation of TBMs in long - distance and large - scale tunnels becomes very complex. - **Limitations of Existing Methods**: Traditional theoretical and empirical models have large errors when dealing with complex geological conditions, and although existing intelligent models have a certain predictive ability, they can often only provide a single reference value and are difficult to handle long - time - series data. ### Solutions To address these challenges, the paper proposes a new hybrid multi - step real - time prediction model - TCN - SENet++, which combines the Temporal Convolutional Network (TCN) and the Squeeze - and - Excitation (SENet) block. The following are the main features of this model: 1. **Temporal Convolutional Network (TCN)**: - **Causal Convolution**: Ensure that future information does not leak into the past and guarantee the causality of the model. - **Dilated Convolution**: Handle long - term memory by increasing the receptive field while maintaining the computational efficiency of the model. - **Residual Connection**: Prevent network performance degradation through cross - layer information transfer. 2. **Squeeze - and - Excitation Block (SENet)**: - **Squeeze**: Compress each feature channel into a scalar through global average pooling. - **Excitation**: Calculate the weights of different feature channels through a fully - connected layer to enhance the influence of important feature channels. - **Scaling**: Add the weights directly to the early features and adjust the importance of the features through a dot - product operation. ### Experimental Verification - **Dataset**: Used the actual data from the Jilin Yinsong Project, covering multiple rock types and geological conditions. - **Performance Comparison**: Compared with other models (such as RNN, LSTM, GRU, TCN, and TCN - SENet+), the results show that the TCN - SENet++ model performs best in terms of mean - squared error (MSE), reducing the error by 18%, 6%, 3%, 1% and 2% respectively. - **New Project Verification**: In new engineering projects, the TCN - SENet++ model shows less error, verifying its effectiveness and applicability. ### Conclusion The TCN - SENet++ model effectively improves the prediction accuracy of the TBM penetration rate by combining the advantages of TCN and SENet, which helps to improve construction efficiency, enhance safety and reduce project costs. This research provides strong technical support for TBM construction.

Multi-step real-time prediction of hard-rock TBM penetration rate combining temporal convolutional network and squeeze-and-excitation block

Real-Time Prediction of TBM Driving Parameters Using Geological and Operation Data<italic/>

Performance Prediction of Hard Rock TBM Based on Extreme Learning Machine

Real-Time Prediction of TBM Driving Parameters Using In Situ Geological and Operation Data

Realtime prediction of hard rock TBM advance rate using temporal convolutional network (TCN) with tunnel construction big data

Prediction of TBM cutterhead speed and penetration rate for high-efficiency excavation of hard rock tunnel using CNN-LSTM model with construction big data

TBM penetration rate prediction based on the long short-term memory neural network

Real-time prediction of TBM penetration rates using a transformer-based ensemble deep learning model

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

Data-driven multi-output prediction for TBM performance during tunnel excavation: An attention-based graph convolutional network approach

TBM performance prediction using LSTM-based hybrid neural network model: Case study of Baimang River tunnel project in Shenzhen, China

Dynamic prediction of penetration rate based on TBM operational data

A Performance-Based Hybrid Deep Learning Model for Predicting TBM Advance Rate Using Attention-ResNet-LSTM

A Real-Time Multiple Tunneling Parameter Prediction Method of TBM Steady Phase Based on Dual Recurrent Neural Networks

Research on Prediction of Excavation Parameters for Deep Buried Tunnel Boring Machine Based on Convolutional Neural Network-Long Short-Term Memory Model

A real-time prediction method for tunnel boring machine cutter-head torque using bidirectional long short-term memory networks optimized by multi-algorithm

Predicting the Performance of Tunnel Boring Machines using Big Operational Data

Supervised Machine Learning Techniques to the Prediction of Tunnel Boring Machine Penetration Rate

Big Data-Based Performance Analysis of Tunnel Boring Machine Tunneling Using Deep Learning

TBM penetration rate prediction ensemble model based on full-scale linear cutting test

Prediction of TBM Tunneling Parameters Through an LSTM Neural Network