Related Experiment Video
Updated: Oct 16, 2025
07:58
Data Processing Methods for 3D Seismic Imaging of Subsurface Volcanoes: Applications to the Tarim Flood Basalt
Published on: August 7, 2017
9.5K
Prediction of Geological Parameters during Tunneling by Time Series Analysis on In Situ Data
Shanglin Liu1, Kaihong Yang1, Jie Cai2
1Key Laboratory of Modern Engineering Mechanics, Tianjin University, Tianjin 300072, China.
Computational Intelligence and Neuroscience
|October 21, 2021
Summary
This study introduces a Long Short-Term Memory (LSTM) neural network to predict geological parameters during tunnel boring machine (TBM) operations. The LSTM method significantly improves prediction accuracy and robustness compared to traditional methods.
Area of Science:
- Geotechnical Engineering
- Artificial Intelligence in Construction
- Time Series Analysis
Background:
- Geological conditions critically influence tunnel boring machine (TBM) performance and safety.
- Acquiring real-time geological information during underground tunneling is challenging.
- Understanding the equipment-geology interaction is vital for efficient TBM operation.
Purpose of the Study:
- To develop a predictive method for geological parameters using TBM real-time data.
- To leverage the sequential nature of in situ data for enhanced geological forecasting.
- To improve the control and safety of underground tunnel construction.
Main Methods:
- Application of Long Short-Term Memory (LSTM) time series neural network.
- Processing of sequential in situ data from TBM operations.
- Comparative analysis with Artificial Neural Network (ANN) for performance evaluation.
Main Results:
- Achieved R-squared values above 0.98 for predicting five geological parameters.
- Demonstrated significantly higher prediction accuracy and robustness of LSTM over ANN.
- Successfully extracted sequential properties from in situ data, reflecting equipment-geology interaction.
Conclusions:
- The proposed LSTM method offers a novel approach for real-time geological information acquisition in TBM construction.
- LSTM's memory structure effectively captures complex equipment-geology dynamics.
- The findings provide a valuable reference for analyzing sequential in situ data in tunneling projects.
