Prediction of Geological Parameters during Tunneling by Time Series Analysis on In-situ Data: A Recent Study

A tunnel boring machine (TBM) is a type of heavy load equipment commonly employed in the building of underground tunnels. In the tunnelling process, geological conditions are critical factors that have a direct impact on the control of construction equipment. The objective of this paper is to predict the geological information ahead of TBM using continuously collected airborne parameters. The long short-term memory (LSTM) time series neural network technique for processing in-situ data was introduced in this research, which focused on the internal linkages between the sequential character of tunnel in-situ data and the continuous interaction between equipment and geology. On the basis of TBM real-time status monitoring data, a method for predicting geological parameters in advance is proposed. The suggested method was used to forecast five geological parameters for a tunnel project in China, and the R2 of the prediction findings for all five geological parameters was greater than 0.98. The LSTM was compared to an artificial neural network (ANN) in terms of performance. The LSTM's prediction accuracy was much greater than the ANN's, and its generalisation and robustness were also superior to the ANN's, indicating that the suggested LSTM method could extract the sequence features of the in-situ data. Through the adoption of the "gate" idea, the rule of equipment-geology interaction was mirrored in the model's memory structure, allowing for precise prediction of geological parameters during tunnelling. The impact of the time frame and prediction distance on the model is also examined. The suggested method offers a novel strategy to obtaining geological information during TBM construction, as well as a point of reference for analysing in-situ data with sequence properties.