The advance rate (AR) is a significant parameter in shield tunneling construction, which has a major impact on construction efficiency. From a practical perspective, it's helpful to establish a predictive model of the AR, which takes into account the instantaneous parameters as well as the past operations. However, for shield tunneling in mixed ground conditions, most researches focused on the average values of AR per ring and neglect the influence of past operations. This article presents a long short-term memory (LSTM) recurrent neural network model, which was developed for the slurry shield tunneling in a mixed ground of round gravel and mudstone in Nanning metro. A temporal aggregated random forest is employed to rank the importance of the explanatory features. The model performances in different ground conditions are investigated. The results show that the LSTM model can be effectively implemented for the AR prediction. A high correlation is observed between predicted and measured AR with a correlation coefficient ($R^{2}$ ) of 0.93. The LSTM based AR predictive model is compared with the random forest (RF) model, the deep feedforward network (DFN) model, and the support vector regression (SVR) model. The comparison shows that the LSTM model has the best performances compared to other models. With one-fourth features, we can achieve a 95% prediction accuracy measured by the $R^{2}$ in the proposed model.

Clogging constitutes a significant obstacle to shield tunneling in mudstone soils. Previous research has focused on investigating the influence of soils and slurry properties on clogging, although little attention has been paid to the impact of tunneling parameters on clogging, and particularly early clogging warning during tunneling. This paper contributes to developing a real-time clogging early-warning approach, based on a self-updating machine learning method. The clogging judgment criteria are based on the statistical characteristics of whole-ring tunneling parameters. The paper proposes the use of random forest (RF) for a real-time self-updating early warning strategy for clogging. The performance of this approach is illustrated through its application to a slurry-pressure-balanced shield tunneling construction of Nanning metro line 1. Results show that the RF-based approach can predict clogging during a ring construction with only four minutes of tunneling data, with an accuracy of 95%. The RF model provided the best performance compared with the other machine learning methods. Furthermore, the RF model can realize an accurate clogging prediction in one ring, using less tunneling data with the self-updating mechanism.

As an important part of urban rail transit, subway tunnels play an important role in alleviating traffic pressure in mega-cities. Identifying and locating damage to the tunnel structure as early as possible has important practical significance for maintaining the long-term safe operation of subway tunnels. Summarizing the current status and shortcomings of the structural health monitoring of subway tunnels, a very economical and effective monitoring program is proposed, which is to use the train vibration response to identify and locate the damage of the tunnel structure. Firstly, the control equation of vehicle–tunnel coupling vibration is established and its analytical solution is given as the theoretical basis of this paper. Then, a damage index based on the cumulative sum of wavelet packet energy change rate (TDI_SC) is proposed, and its process algorithm is given. Through the joint simulation of VI-Rail and ANSYS, a refined 3D train-tunnel coupled vibration model is established. In this model, different combined conditions of single damage and double damage verify the validity of the damage index. The effectiveness of this damage index was further verified through model tests, and the influence of vehicle speed and load on the algorithm was discussed. Numerical simulation and experimental results show that the TDI_SC can effectively locate the damage of the tunnel structure and has good robustness.

Under the dual effects of the rapid growth of tunnel mileage and operating years, the application and research of tunnel crack identification based on machine vision are increasing with the vigorous development of machine vision. However, due to the complex environment in tunnels, it is difficult to quickly obtain tunnel lining cracks via computer visions in the tunnel. Therefore, this paper presents the design of a fast acquisition system with the geometric feature analysis for tunnel lining cracks, which has been integrated into a tunnel fast inspection vehicle with a machine vision module. Through the research on the image acquisition system of the tunnel lining, the parameter selection of the crack shooting hardware system is determined, and the fast calculation method of shooting parameters is proposed. The geometric characteristic analysis of the tunnel lining crack image is employed to calculate crack width and determine the optimal gray value of crack extraction. Field tests have been conducted in the highway tunnels in Zhejiang and Yunnan provinces in China and the result indicates that the proposed approach yields much better performance in the detection efficiency, whose time of detection is only 1%, and the number of personnel required is only 40% of the traditional pure manual method. Compared with similar systems, it also has significant advantages in crack resolution and detection speed. This research provides a means of rapid acquisition of tunnel cracks and laying a foundation for the evaluation of the service performance of the tunnel.

In order to overcome the difficulty in rapid detection for expressway tunnels, the coherence calculation of dual-frequency radar signal in the time domain is proposed to suppress the interference. A dual-frequency (400 MHz and 900 MHz) GPR and a manipulator are developed. In the horizontal direction, it has a −90°~90° range, a vertical rotation range from 0 to 90°, a length range from 5~9.5 m, and an antenna rotation between −40°~40°. The mobile scanning has been realized in the expressway tunnel, taking the vehicle as the mobile carrier. The research shows that: This equipment realizes the rapid detection of the tunnel without affecting the expressway’s opening. The imaging algorithm recognizes the structural thickness, reinforcement distribution characteristics, and structural diseases. According to the detection carried out in Zhejiang, China, the spacing between reinforcement in the second lining is 4 cm, and the thickness of the structure is about 0.55 m. However, the reinforcement has deformed badly, and the defects are also discerned in the structure. Compared with the traditional handheld radar detection, the equipment dramatically reduces the labor demand, time, and cost and can meet engineering needs. With the proposed method, the detection time is reduced to 12 min/km from 0.5 d/km, and the cost is reduced by more than two times. Furthermore, during the detection, the traffic can be maintained normally.

The safety of tunneling with shield tunnel boring machines largely depends on the tunnel face pressure, which is currently decided by human operators empirically. Face pressure control is vulnerable to human misjudgment and human errors can cause severe consequences, especially in difficult ground conditions. From a practical perspective, it is therefore beneficial to have a model capable of predicting the tunnel face pressure given operation and the changing geology. In this paper, we propose such a model based on deep learning. More specifically, a long short‐term memory (LSTM) recurrent neural network is employed for tunnel face pressure prediction. To correlate with PLC data, linear interpolation is employed to transform the borehole geological data into sequential geological data according to the shield machine position. The slurry pressure in the excavation chamber (SPE) is taken as the output in the case study of Nanning Metro, which is confronted with the clogging problem due to the mixed ground of mudstone and round gravel. The LSTM‐based SPE prediction model achieved an overall MAPE and RMSE of 3.83% and 10.3 kPa, respectively, in mudstone rich ground conditions. Factors that influence the model, including different kinds and length of input data and comparison with the traditional machine learning‐based model, are also discussed.