Tunnel Boring Machine Tunneling Research Articles

Prediction of rock mass classification in tunnel boring machine tunneling using the principal component analysis (PCA)–gated recurrent unit (GRU) neural network

Prediction of rock mass classification in tunnel boring machine tunneling using the principal component analysis (PCA)–gated recurrent unit (GRU) neural network

АНАЛИТИЧЕСКИЙ ПОДХОД К ОПРЕДЕЛЕНИЮ ПРОДОЛЬНОГО СМЕЩЕНИЯ СУЩЕСТВУЮЩЕГО СБОРНОГО ТОННЕЛЯ ПРИ ВЫЕМКЕ КОТЛОВАНА ПО ВСЕЙ ДЛИНЕ

In the realm of urban construction employing excavation techniques, safeguarding existing underground structures from detrimental consequences arising from surface construction operations poses a formidable challenge. The reduction of loads due to excavation activities can induce unintended responses, potentially jeopardizing subterranean infrastructure, particularly high-safety-demanding structures like Tunnel Boring Machine (TBM) tunnels. This article introduces an uncomplicated method for ascertaining the axial displacement of TBM tunnels amidst concurrent surface excavation activities. Primarily, the approach entails the identification of stress variations encountered during soil excavation at the tunnel face. Subsequently, employing the solutions derived for the determination of tunnel deformation subjected to concentrated loads, the deformation incurred by the tunnel due to alterations in excavation-induced stress is quantified. The analytical outcomes are meticulously juxtaposed against results generated from a three-dimensional computational model. The comparative analysis demonstrates that the displacement values and axial deviations calculated using the proposed analytical method exhibit only marginal disparities of 4,3% and 1%, respectively, when compared to those obtained through finite element analysis. This study underscores the efficient predictive capabilities of the analytical method in assessing tunnel deformations, enabling a preliminary estimation of critical parameters associated with the excavation pit. These findings have significant implications for mitigating adverse impacts on existing subterranean infrastructure in densely populated urban areas.

Grouped machine learning methods for predicting rock mass parameters in a tunnel boring machine‐driven tunnel based on fuzzy C‐means clustering

AbstractTo guarantee safe and efficient tunneling of a tunnel boring machine (TBM), rapid and accurate judgment of the rock mass condition is essential. Based on fuzzy C‐means clustering, this paper proposes a grouped machine learning method for predicting rock mass parameters. An elaborate data set on field rock mass is collected, which also matches field TBM tunneling. Meanwhile, target stratum samples are divided into several clusters by fuzzy C‐means clustering, and multiple submodels are trained by samples in different clusters with the input of pretreated TBM tunneling data and the output of rock mass parameter data. Each testing sample or newly encountered tunneling condition can be predicted by multiple submodels with the weight of the membership degree of the sample to each cluster. The proposed method has been realized by 100 training samples and verified by 30 testing samples collected from the C1 part of the Pearl Delta water resources allocation project. The average percentage error of uniaxial compressive strength and joint frequency (Jf) of the 30 testing samples predicted by the pure back propagation (BP) neural network is 13.62% and 12.38%, while that predicted by the BP neural network combined with fuzzy C‐means is 7.66% and 6.40%, respectively. In addition, by combining fuzzy C‐means clustering, the prediction accuracies of support vector regression and random forest are also improved to different degrees, which demonstrates that fuzzy C‐means clustering is helpful for improving the prediction accuracy of machine learning and thus has good applicability. Accordingly, the proposed method is valuable for predicting rock mass parameters during TBM tunneling.

Seismic Isolation Materials for Bored Rock Tunnels: A Parametric Analysis

Most recent tunnel designs rely on more thorough analyses of the intricate rock interactions. The three principal techniques for excavating rock tunneling are drill-and-blast for complete or partial cross-sections, TBM only for circular cross-sections with full faces, and road header for small portions. Tunnel-boring machines (TBM) are being utilized to excavate an increasing number of tunnels. Newer studies have demonstrated that subterranean structures such as tunnels produce a variety of consequences during and after ground shaking, challenging the long-held belief that they are among the most earthquake-resistant structures. Consequently, engineering assessment has become crucial for these unique structures from both the geotechnical and structural engineering standpoints. The designer should evaluate the underground structure’s safety to ensure it can sustain various applied loads, considering both seismic loads and temporary and permanent static loads. This paper investigates how adding elastic, soft material between a circular tunnel and the surrounding rock affects seismic response. To conduct the study, Midas/GTS-NX was used to model the TBM tunnel and the nearby rock using the finite element (F.E.) method to simulate the soil–tunnel interactions. A time–history analysis of the El Centro (1940) earthquake was used to calculated the stresses accumulated in the tunnels during seismic episodes. Peak ground accelerations of 0.10–0.30 g, relative to the tunnel axis, were used for excitation. The analysis utilized a time step of 0.02 s, and the duration of the seismic event was set at 10 s. Numerical models were developed to represent tunnels passing through rock, with the traditional grout pea gravel vs. isolation layer. A parametric study determined how isolation material characteristics like shear modulus, Poisson’s ratio, and unit weight affect tunnel-induced stresses. In the meantime, this paper details the effects of various seismic isolation materials, such as geofoam, foam concrete, and silicon-based isolation material, to improve protection against seismic shaking. The analysis’s findings are discussed, and how seismic isolation affects these important structures’ performance and safety requirements is explained.

Investigation on Disaster Mechanism of Diversion Tunnel Induced by Gripper TBM in Hydrokarst Erosion Stratum and Engineering Measures

In gripper tunnel boring machine (TBM) tunneling through complex geological formations, the safe and efficient recovery from large-scale collapses remains a formidable challenge. In this study, we investigate the causes of a 1246 m3 collapse that occurred during the gripper TBM tunneling in the diversion tunnel in Xinjiang, China. Various techniques including TSP seismic waves, CFC advanced water exploration, laboratory point load tests and packer permeability tests were employed for thorough research. The examination discloses that the water softening in biotite-quartz schist in fractured zones contributes significantly to the loosening and dislocation of rock layers along joints. The gripper TBM’s cutterhead exacerbates this process through cutting action and vibrations, causing large-scale instability and eventual rock mass collapse. To tackle this engineering problem, we propose a three-step treatment scheme comprising “Reinforcement-Backfill-Re-excavation”. Furthermore, we propose a technique to handle TBM collapses by creating a “protective shell” within the cavity. The safety and feasibility of these proposed solutions were thoroughly validated through numerical simulations. Also, we utilized the Hoek-Brown theory and Rostami prediction formula to establish recommended values for the total thrust and total torque of the TBM during the collapsed section. The proposed treatment scheme and estimated parameters were successfully applied, resulting in a comprehensive solution from collapse handling to tunneling. This study offers valuable details on effectively managing large-scale collapses in gripper TBM tunneling, which can be useful for similar tunnel engineering and improve safety and efficiency.

대심도 Shield TBM 터널의 IoT 스마트 안전관리시스템 구축에 관한 연구

The purpose of this study is to establish a smart safety management system during deep-seated Shield TBM(Tunnel Boring Machine)tunnel construction. During this work, the site is continuously managed and controlled audio-visually to identify and discover unexpected situations in advance. This study aims to contribute to preventing accidents during TBM (Tunnel Boring Machine) construction while performing efficient work by responding accurately. The system is composed of turn gates and vascular recognizers that can control workers'access, smart tags for tracking workers' location in real-time, emergency call systems, Workers motion sensor to detecting workers who are not moving, monitoring through CCTV cameras, PTZ (Pan Tilt Zoom), DOME, sensing systems detecting four major gases (02, CO, H2S, CH4) that can cause serious accident, loudspeakers, amplifiers, and two-way interphones, etc.
 As a result of this study, the IoT Smart Safety Management System for the Shield TBM Tunnel Construction in the deep underground construction site has been materialized to establish the Safety Management Control Center, safety management system for TBM Construction Machinery, tunnel internal safety management system, network system, and can make immediate response in case of emergency. This system can guarantee safety of workers who provide labor within the poor working environment of deep underground spaces. The IoT-based smart safety management system established in this study is expected to contribute a IoT to preventing industrial accidents at the site of underground deep-seated TBM tunnel construction (earth pressure type and Slurry shield TBM or Open TBM) over 40m in the future.

Predicting advance rate and cutter life in TBM tunnelling using evolutionary algorithm

ABSTRACT Tunnelling with a Tunnel Boring Machine (TBM) offers several advantages over conventional tunnelling methods when covering long distances, particularly regarding time, safety, and environmental impact. Prior to starting a TBM project, performance analysis is a crucial stage necessary to inform the initial investment decision and project cost estimation. However, this analysis requires extensive fieldwork and laboratory studies. The Advance Rate (AR), calculated from the Utilization Factor (U) – the ratio of excavation time to total project time – is a widely accepted measure of TBM performance. Total project time encompasses various parameters, including boring time, cutter inspection and replacement times, and constant factors like re-gripping, service, and maintenance times. In this study, we employed analytical methods, statistical analysis, and genetic programming approaches to evaluate the AR and thus enhance TBM performance analysis. Gamma tests were conducted to identify interrelated parameters in TBM operations. The results from these tests showed that the Applied Force per Cutter (FN) and penetration values (together forming the Field Penetration Index or FPI) are useful in estimating cutter consumption. While FN can be determined through laboratory tests, its field determination is challenging. We have developed several charts and a function derived from genetic programming to estimate the AR of a TBM. The study demonstrated that this function, formulated using genetic programming, possesses strong predictive capability and can significantly improve TBM performance analysis.

Water saturation effects on rock abrasivity and TBM tunneling efficiency

Tunnel boring machine (TBM) tunneling efficiency is influenced by surrounding rocks, TBM design, and interaction between the TBM cutter and rocks. Using a case study from a water conveyance tunnel project located in northwestern China and involving five different lithologies (metamorphic andesite, sandstone, granite, quartz schist, and quartz diorite), this work investigates the water saturation effects on Cerchar abrasivity index (CAI) and its correlations with other physical and mechanical rock properties and TBM disc cutter and energy consumption. The results show that water saturation reduces CAI for all examined rocks except for the quartz diorite at the 90 % confidence level. CAI of saturated rocks (CAIsat) has statistically better correlations with weighted Mohs hardness value of rock minerals (Ha), equivalent quartz content (EQC), and uniaxial compressive strength (UCS), than CAI of dry rocks (CAIdry). The in-situ construction data analyses suggest that the average disc cutter consumption increases exponentially with the rise of CAI values in both dry and saturated conditions, and the water saturation effects significantly reduce the disc cutter consumption within medium-hard and hard surrounding rocks. Both disc cutter and specific energy consumption have a quadratic relationship with the ratio of the maximum cutter spacing to penetration depth (Smax/P), and water saturation can significantly shift the quadratic relationships. Optimal Smax/P ranges for minimizing the disc cutter and energy consumption are recommended for different rocks under dry and saturated conditions. The findings from this study can support estimating cutter and energy consumption with considerations of groundwater effects.

K-means-based heterogeneous tunneling data analysis method for evaluating rock mass parameters along a TBM tunnel

Rapid and accurate judgment of the rock mass condition is the key to guaranteeing the safety and efficiency of tunnel boring machine (TBM) tunneling. This paper proposes a method for evaluating rock mass parameters based on K-means clustering, grouping tunneling areas according to the values of TBM tunneling parameters. A dataset including rock mass and TBM tunneling data is treated by logistic normalization and principal component analysis (PCA), and large volumes of tunneling data with different features are transformed into appropriate volumes of dimensionless data. K-means clustering is used, samples are grouped according to the values of tunneling data, and the specific ranges as defined by clustering are regarded as the unified evaluated results of each group. Based on the C1 part of the Pearl Delta water resources allocation project, 100 training samples and 30 testing samples were field-collected, and the proposed method was realized by the training samples and verified by the testing samples. The evaluation accuracies of uniaxial compressive strength (UCS), and joint frequency (Jf) were 90%, and 86.7% respectively, demonstrating that the evaluation had acceptable values, and the proposed method was greatly helpful for judging rock conditions.

Transfer learning for collapse warning in TBM tunneling using databases in China

Tunnel boring machine (TBM) has been widely used for tunnelling in harsh environments with deeply embedded tunnels and hard rocks. The collapse of surrounding rocks seriously jeopardizes the safety of the machine and the driver. Prediction of TBM tunneling response is vitally important to alert the hazard in advance, which is still a challenge topic, especially for a new-constructing tunnel where scarce data are available. This study proposes a transfer learning model that is capable to predict TBM tunneling responses for a new project using the boring data obtained from the other existing tunnels. The key problem of transfer learning model, i.e., what to transfer, is first studied based on the big data obtained from Yin-song project and Yin-chao project in China. Then, when to transfer is investigated by comparing with the machine learning model that developed using only the data in the new tunnel. Finally, possible collapses in the new Yin-chao project are analyzed using the TBM responses that obtained from the transfer learning model. Results show that the types of transferred TBM parameters significantly affect the prediction accuracy. A filter procedure is proposed to determine the number and types of parameters to be transferred. The predicted accuracy for the torque and thrust reach 93 % and 95 %, respectively, for the first 100 steps from Ch. 66060 m to Ch. 65366 m in the new Yin-chao project when the most important 10 parameters are used in the transfer learning model. The predicted accuracy of the transfer learning model is better than that of the machine learning model in the early stage of the new tunnelling, especially in the first 2000 steps. The proposed model shows superior capability in prediction of the TBM responses and in collapse warning for a new tunnelling project, which paves the way for an automatic construction of TBM.

Experimental and numerical studies on the shear mechanical behavior of rock joints under normal vibration loads

Jointed rock masses of roadways in underground coal mines are heavily disturbed by rock-cutting vibrations from Tunnel Boring Machines (TBMs), posing significant challenges in controlling the deformation and damage of the surrounding rock. To investigate the mechanical behavior of rock joints under normal vibration loads, a series of laboratory vibration experiments were conducted with varying frequencies and amplitudes, based on in-situ vibration characteristics during TBM rock-cutting. Subsequently, the relative damage coefficient Dv and relative shear strength ratio Rτ were proposed to quantify the damage degree of rock joints. The findings revealed a non-linear growth of rock joints damage with increasing vibration frequency and amplitude. The vibration frequency and amplitude are categorized into three and two stages, respectively, depending on the damage of rock joints. Additionally, the vibration failure mechanism of rock joints was elucidated using the discrete element method. It was observed that the failure of weakly joints is controlled by both the “filler-rock” interfaces and “filler-filler”, while strongly joints fail solely due to the “filler-rock” interfaces, confirming the agreement with laboratory experiments. Moreover, effective grouting reinforcement was proven to significantly enhance the vibration resistance of in-situ jointed rock masses, preventing vibration collapse resulting from TBM tunneling disturbance.

On-site measurement and environmental impact of vibration caused by construction of double-shield TBM tunnel in urban subway

The vibration generated during the construction of subway tunnels with double-shield tunnel boring machine (TBM) has a significant impact on the environment, which has caused multiple complaints from residents. Taking a double-shield TBM tunnel project as the background, vibration measurements were conducted by installing vibration sensors on-site. By combining theoretical methods—such as normalization, polynomial fitting prediction, and gray correlation analysis—the vibration characteristics, impact range on the environment, and factors affecting the vibration of TBM construction were studied. The key research results included: (1) The amplified zone of X and Y vibration acceleration occurred on the left-hand side of the tunnel from 3.15 to 13.85 m, but rapidly decayed away from the amplification zone. (2) The impact range of TBM vibrations on residential areas at night and during the day was studied according to the official “Urban Regional Environmental Vibration Standard” and it was found to be larger at night than during the day. (3)The main factors affecting the TBM vibration level was studied—including the cutter-head torque, TBM thrust, cutter-head speed, penetration, field penetration index (FPI) and so on. In summary, when the double-shield TBM construction tunnel is adjacent to residential areas, the vibration generated exceeds the national standard limit. In order to reduce the impact of TBM vibration on residential areas, excavation parameters such as cutter head torque, TBM thrust, cutter head speed, and penetration should be appropriately reduced.

Nitrogen Dioxide Gas Levels in TBM Tunnel Construction with Diesel Locomotives Based on Directive 2017/164/EU

Directive 2017/164/EU proposed a drastic reduction of nitrogen monoxide (NO) and nitrogen dioxide (NO2) levels, thereby fortifying the health protection framework within the mining industry. Despite the commendable record of non-road emissions standards (Stage IV and V) in continuing to reduce NOx emissions, concerns remain about compliance with the directive’s strict limits, particularly in demanding tunnels and mining fields. To illustrate this problem, this study undertakes a comprehensive assessment of the practical feasibility surrounding the implementation of these proposed limits in a 6.2 internal diameter tunnel-boring machine (TBM) tunnel constructed with Stage III emission locomotives. The results cast light upon the formidable challenges entailed in achieving strict compliance with the envisioned limits, with a substantial number of measurements notably surpassing these thresholds, primarily concerning NO2 emissions from Stage III engines. To address these challenges, this study highlights the key role of moving to Stage IV-V locomotives or introducing electric locomotives to effectively reduce NOx emissions, ensure compliance with the directive, and avoid delays in tunnel construction.

Safety and environmental impact control of cross passage construction in soft soil strata using tunnel boring machine method

The construction of cross passages using the tunnel boring machine (TBM) method represents an emerging construction technique with numerous advantages. However, owing to the scarcity of application instances, the safety control methodologies and the regulatory patterns concerning environmental impacts remain inconclusive. In this study, a cross passage excavated using the TBM method—the first of its kind in the Tianjin area—was investigated. We identified the key risk control measures for the construction and analysed the TBM operating parameters, monitored ground and building settlements, and monitored mainline tunnel deformations and mechanical responses, revealing the ground and tunnel structure deformation patterns. The following conclusions are drawn. (1) The ground surrounding the cross-passage break-out opening was stabilised by performing secondary grouting and small-range freezing, and the break-in opening was excavated using a completely enclosed steel sleeve. These measures prevented water and sand inflows during the excavation of the break-out and break-in openings in the silt and silty sand strata. (2) The torsional moment of the cutter disc was large during the break-out phase. Break-out mainline tunnel displacement monitoring data indicated that the thrust had a significant effect on the mainline tunnel during the break-out phase. (3) The TBM tunnelling caused ground loss. The ground settlement exhibited a U-shaped distribution along the cross-passage axis, with the maximum settlement being 10 mm. (4) During the break-out phase, the deformation of the break-out mainline tunnel exhibited a duck-egg-shaped distribution. The clearance convergence of the break-out mainline tunnel was within ± 4, and the clearance convergence of the break-in mainline tunnel was controlled within ± 1 mm.

Fast perception of rock mass strength and integrity in TBM tunnelling using in-situ penetration test

Tunnel boring machine (TBM) is the preferred method in the construction of long-distance and deep-buried tunnel due to its advantages in safety and high efficiency. However, the TBM method is very sensitive to the condition of the surrounding rock. The strength and integrity characteristics of rock mass have great influence on the safe and efficient tunnelling. To provide a fast perception method for these characteristics during the TBM tunnelling process, an in-situ penetration testing system is proposed in the present study. The development and testing procedure of the system are elaborated. The system is applied to a TBM-excavated water conveyance tunnel to continuously perceive the strength and integrity characteristics of the surrounding rock. Firstly, the force-penetration (F-P) responses of the surrounding rock under penetration are qualitatively studied. Then, series of penetration indices are summarized to quantify the characteristics of the F-P responses. Finally, the quantitative relationships between the rock strength and the penetration indices are studied. The results indicate that the in-situ penetration testing system can perceive the rock mass strength and integrity characteristics qualitatively and quantitatively. On the qualitative aspect, the F-P responses can be categorized into four types (Type I to Type IV), which correspond to the level of rock mass strength and integrity. On the quantitative aspect, a single penetration index cannot represent the complete information of the F-P response. The descriptive classification of the rock mass strength (class I to III) is established through multiple penetration indices. The in-situ penetration testing system can be used as a method to quickly perceive the strength and integrity characteristics of the surrounding rock in TBM tunnelling.

A performance-based hybrid deep learning model for predicting TBM advance rate using Attention-ResNet-LSTM

The technology of tunnel boring machine (TBM) has been widely applied for underground construction worldwide; however, how to ensure the TBM tunneling process safe and efficient remains a major concern. Advance rate is a key parameter of TBM operation and reflects the TBM-ground interaction, for which a reliable prediction helps optimize the TBM performance. Here, we develop a hybrid neural network model, called Attention-ResNet-LSTM, for accurate prediction of the TBM advance rate. A database including geological properties and TBM operational parameters from the Yangtze River Natural Gas Pipeline Project is used to train and test this deep learning model. The evolutionary polynomial regression method is adopted to aid the selection of input parameters. The results of numerical experiments show that our Attention-ResNet-LSTM model outperforms other commonly-used intelligent models with a lower root mean square error and a lower mean absolute percentage error. Further, parametric analyses are conducted to explore the effects of the sequence length of historical data and the model architecture on the prediction accuracy. A correlation analysis between the input and output parameters is also implemented to provide guidance for adjusting relevant TBM operational parameters. The performance of our hybrid intelligent model is demonstrated in a case study of TBM tunneling through a complex ground with variable strata. Finally, data collected from the Baimang River Tunnel Project in Shenzhen of China are used to further test the generalization of our model. The results indicate that, compared to the conventional ResNet-LSTM model, our model has a better predictive capability for scenarios with unknown datasets due to its self-adaptive characteristic.

Rock fragmentation indexes reflecting rock mass quality based on real-time data of TBM tunnelling

Perception of rock condition (RC) is a challenge in tunnel boring machine (TBM) construction due to lack of space and time to observe and detect RC. To overcome this problem, this study aims to extract a new rock fragmentation index (RFI) that can reflect RC from real-time rock fragmentation data of the TBM. First, a comprehensive review of existing rock fragmentation models is conducted, which leads to some candidate RFIs that can reflect RC. Next, these candidate RFIs are investigated using data from 12,237 samples from a well-monitored tunnel boring process of the TBM in a 20,198 m tunnel. Further, a new RFI system is recommended as the parameter involving the optimal models. Finally, a preliminary study of the relationship between these RFIs and RC is carried out, and it is shown that these RFIs can reflect RC to a large extent. In the TBM boring process, these RFIs can be extracted from real-time TBM fragmentation data and used to predict the RC in the field. Therefore, the challenge of RC perception is solved with this new RFI system. The new RFI system offers significant potential for the real-time rock classification, prediction of the surrounding rock collapse potential, and selection of control parameters or support measures during TBM construction. This will be the key to improving TBM construction performance.

Time series clustering-enabled geological condition perception in tunnel boring machine excavation

For automatic geology perception, this paper develops a DTW-Kmedoids-enabled time series clustering framework to gain a clear insight into the geological condition ahead of the tunnel boring machine (TBM) head. The core idea of DTW-Kmedoids is to incorporate dynamic time warping (DTW) as a distance metric into a popular clustering algorithm Kmedoids, aiming to aggregate time series data reflecting similar conditions of surrounding geology into the same cluster. It has been verified in a sequence dataset recording 377 operational rings from a TBM tunneling project in Singapore. Results indicate that the proposed DTW-Kmedoids algorithm outperforms Kmedoids and softDTW-Kmedoids in partitioning the whole dataset into four meaningful clusters on behalf of four geological conditions. Moreover, it shows merits in great robustness to handle varying-length temporal sequences and even fragmentary data instead of the full observations. Furthermore, an online clustering mechanism is well designed to adaptively update the cluster center, which can accomplish the dynamic recognition of the forward geological circumstances ring by ring. In short, the practical value behind the hybrid time series clustering is to automatically and quickly perceive lithology categories along the designed tunnel trail with no requirement of data labeling, possibly allowing for further inference of geological risk and guidance in safe excavation.

Applications of Machine Learning in Mechanised Tunnel Construction: A Systematic Review

Tunnel Boring Machines (TBMs) have become prevalent in tunnel construction due to their high efficiency and reliability. The proliferation of data obtained from site investigations and data acquisition systems provides an opportunity for the application of machine learning (ML) techniques. ML algorithms have been successfully applied in TBM tunnelling because they are particularly effective in capturing complex, non-linear relationships. This study focuses on commonly used ML techniques for TBM tunnelling, with a particular emphasis on data processing, algorithms, optimisation techniques, and evaluation metrics. The primary concerns in TBM applications are discussed, including predicting TBM performance, predicting surface settlement, and time series forecasting. This study reviews the current progress, identifies the challenges, and suggests future developments in the field of intelligent TBM tunnelling construction. This aims to contribute to the ongoing efforts in research and industry toward improving the safety, sustainability, and cost-effectiveness of underground excavation projects.

Seismic ahead-prospecting based on deep learning of retrieving seismic wavefield

Unknown geology ahead of the tunnel boring machine (TBM) brings a large safety risk for tunnel construction. Seismic ahead-prospecting using TBM drilling noise as a source can achieve near-real-time detection, meeting the requirements of TBM rapid drilling. Seismic wavefield retrieval is the key data processing step for the efficient utilization of TBM drilling noise. The traditional solution is based on cross-correlation to extract reflected waves, but the reference waves remain in the result, disturbing the imaging and interpretation of the adverse geology. To solve this problem, the deep learning method was introduced in wavefield retrieval to improve the accuracy of geological prospecting. We trained a deep neural network (DNN) with its strong nonlinear mapping capability to transform seismic data from TBM drilling noise to data from the active source. The issue lies in its features for this specific tunnel task, including the decay of the seismic signal with time and the incomplete spatial correspondence. Thus, we improved a classical DNN with the time constraint as an additional input, and an additional pre-decoder to enlarge the receptive field. Additionally, a loss function weighted by the ground truth and time constraint is improved to achieve an accurate retrieval of the effective signal, considering the little effective information in tunnel data. Finally, the workflow of the proposed method was given, and a dataset designed with reference to the field case was employed to train the network. The proposed method accurately retrieved the reflection signal with higher dominant frequencies, which helped improve the accuracy of imaging. Numerical simulations and imaging on typical geological models show that the proposed method can suppress reference waves and get more accurate results with fewer artifacts. The proposed method has been applied in the Gaoligongshan Tunnel and imaged two abnormal zones, providing meaningful geological information for TBM drilling and tunnel construction.