Multi-source collaborative advanced geological exploration method based on TBM construction

Abstract

The invention discloses a multi-source collaborative advanced geological exploration method based on TBM construction, and belongs to the technical field of tunnel engineering geological exploration. A three-dimensional geological model is constructed through multi-scale collaborative operation of directional long exploration drilling, branch short exploration drilling and in-hole transient electromagnetic detection; the directional long exploration adopts a secondary drilling structure and a screw motor directional drilling technology to realize accurate control of a deep stratum, the branch short exploration realizes dynamic and rapid verification through a TBM integrated drilling machine, meanwhile, in-hole transient electromagnetic exploration adopts three-component measurement and an intelligent inversion algorithm to fill the blank of geological information between drill holes, and by means of mutual data correction of the three methods, the detection accuracy is improved. A three-in-one detection system of drilling, geophysical prospecting and intelligent analysis is formed, and then accurate advanced prediction is achieved, so that the problems that a traditional single detection technology is low in precision and insufficient in coverage range are effectively solved, the water inrush risk is remarkably reduced, construction safety is guaranteed, and the method is effectively suitable for deep tunnel TBM safety construction.

Description

Technical Field

[0001] The present invention relates to the technical field of tunnel engineering geological detection, and in particular to a multi-source collaborative advanced geological detection method based on TBM construction. Background Art

[0002] Traditional TBM construction often uses a single drilling or geophysical exploration method, which has the following defects:

[0003] The borehole detection range is limited, with geological blind spots. It can only reflect the geological conditions near the borehole path and may miss lateral anomalies such as hidden faults.

[0004] A single geophysical exploration method is greatly affected by geological conditions, construction environment, etc., and has the risk of multiple solutions and misjudgment.

[0005] Isolated data lacks verification, and there is no dynamic cross-verification mechanism between drilling and geophysical data, resulting in a high risk of water inrush;

[0006] TBM construction is fast. If the geological conditions ahead of the working face are not detected and predicted in advance, it is very likely to cause disasters such as machine jams and sudden water gushing. Therefore, a multi-source collaborative advanced geological detection method based on TBM construction is proposed. Summary of the Invention

[0007] The purpose of the present invention is to provide a multi-source collaborative advanced geological exploration method based on TBM construction to solve the problems raised in the above background technology.

[0008] To achieve the above-mentioned object, the present invention provides the following technical solution: a multi-source collaborative advanced geological exploration method based on TBM construction, comprising the following steps:

[0009] S1. Conduct directional long-distance exploration before TBM starts to obtain long-distance axial geological data along the TBM excavation direction;

[0010] S2. During TBM excavation, branch short exploration is carried out using the equipment's built-in advanced geological drilling rig;

[0011] S3. Conduct transient electromagnetic surveys in long and short boreholes to generate a 30m radial three-dimensional geological profile.

[0012] S4. Compare drill cuttings, electromagnetic anomalies, and drilling parameters through a data fusion platform to identify water-rich areas and structural fracture zones;

[0013] S5. Dynamically update the BIM geological model. When the three results in S1-S4 are consistent, excavation is allowed. If the three results are different, the emergency plan is activated.

[0014] As a further preferred embodiment of the present technical solution: in S1, the directional long-range exploration adopts a φ89mm screw motor for directional drilling, the drilling depth is ≥600m, and the final hole diameter is φ94mm;

[0015] Among them, it includes secondary drilling structure and grouting casing, the grouting casing is φ127mm, and the pressure resistance is ≥1.5MPa;

[0016] Among them, the directional long-distance drilling uses a hole-opening instrument for hole-opening orientation, and its trajectory control accuracy is: the error range of the hole-opening inclination and azimuth angle does not exceed ±0.5°.

[0017] As a further preferred embodiment of the present technical solution: the deviation between the drilling trajectory and the set value is no more than 3m, and a cable drill rod is used to realize the return of real-time inclination measurement data every 3m.

[0018] As a further preferred embodiment of the present technical solution: in S2, the branch short probe is integrated into a hydraulic drilling rig of the TBM saddle frame, and the drilling depth is ≤100m;

[0019] Among them, the branch short exploration drilling distance is 100m / time, the excavation distance is 70m, and a 30m advance safety distance is left. The drilling arrangement is 3 holes evenly distributed along the arch in the range of -90° to 90°, cross-verified, and the elevation angle is 0° to 15°.

[0020] As a further preferred embodiment of the present technical solution: in S3, the in-hole transient electromagnetic detection adopts a mining transient electromagnetic instrument with an outer diameter of φ73mm;

[0021] It includes a three-component measurement system with a radial detection radius of 30m and supports 500m hole depth push.

[0022] As a further preferred embodiment of the present technical solution: the in-hole transient electromagnetic detection adopts a convolutional neural network inversion algorithm, the resistivity and resolution are both ≤0.5m, the low-resistance anomaly threshold is set to ρ<15Ω·m, and the data collection interval is ≤1m.

[0023] As a further preferred embodiment of the present technical solution: in-hole transient electromagnetic detection can realize three-dimensional geological imaging within a radial range of 30m of the borehole, determine low-resistance anomalies and locate their spatial orientation in real time, and support continuous deep hole detection.

[0024] As a further preferred embodiment of the present technical solution: in S4, the data fusion platform is used to integrate drilling parameters, electromagnetic signals and geological data in real time, thereby generating a three-dimensional risk warning model;

[0025] The data fusion processing in the data fusion platform includes the following steps:

[0026] A1. Data collection;

[0027] A2. Data processing;

[0028] A3. Decision analysis;

[0029] A4. Dynamic verification and model update;

[0030] A5. Dynamic update of BIM geological model;

[0031] A6. Engineering application optimization strategy;

[0032] A7. Computing resource configuration;

[0033] A8. Visual decision support;

[0034] Among them, in A1, data collection includes drilling data, geophysical data and construction data;

[0035] Drilling data: particle size analysis of rock cuttings samples from directional long boreholes, multiple real-time parameters including drilling torque, drilling speed, and pump pressure, hole position deviation from branch short boreholes, and rock cuttings mineral composition test data;

[0036] Geophysical data: three-component resistivity data and decay time series from in-hole transient electromagnetic detection, and low-resistance anomaly distribution within a radial range of 30 m;

[0037] Construction data: TBM excavation speed, cutterhead speed, shield pressure equipment and other operating parameters, as well as auxiliary data such as grouting pressure and casing pressure resistance;

[0038] Among them, in A2, data processing includes preprocessing and feature extraction;

[0039] Preprocessing: Sliding window filtering is used to remove the impulse noise of drilling parameters, interpolation algorithm is used to fill the missing values ​​of electromagnetic data, and the cuttings composition is standardized and coded;

[0040] Feature extraction: The spatial coordinates and morphological features of low-resistance anomalies are extracted from electromagnetic signals, and mutation points are extracted from drilling parameters as structural fracture zone features.

[0041] As a further preferred embodiment of this technical solution: In A3, a "three-dimensional risk warning model" is constructed, integrating drilling hole trajectories, geophysical anomalies, and construction parameter thresholds. A visual risk cloud map is generated through spatial interpolation, marking multiple high-risk areas such as water-rich areas and faults. The decision analysis specifically includes:

[0042] B1. Core Fusion Algorithms and Technologies:

[0043] Using multi-source data spatiotemporal registration technology, with the TBM face as the coordinate origin, a unified geographic reference system is established to spatially align the directional long borehole trajectory, the fan-shaped distribution of branch short boreholes, and the transient electromagnetic 3D profile to ensure data coordinate consistency.

[0044] In the temporal dimension, the drilling parameters, electromagnetic detection, and TBM excavation progress are matched through a clock synchronization mechanism to form a spatiotemporal coupled dataset.

[0045] B2. Machine Learning Fusion Model:

[0046] Through random forest risk assessment: the input variables include multiple features, and 500 decision trees are used for ensemble learning to output the probability of water inrush;

[0047] Convolutional neural network inversion: 3D CNN is used to automatically identify the morphology of low-resistance anomalies based on transient electromagnetic data. The model is trained using drilling verification data to reduce the positioning error of water-rich areas to less than 1m.

[0048] B3. Evidence-theoretic decision-making fusion:

[0049] Establish confidence matrices for drilling, geophysical exploration, and construction parameters;

[0050] Transient electromagnetic detection shows a resistivity of <10Ω·m with a confidence level of 0.7;

[0051] When the TBM is drilling, the torque suddenly increases by 20%, with a confidence level of 0.6;

[0052] By integrating multi-source evidence through DS evidence theory, when the comprehensive confidence level is greater than 0.7, it is determined to be a high-risk area and the emergency plan is triggered.

[0053] As a further preferred embodiment of the present technical solution: in A4, the specific steps of dynamic verification and model update include:

[0054] C1. Each time a detection cycle is completed, the following data are automatically compared;

[0055] C2. The consistency of mineral composition between long-hole cuttings and short-hole cuttings;

[0056] C3, spatial consistency between the transient electromagnetic anomaly area and the structural belt revealed by drilling;

[0057] C4. Time correlation between the mutation point of drilling parameters and geophysical anomalies;

[0058] C5. When the consistency of the three-source data is less than 80%, start supplementary detection.

[0059] As a further preferred embodiment of this technical solution: In A5, the BIM geological model is dynamically updated to import the integrated risk data into the BIM model, marking the risk level with different colors and linking it with the TBM construction progress, thus realizing the closed-loop management of "detection-excavation-verification";

[0060] When new structures are discovered, the model prediction parameters are automatically corrected and subsequent detection plans are updated;

[0061] In A6, an abnormal early warning mechanism for drilling data is established: when the drilling trajectory deviation is greater than 3m or the inclination data return is interrupted for more than 5 minutes, a manual verification is triggered;

[0062] Verification of the validity of geophysical data: The reliability of the data is determined by the repeatability of the resistivity of multiple detections in the same borehole.

[0063] As a further preferred embodiment of this technical solution: In A7, computing resource configuration uses edge computing nodes to process real-time drilling parameters, and cloud servers run CNN inversion and random forest models, realizing a hybrid architecture of "local fast response + cloud-based deep analysis", controlling data processing delay within 10 seconds;

[0064] In A8, a three-dimensional interactive interface has been developed to support construction personnel in viewing geological data slices at any location through touch operations. Clicking on abnormal areas can link and display multiple multi-dimensional information such as drilling cuttings photos, electromagnetic waveforms, and drilling parameter curves to assist on-site decision-making.

[0065] Compared with the prior art, the present invention has the following beneficial effects:

[0066] The present invention constructs a three-dimensional geological model through the multi-scale coordinated operation of directional long-exploration drilling, branch short-exploration drilling and in-hole transient electromagnetic detection. The directional long-exploration adopts a two-level drilling structure and screw motor directional drilling technology to achieve precise control of deep strata, while the branch short-exploration realizes dynamic and rapid verification through the TBM integrated drilling rig. At the same time, the in-hole transient electromagnetic detection adopts three-component measurement and intelligent inversion algorithm to fill the geological information gap between boreholes. Through the mutual calibration of data from the three methods, a three-in-one detection system of "drilling-geophysical exploration-intelligent analysis" is formed, thereby realizing accurate and advanced prediction. This effectively solves the problems of low precision and insufficient coverage of traditional single detection technology, significantly reduces the risk of water inrush, ensures construction safety, and is effectively suitable for safe TBM construction in deep tunnels. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 Schematic diagram of the process of the multi-source collaborative advanced geological exploration method based on TBM construction of the present invention;

[0068] Figure 2 Schematic diagram of the planar structure of a directional long exploration hole in the multi-source collaborative advanced geological exploration method based on TBM construction of the present invention;

[0069] Figure 3 Schematic diagram of the cross-sectional structure of a directional long exploration hole in the multi-source collaborative advanced geological exploration method based on TBM construction of the present invention;

[0070] Figure 4 Schematic diagram of the planar structure of branch short exploration holes in the multi-source collaborative advanced geological exploration method based on TBM construction of the present invention;

[0071] Figure 5 Schematic diagram of the cross-sectional structure of a branch short exploration hole in the multi-source collaborative advanced geological exploration method based on TBM construction of the present invention;

[0072] Figure 6 This is a schematic diagram of the structure of the in-hole transient electromagnetic detection in the multi-source collaborative advanced geological detection method based on TBM construction of the present invention;

[0073] Figure 7 Schematic diagram of the data fusion process in the multi-source collaborative advanced geological exploration method based on TBM construction of the present invention;

[0074] Figure 8 It is a flow chart of dynamic verification and model updating in the multi-source collaborative advanced geological exploration method based on TBM construction of the present invention. DETAILED DESCRIPTION

[0075] The technical solutions in the embodiments of the present invention will be described clearly and completely below with reference to the accompanying drawings in the embodiments of the present invention.

[0076] Example

[0077] See also Figures 1-8 The present invention provides a technical solution: a multi-source collaborative advanced geological exploration method based on TBM construction, comprising the following steps:

[0078] S1. Conduct directional long-distance exploration before TBM starts to obtain long-distance axial geological data along the TBM excavation direction;

[0079] S2. During TBM excavation, branch short exploration is carried out using the equipment's built-in advanced geological drilling rig;

[0080] S3. Conduct transient electromagnetic surveys in long and short boreholes to generate a 30m radial three-dimensional geological profile.

[0081] S4. Compare drill cuttings, electromagnetic anomalies, and drilling parameters through a data fusion platform to identify water-rich areas and structural fracture zones;

[0082] S5. Dynamically update the BIM geological model. When the three results in S1-S4 are consistent, excavation is allowed. If the three results are different, the emergency plan is activated.

[0083] In this embodiment, specifically, the directional long exploration holes include 1# exploration hole (also a geophysical exploration hole), 2# exploration hole and 3# exploration hole.

[0084] In this embodiment, specifically: in S1, directional long exploration adopts φ89mm screw motor for directional drilling, the drilling depth is ≥600m, and the final hole diameter is φ94mm;

[0085] Among them, it includes secondary drilling structure and grouting casing, the grouting casing is φ127mm, and the pressure resistance is ≥1.5MPa;

[0086] In this embodiment, specifically: a secondary drilling structure (φ153mm hole expanded to 11m and then a φ127mm grouting casing is inserted), and a grouting pressure ≥2MPa is used to ensure the stability of the hole wall.

[0087] In this embodiment, specifically: the directional long-range drilling uses a hole-opening instrument to perform hole-opening orientation, and its trajectory control accuracy is: the error range of the hole-opening inclination angle and azimuth angle does not exceed ±0.5°.

[0088] In this embodiment, specifically: the deviation between the drilling trajectory and the set value is no more than 3m, and a cable drill pipe is used to realize the return of real-time inclination measurement data every 3m.

[0089] In this embodiment, specifically: in S2, the branch short probe is integrated into the hydraulic drilling rig of the TBM saddle frame, and the drilling depth is ≤100m;

[0090] Among them, the branch short exploration drilling distance is 100m / time, the excavation distance is 70m, and a 30m advance safety distance is left. The drilling arrangement is 3 holes evenly distributed along the arch in the range of -90° to 90°, cross-verified, and the elevation angle is 0° to 15°.

[0091] In this embodiment, specifically: in S3, the in-hole transient electromagnetic detection uses a mining transient electromagnetic instrument with an outer diameter of φ73mm;

[0092] It includes a three-component measurement system with a radial detection radius of 30m and supports 500m hole depth push.

[0093] In this embodiment, specifically: the three-component transient electromagnetic uses a YCSZ-DFS-009 probe (outer diameter 73mm) to achieve three-component measurement of the X, Y, and Z axes in the borehole, and cooperates with the convolutional neural network inversion algorithm to achieve a resistivity resolution of 0.5m.

[0094] In this embodiment, specifically: the in-hole transient electromagnetic detection adopts a convolutional neural network inversion algorithm, the resistivity and resolution are both ≤0.5m, the low-resistance anomaly threshold is set to ρ<15Ω·m, and the data collection interval is ≤1m.

[0095] In this embodiment, specifically: in-hole transient electromagnetic detection can achieve three-dimensional geological imaging within a radial range of 30m from the borehole, determine low-resistance anomalies (ρ < 15Ω·m) in real time and locate their spatial orientation, and support continuous detection of deep holes (500m level).

[0096] In this embodiment, specifically: in S4, the data fusion platform is used to integrate drilling parameters, electromagnetic signals and geological data in real time, thereby generating a three-dimensional risk warning model;

[0097] The data fusion processing in the data fusion platform includes the following steps:

[0098] A1. Data collection;

[0099] A2. Data processing;

[0100] A3. Decision analysis;

[0101] A4. Dynamic verification and model update;

[0102] A5. Dynamic update of BIM geological model;

[0103] A6. Engineering application optimization strategy;

[0104] A7. Computing resource configuration;

[0105] A8. Visual decision support;

[0106] Among them, in A1, data collection includes drilling data, geophysical data and construction data;

[0107] Drilling data: particle size analysis of rock cuttings samples, real-time parameters such as drilling torque, drilling speed, and pump pressure from directional long exploration holes (depth ≥ 600m); hole position deviation and rock cuttings mineral composition test data from branch short exploration holes (depth ≤ 100m);

[0108] Geophysical data: three-component (X, Y, Z axis) resistivity data and decay time series of in-hole transient electromagnetic detection, and low-resistance anomaly distribution within a radial range of 30 m (threshold < 15);

[0109] Construction data: TBM operating parameters such as excavation speed, cutterhead speed, shield pressure, and auxiliary data such as grouting pressure and casing pressure resistance (≥1.5MPa);

[0110] Among them, in A2, data processing includes preprocessing and feature extraction;

[0111] Preprocessing: Sliding window filtering is used to remove impulse noise from drilling parameters, interpolation algorithms are used to fill missing values ​​in electromagnetic data, and cuttings composition is standardized (including quartz content and water content quantification).

[0112] Feature extraction: The spatial coordinates (three-dimensional azimuth, radial distance) and morphological characteristics (anomaly size and continuity) of low-resistance anomalies are extracted from electromagnetic signals, and mutation points (including sudden increase in torque and sudden drop in pump pressure) are extracted from drilling parameters as characteristics of the structural fracture zone.

[0113] In this embodiment, specifically: in A3, a "3D risk warning model" is constructed, integrating drilling hole trajectories, geophysical anomalies, and construction parameter thresholds. A visual risk cloud map is generated through spatial interpolation, marking high-risk areas such as water-rich areas and faults. The decision analysis specifically includes:

[0114] B1. Core Fusion Algorithms and Technologies:

[0115] Using multi-source data spatiotemporal registration technology, with the TBM face as the coordinate origin, a unified geographic reference system (e.g., WGS84) is established. The directional long borehole trajectory (error ≤ 3m), the fan-shaped distribution of branch short boreholes (ranging from -90° to 90° from the vault), and the transient electromagnetic 3D profile (radial 30m) are spatially aligned to ensure data coordinate consistency.

[0116] In the temporal dimension, a clock synchronization mechanism was used to match drilling parameters (inclination data every 3 meters), electromagnetic detection (acquisition interval ≤ 1 meter), and TBM excavation progress (millimeter-level positioning) to form a spatiotemporally coupled dataset.

[0117] B2. Machine Learning Fusion Model:

[0118] Through random forest risk assessment: Input variables include electromagnetic resistivity (resolution ≤ 0.5m), cuttings moisture content, drilling torque mutation amplitude and other 20+ features. Through 500 decision tree ensemble learning, the output water inrush probability (accuracy 0.25%) is output;

[0119] Convolutional Neural Network (CNN) inversion: Using 3D CNN to automatically identify low-resistance anomalies in transient electromagnetic data, and combining it with drilling verification data to train the model, the positioning error of water-rich areas is less than 1m (traditional methods are 3-5m).

[0120] B3. Evidence-theoretic decision-making fusion:

[0121] Establish a confidence matrix for drilling, geophysical exploration, and construction parameters (the mud content of rock fragments obtained by directional long-range exploration is greater than 30%, and the confidence level is 0.8);

[0122] Transient electromagnetic detection shows a resistivity of <10Ω·m with a confidence level of 0.7;

[0123] When the TBM is drilling, the torque suddenly increases by 20%, with a confidence level of 0.6;

[0124] By integrating multi-source evidence through DS evidence theory, when the comprehensive confidence level is greater than 0.7, it is determined to be a high-risk area and the emergency plan is triggered.

[0125] In this embodiment, specifically: in A4, the specific steps of dynamic verification and model update include:

[0126] C1. After each detection cycle (70m excavation), the following data are automatically compared:

[0127] C2. The consistency of mineral composition between long-hole cuttings and short-hole cuttings;

[0128] C3, spatial consistency between the transient electromagnetic anomaly area and the structural belt revealed by drilling;

[0129] C4. Time correlation between the mutation point of drilling parameters and geophysical anomalies;

[0130] C5. If the consistency of the three-source data is less than 80%, start supplementary detection (for example, increase the number of short exploration holes).

[0131] In this embodiment, specifically: in A5, the BIM geological model is dynamically updated, and the integrated risk data is imported into the BIM model. The risk levels are marked with different colors (green ≤ 0.3% probability of water inrush, yellow 0.3%-1%, and red > 1%), and are linked to the TBM construction progress to achieve a closed-loop management of "detection-excavation-verification";

[0132] When a new structure is discovered (fault with a drop greater than 0.5m), the model prediction parameters are automatically corrected and the subsequent exploration plan is updated (adjusting the drilling angle for the next cycle of the long exploration hole);

[0133] In A6, an abnormal early warning mechanism for drilling data is established: when the drilling trajectory deviation is greater than 3m or the inclination data return is interrupted for more than 5 minutes, a manual verification is triggered;

[0134] Verification of the validity of geophysical data: The reliability of the data is determined by the resistivity repeatability (deviation ≤ 5%) of multiple detections in the same borehole.

[0135] In this embodiment, specifically: in A7, computing resources are configured using edge computing nodes to process real-time drilling parameters (inclination data every 3 meters), and cloud servers run CNN inversion and random forest models, achieving a hybrid architecture of "local rapid response + cloud-based deep analysis" to control data processing latency within 10 seconds;

[0136] In A8, a three-dimensional interactive interface has been developed to support construction personnel in viewing geological data slices at any location through touch operations. By clicking on abnormal areas, multi-dimensional information such as drilling cuttings photos, electromagnetic waveforms, and drilling parameter curves can be displayed to assist on-site decision-making.

[0137] In this embodiment, specifically: drilling parameters include pump pressure, torque, and drilling speed, etc., electromagnetic data include resistivity and decay time, etc., and geological information includes cuttings particle size and water inflow, etc.

[0138] In this embodiment, specifically: the water inrush risk probability is calculated by an AI model (based on a random forest algorithm). When the results of the three methods are consistent, the risk probability is ≤0.3%. Among them, the random forest algorithm is an existing technology and will not be described in detail here.

[0139] Experimental example

[0140] Directional long-distance exploration construction:

[0141] Drilling setting: Directional drilling is carried out according to the set azimuth, inclination and hole depth;

[0142] Drilling process:

[0143] First-level drilling: φ120mm PDC drill bit drilling to 11m;

[0144] Secondary hole expansion: 153mm drill bit expands the hole to 11m, inserts 127mm casing and injects grout (cement slurry water-cement ratio 0.8:1);

[0145] Directional drilling: φ89mm screw motor with cable drill pipe, inclination data is transmitted back every 3m, and the final hole trajectory deviation is less than 0.3%;

[0146] Transient electromagnetic detection:

[0147] In-hole transient electromagnetic detection is carried out by pushing the transmitting coil and receiving probe into the borehole through the drilling rig, and three-component measurement is performed point by point. The secondary field of the vertical component B along the drilling direction is used to analyze the possible low-resistance abnormal area around the borehole. The spatial orientation of the abnormality relative to the borehole is analyzed by the secondary field of two sets of horizontal components X and Y perpendicular to the borehole and orthogonal to each other (the X component is perpendicular to the hole and points to the right, and the Y component is perpendicular to the hole and points downward). Finally, a cylindrical detection area is formed with the borehole as the center and within a certain radial distance range. The excitation of the electromagnetic field and the data interaction and storage and recording are all completed in the borehole, and the three-component transient signal measurement is carried out by the mobile terminal outside the hole.

[0148] Branch short probe verification:

[0149] Equipment parameters: TBM comes with a hydraulic drill (torque 4000N·m), drilling diameter φ75mm;

[0150] Detection process:

[0151] Detect 100m, dig 70m, leave 30m ahead safety distance, and construct 3 boreholes;

[0152] Risk Decision-Making:

[0153] Data fusion: The platform integrates long-range resistivity data, short-range rock cuttings analysis, and TBM excavation parameters;

[0154] AI judges: The probability of water inrush risk is 0.25%, and generates a "green pass" instruction.

[0155] In summary, the present invention can effectively control risks. Referring to experimental examples, the water inrush accident rate has been reduced from the industry average of 2.7% to below 0.3%. The present invention can also greatly improve efficiency, shortening the comprehensive detection time from the traditional 24 hours to 5 hours to 8 hours, and reducing TBM downtime by 60%. At the same time, the present invention has achieved breakthroughs in precision, with the structure identification accuracy reaching 0.3m (traditional geophysical exploration methods are 3m-5m), and the positioning error of water-rich areas is less than 1m.

[0156] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A multi-source collaborative advanced geological exploration method based on TBM construction, characterized by: The following steps are involved: S1. Conduct directional long-distance exploration before TBM starts to obtain long-distance axial geological data along the TBM excavation direction; S2. During TBM excavation, branch short exploration is carried out using the equipment's built-in advanced geological drilling rig; S3. Conduct transient electromagnetic surveys in long and short boreholes to generate a 30m radial three-dimensional geological profile. S4. Compare drill cuttings, electromagnetic anomalies, and drilling parameters through a data fusion platform to identify water-rich areas and structural fracture zones; S5. Dynamically update the BIM geological model. When the three results in S1-S4 are consistent, excavation is allowed. If the three results are different, the emergency plan is activated.

2. The multi-source collaborative advanced geological exploration method based on TBM construction according to claim 1 is characterized by: In S1, directional long-range exploration uses a φ89mm screw motor for directional drilling, with a drilling depth of ≥600m and a final hole diameter of φ94mm; Among them, it includes secondary drilling structure and grouting casing, the grouting casing is φ127mm, and the pressure resistance is ≥1.5MPa; Among them, the directional long-distance drilling uses a hole-opening instrument for hole-opening orientation, and its trajectory control accuracy is: the error range of the hole-opening inclination and azimuth angle does not exceed ±0.5°.

3. The multi-source collaborative advanced geological exploration method based on TBM construction according to claim 1 is characterized by: In S2, the branch short probe is integrated into the hydraulic drilling rig of the TBM saddle, and the drilling depth is ≤100m; Among them, the branch short exploration drilling distance is 100m / time, the excavation distance is 70m, and a 30m advance safety distance is left. The drilling arrangement is 3 holes evenly distributed along the arch in the range of -90° to 90°, cross-verified, and the elevation angle is 0° to 15°.

4. The multi-source collaborative advanced geological exploration method based on TBM construction according to claim 1 is characterized in that: In S3, the in-hole transient electromagnetic detection uses a mining transient electromagnetic instrument with an outer diameter of φ73mm; It includes a three-component measurement system with a radial detection radius of 30m and supports 500m hole depth push.

5. The multi-source collaborative advanced geological exploration method based on TBM construction according to claim 4 is characterized in that: The in-hole transient electromagnetic detection adopts the convolutional neural network inversion algorithm, with the resistivity and resolution both ≤0.5m, the low-resistance anomaly threshold set to ρ <15Ω·m, and the data collection interval ≤1m.

6. The multi-source collaborative advanced geological exploration method based on TBM construction according to claim 1 is characterized in that: In S4, the data fusion platform is used to integrate drilling parameters, electromagnetic signals and geological data in real time to generate a three-dimensional risk warning model; The data fusion processing in the data fusion platform includes the following steps: A1. Data collection; A2. Data processing; A3. Decision analysis; A4. Dynamic verification and model update; A5. Dynamic update of BIM geological model; A6. Engineering application optimization strategy; A7. Computing resource configuration; A8. Visual decision support; Among them, in A1, data collection includes drilling data, geophysical data and construction data; Drilling data: particle size analysis of rock cuttings samples from directional long boreholes, multiple real-time parameters including drilling torque, drilling speed, and pump pressure, hole position deviation from branch short boreholes, and rock cuttings mineral composition test data; Geophysical data: three-component resistivity data and decay time series from in-hole transient electromagnetic detection, and low-resistance anomaly distribution within a radial range of 30 m; Construction data: TBM excavation speed, cutterhead speed, shield pressure equipment and other operating parameters, as well as auxiliary data such as grouting pressure and casing pressure resistance; Among them, in A2, data processing includes preprocessing and feature extraction; Preprocessing: Sliding window filtering is used to remove the impulse noise of drilling parameters, interpolation algorithm is used to fill the missing values ​​of electromagnetic data, and the cuttings composition is standardized and coded; Feature extraction: The spatial coordinates and morphological features of low-resistance anomalies are extracted from electromagnetic signals, and mutation points are extracted from drilling parameters as structural fracture zone features.

7. The multi-source collaborative advanced geological exploration method based on TBM construction according to claim 6 is characterized by: In A3, a "3D risk warning model" was constructed, integrating drilling hole trajectories, geophysical anomalies, and construction parameter thresholds. A visual risk cloud map was generated through spatial interpolation, marking multiple high-risk areas such as water-rich areas and faults. The decision analysis specifically includes: B1. Core Fusion Algorithms and Technologies: Using multi-source data spatiotemporal registration technology, with the TBM face as the coordinate origin, a unified geographic reference system is established to spatially align the directional long borehole trajectory, the fan-shaped distribution of branch short boreholes, and the transient electromagnetic 3D profile to ensure data coordinate consistency. In the temporal dimension, the drilling parameters, electromagnetic detection, and TBM excavation progress are matched through a clock synchronization mechanism to form a spatiotemporal coupled dataset. B2. Machine Learning Fusion Model: Through random forest risk assessment: the input variables include multiple features, and 500 decision trees are used for ensemble learning to output the probability of water inrush; Convolutional neural network inversion: 3D CNN is used to automatically identify the morphology of low-resistance anomalies based on transient electromagnetic data. The model is trained using drilling verification data to reduce the positioning error of water-rich areas to less than 1m. B3. Evidence-theoretic decision-making fusion: Establish confidence matrices for drilling, geophysical exploration, and construction parameters; Transient electromagnetic detection shows a resistivity of <10Ω·m with a confidence level of 0.7; When the TBM is drilling, the torque suddenly increases by 20%, with a confidence level of 0.6; By integrating multi-source evidence through DS evidence theory, when the comprehensive confidence level is greater than 0.7, it is determined to be a high-risk area and the emergency plan is triggered.

8. The multi-source collaborative advanced geological exploration method based on TBM construction according to claim 6 is characterized by: In A4, the specific steps of dynamic verification and model update include: C1. Each time a detection cycle is completed, the following data are automatically compared; C2. The consistency of mineral composition between long-hole cuttings and short-hole cuttings; C3, spatial consistency between the transient electromagnetic anomaly area and the structural belt revealed by drilling; C4. Time correlation between the mutation point of drilling parameters and geophysical anomalies; C5. When the consistency of the three-source data is less than 80%, start supplementary detection.

9. The multi-source collaborative advanced geological exploration method based on TBM construction according to claim 6, characterized in that: In A5, the BIM geological model is dynamically updated, integrating risk data into the BIM model, marking risk levels with different colors and linking them to the TBM construction progress, achieving a closed-loop management of "detection-excavation-verification"; When new structures are discovered, the model prediction parameters are automatically corrected and subsequent detection plans are updated; In A6, an abnormal early warning mechanism for drilling data is established: when the drilling trajectory deviation is greater than 3m or the inclination data return is interrupted for more than 5 minutes, a manual verification is triggered; Verification of the validity of geophysical data: The reliability of the data is determined by the repeatability of the resistivity of multiple detections in the same borehole.

10. The multi-source collaborative advanced geological exploration method based on TBM construction according to claim 6, characterized in that: In the A7, computing resources are configured using edge computing nodes to process real-time drilling parameters, while cloud servers run CNN inversion and random forest models, achieving a hybrid architecture of "local rapid response + cloud-based in-depth analysis" and keeping data processing latency within 10 seconds. In A8, a three-dimensional interactive interface has been developed to support construction personnel in viewing geological data slices at any location through touch operations. Clicking on abnormal areas can link and display multiple multi-dimensional information such as drilling cuttings photos, electromagnetic waveforms, and drilling parameter curves to assist on-site decision-making.

Images