Shield front karst fissure grouting plugging method based on feedforward recognition and directional regulation and control

Abstract

The invention discloses a shield front karst fissure grouting plugging method based on feedforward identification and directional regulation and control, which comprises the steps of front geological information acquisition, target area identification, grouting path planning, parameter adaptive regulation and control, directional grouting execution and plugging effect feedback. Feature fusion processing of geometric features of the cavity region from a time domain to a frequency domain is realized, geological types and features of the cavity region are analyzed in combination with data, and remarkable reduction of cavity three-dimensional coordinate positioning errors and remarkable improvement of boundary recognition precision are realized; in the drilling process, the boundary of the cavity is sensed in real time, synchronous grouting and air exhaust are achieved through automatic deviation correction of the posture adjusting mechanism and the double-through-hole design, air pressure fluctuation in the cavity is controlled within a safe and stable range, and the risk of cavity collapse caused by air pressure changes is effectively prevented.

Description

Technical Field

[0001] This invention relates to a method for grouting and sealing fissures, specifically a method for grouting and sealing karst fissures in front of a tunnel boring machine based on feedforward identification and directional control, belonging to the field of underground engineering geological disaster detection and control technology. Background Technology

[0002] During the construction and excavation of underground transportation tunnels, geological hazards such as karst fissures seriously affect the progress and safety of the project. However, traditional ground-penetrating radar detection has obvious shortcomings: it relies on manual analysis of reflected signals, and the accuracy of cavity identification is low in complex geological conditions such as water-rich strata and high-resistivity rock masses. It is easy to misidentify pipelines and fissures as cavities. Moreover, a single frequency cannot distinguish the differences in electromagnetic response between cavities and the surrounding medium, resulting in many noise points in the 3D model of cavities and modeling errors exceeding 20%. At the same time, the treatment methods for cavities and fissures lack systematic and differentiated measures. They are mostly operated according to uniform standards, which are not adaptable enough and make it difficult to ensure the treatment quality under different geological conditions.

[0003] Small drilling operations lack safety when filling cavities: the drilling path and filling point rely heavily on preset paths and manual remote control. In complex geological conditions, sudden changes in drilling resistance can easily cause the drill bit to deviate. Furthermore, because the attitude deviation cannot be detected in real time, the success rate of accurately reaching the center of the cavity is extremely low. This not only increases the cost of secondary drilling but may also induce cavity collapse due to repeated disturbances. The problem is even more prominent in sensitive geological conditions such as extremely thin coal seams and soft soil strata.

[0004] Mainstream filling methods have several problems: with only a single grouting channel, it is difficult to expel air from cavities, which can lead to collapse and media disturbance due to unstable air pressure, and also leave air to form filling blind spots, affecting quality; there is no quality inspection after the initial filling of the tunnel boring machine, and the working face often collapses due to insufficient filling; in addition, the electromagnetic properties of traditional cement-based materials are not much different from those of the rock mass, making it difficult to distinguish between filled and unfilled areas using conventional ground-penetrating radar during secondary inspection, resulting in low repair positioning accuracy and poor inspection effect.

[0005] Ground-penetrating radar (GPR) is an advanced geophysical exploration technology. Its principle is to use pulse waves to non-destructively probe geological structures. Due to its excellent performance and wide detection range, it is widely used in archaeology, geological exploration, and transportation engineering. GPR can effectively identify cavities and fissures ahead during tunnel boring machine (TBM) excavation. By detecting underground areas including soil, rock, and potential cavities and fissures, it provides a way to further understand the geological structure and characteristics of the project's front end. The principle is that different media (soil, rock, and cavities) have significantly different characteristics in reflecting, refraction, and scattering electromagnetic waves. When electromagnetic waves encounter these media layers, especially abnormal structures such as cavities, some electromagnetic waves are reflected and received by the detector. The GPR system captures these reflected electromagnetic waves and meticulously analyzes the amplitude and frequency changes of the received waves, as well as the distribution characteristics of abnormal signals, thereby accurately inferring the location, size, and surrounding medium properties and structural state of cavities beneath the road.

[0006] In existing technologies, such as the full-face grouting shield tunneling method for water-rich rock strata under sensitive structures disclosed in CN107642364A, a full-face grouting process is used. Grout is injected into a soil chamber and pressure is maintained to seal bedrock fissures. This method focuses on using the shield tunneling system itself for large-scale grouting, but it lacks the ability to accurately identify karst fissures in front of the shield, plan individualized grouting paths, and dynamically control the grouting process, making it difficult to achieve differentiated and precise grouting for fissures of different shapes. Another example is the double-layer pipe fixed-point grouting and sealing device for shield tunnels in karst environments disclosed in CN117514201A, which provides a double-layer pipe fixed-point grouting device. Its inner and outer pipes work together to achieve integrated drilling and grouting and fixed-point grouting. While the device improves the mobility of grouting, its grouting strategy still largely depends on preset paths and manual intervention, lacking intelligent planning and adaptive control capabilities based on accurate geological exploration ahead. In particular, when dealing with complex and extended fracture networks, the success rate of a single grouting and the uniformity of coverage need to be improved.

[0007] In summary, existing karst fissure grouting technology in front of tunnel boring machines has the following limitations: inaccurate analysis of geological exploration data makes it difficult to reconstruct the fine three-dimensional structure of fissures; poor adaptability of grouting path planning to geological conditions, easily leading to grouting blind spots or stress concentrations; poor matching between grouting parameters (grout type, pressure, flow rate) and the dynamically changing grouting environment; and a lack of effective real-time feedback and quantitative evaluation methods for grouting effects. Although there are relevant research results on underground cavity detection and filling, there are still shortcomings in the advanced detection, cavity filling, and systematic filling schemes for underground tunnel cavities in the unexcavated working face. Summary of the Invention

[0008] The purpose of this invention is to provide a method for grouting and sealing karst fissures in front of a tunnel boring machine based on feedforward identification and directional control, in order to solve the problems of inconvenience, low detection accuracy and insufficient safety of existing underground tunnel karst fissure detection methods.

[0009] This invention achieves the above objective through the following technical solution: a method for grouting and sealing karst fissures in front of a tunnel boring machine based on feedforward identification and directional control, comprising the following steps:

[0010] S1. Obtaining geological information ahead: By sensing engineering parameters of the pilot borehole and scanning the strata ahead of the shield tunnel using ground-penetrating radar or ultrasonic waves, the mechanical and physical properties of the strata are obtained.

[0011] S2. Target area identification: Based on the morphological characteristics of karst fissures, study the development characteristics and extension properties of karst fissures, and determine the three-dimensional morphology and filling status of karst caves.

[0012] S3. Grouting path planning: Based on the detection results, the area to be grouted is automatically identified, and the optimal grouting path and hole layout scheme are generated according to the adaptive prediction algorithm of karst fracture configuration.

[0013] S4. Parameter adaptive control: Matches grout type, concentration, grouting pressure and flow rate with the geological conditions of the target area to control the directional injection device;

[0014] S5. Directional grouting is carried out, and differentiated grouting is performed along the planned path. During this process, the sealing effect is judged by the changes in grouting reaction force, return flow rate or permeability, and the relevant grouting parameters are dynamically adjusted.

[0015] S6. Sealing effect feedback: Magnetic tracer is added evenly during the grouting process. After detection, a three-dimensional grouting model is generated and compared with the cavity model to determine the grouting quality.

[0016] As a further aspect of the present invention: In S2, when identifying the three-dimensional morphology and filling state of the karst cave, the three-dimensional spatial geometric center coordinates and boundary range of the cave are identified and output through multi-dimensional signal verification combined with three-dimensional modeling constraints, specifically including:

[0017] The system collects multi-dimensional detection information from ground-penetrating radar, and performs spatiotemporal matching of multi-dimensional signals through time-domain and frequency-domain feature fusion algorithms to generate a weighted cavity feature map. The weights are dynamically adjusted based on the amplitude of the reflected waves.

[0018] The geometric center of the cavity is determined by combining the minimum circumscribed sphere with the axis-aligned bounding box. The coordinate parameters of the minimum three-dimensional cube containing the cavity and the radius parameters of the sphere are output. A spatial coordinate system with the ground-penetrating radar as the origin is established and used as a reference to control the drilling trajectory of the drill bit.

[0019] As a further aspect of the present invention: In S3, the grouting path and hole arrangement are optimized using an adaptive prediction algorithm based on the fracture development law, extension characteristics, and three-dimensional cavity configuration, specifically including:

[0020] By reconstructing ground-penetrating radar images, the fracture orientation, dip angle, opening and extension length, as well as the three-dimensional coordinates, volume, wall thickness and connectivity topology of cavities are quantitatively extracted.

[0021] By combining ground stress data, the direction of crack propagation is predicted using a fracture finite element model, and the key support points of the cavity are identified through cavity stability analysis.

[0022] A hole position constraint model is constructed to avoid support points and stress concentration areas at the crack tips. The path follows the crack direction, and the hole spacing is dynamically adjusted according to density. After adaptive iterative optimization, with the goal of grout coverage ≥90% and disturbance stress increment ≤0.1MPa, the final path and hole position coordinates are output.

[0023] As a further aspect of the present invention: In S4, the parameter adaptive control is based on the improved FCM algorithm. According to the geological type discrimination system, different strata types and water-rich states are characterized. Then, according to the slurry decision system, the optimal slurry ratio and setting time parameters are recommended in combination with the geological type and related factors to determine the optimal ratio.

[0024] As a further aspect of the present invention: In S5, during differentiated grouting, a small drilling machine is used for path drilling, and the drilling is based on underground drilling path planning. The drilling direction of the small drilling machine is pre-planned and corrected in real time according to the acquired potential fractures, cavities and stratum attribute data to ensure the smooth progress of the drilling process. The endpoint is planned according to the safe grouting point given by the prediction algorithm to ensure the safety and reliability of the grouting work of the small drilling machine.

[0025] As a further aspect of the present invention: In S5, during directional grouting, by analyzing geological characteristics and geological composition conditions, a multi-factor decision-making method is adopted to make a reasonable matching decision on the grouting composition ratio, so as to ensure that the injected grout can play a role in reinforcement and filling without affecting or damaging the rock strata in the area. The multi-factor decision-making method includes, but is not limited to, extraction of geological and physical characteristics of the target area, analysis of strata lithology and composition, control of engineering functional requirements, and consideration of construction constraints and environmental compatibility requirements.

[0026] As a further aspect of the present invention: In S6, the sealing effect feedback adopts multiple sensors to monitor the grouting effect in real time and adjust the parameters in a closed loop. The sensors used include, but are not limited to, pressure sensors, seepage rate sensors and backflow detectors installed at the front end of the small drilling machine. In addition, an exhaust pipe is installed at the front end of the small drilling machine to facilitate the discharge of gas in the cracks and cavities during grouting, and to prevent the collapse of cavities and cracks due to pressure imbalance in the space.

[0027] As a further aspect of the present invention: In S6, the magnetic tracer is added under the following conditions: it is added uniformly without affecting the properties of the slurry itself, so that the geometric morphology of the filler is easier to detect after the slurry solidifies, which is conducive to the subsequent generation of a three-dimensional model of the filler and comparison with the three-dimensional model of the spatial geometry detected in the cavity and crack detection stage to determine the filling quality of the cavity area; at the same time, the magnetic tracer remaining inside is also conducive to the subsequent maintenance work of cavities and cracks.

[0028] As a further aspect of the present invention, the method also includes fusing feedforward detection data with historical tunneling data to construct a geological-grouting response database, which is then used to optimize the grouting strategy for subsequent tunneling sections; and storing fracture and cavity models in a cavity stress analysis and grouting point decision database, with the processor retrieving models from the database for training and analysis during idle periods, so that the performance of the grouting decision model can continuously improve as the amount of tunneling work increases.

[0029] The beneficial effects of this invention are:

[0030] 1) This invention converts ground-penetrating radar data into a spatial three-dimensional model, realizes the fusion processing of the geometric features of the cavity area from the time domain to the frequency domain, and combines data analysis of the geological type and characteristics of the cavity area, thereby achieving a significant reduction in the three-dimensional coordinate positioning error of the cavity and a significant improvement in the boundary identification accuracy.

[0031] 2) This invention effectively solves the positioning deviation problem caused by single signal in the prior art, significantly reduces drilling path deviation, and avoids repeated drilling or missed drilling. The path planning system analyzes the cavity and geological conditions in the path and autonomously plans the drilling path. The small drilling machine carries out drilling work according to the path planned by the system. The specially designed drill bit at the front end integrates a sound wave sensor and a gas pressure monitoring device to sense the cavity boundary in real time during the drilling process and automatically correct the deviation through the attitude adjustment mechanism.

[0032] 3) The dual-hole design of this invention enables simultaneous grouting and air extraction, controlling the air pressure fluctuations within the hole within a safe and stable range, effectively preventing the risk of cavity collapse caused by air pressure changes. Compared with traditional grouting processes, the collapse accident rate is significantly reduced. The front-end and mid-section ground radar of the tunnel boring machine work in conjunction with the magnetic gradient instrument to form a closed-loop control of "synchronous exploration and filling". The magnetic tracer particles incorporated into the filling material significantly enhance the recognition signal strength of the magnetic gradient instrument for the filling material, greatly improving the filling success rate and significantly reducing the risk of subsequent ground subsidence.

[0033] 4) This invention deeply integrates modules such as multi-factor decision-making, three-dimensional modeling, intelligent drilling, and air pressure control to form a complete automated system. By adjusting the parameters of the geological radar and the drilling strategy, it can adapt to different geological environments and borehole sizes. Its application scope covers multiple fields such as urban underground engineering, tunnel construction, and mine restoration, and it has significant engineering practical value and economic benefits. Attached Figure Description

[0034] Figure 1 This is the overall flowchart of the present invention;

[0035] Figure 2 This is a schematic diagram of an underground path planning system.

[0036] Figure 3 Intelligent decision-making flowchart for slurry composition;

[0037] Figure 4 This is a flowchart illustrating the key steps and intelligent feedback mechanism of grouting. Detailed Implementation

[0038] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0039] Example 1: A method for grouting and sealing karst fissures in front of a tunnel boring machine based on feedforward identification and directional control. This method includes the following steps:

[0040] S1. Obtaining geological information ahead: By sensing engineering parameters of the pilot borehole and scanning the strata ahead of the shield tunnel using ground-penetrating radar or ultrasonic waves, the mechanical and physical properties of the strata are obtained.

[0041] During the construction of the pilot borehole, inversion sensing is performed based on drilling engineering data. Borehole geophysical exploration tests are conducted using methods such as borehole ground-penetrating radar and acoustic waves to obtain comprehensive stratigraphic data, identify potential geological hazards such as karst fissures, and acquire data on potential karst fissures and stratigraphic properties. For convenience, this invention uses ground-penetrating radar electromagnetic scanning for fine detection. By leveraging the underground propagation of high-frequency electromagnetic waves and the differences in electronic characteristics of the underground medium, scanning imaging is achieved, capturing potential fissures, cavities, and geological feature data within the scanning range at the front of the tunneling face. This provides a basis for subsequent trajectory planning and grouting mix design. Based on this method, other detection methods can also be used to meet the corresponding directional identification and sealing requirements.

[0042] S2. Target area identification: Based on the morphological characteristics of karst fissures, study the development characteristics and extension properties of karst fissures, and determine the three-dimensional morphology and filling status of karst caves.

[0043] Based on the detection results, the system automatically identifies cavities and fissures that require grouting. After standardizing and denoising the collected cavity information, the system extracts the three-dimensional features of the cavity area, converts the original detected signal into a quantifiable grouting target area, and performs three-dimensional imaging of the area. The system then uses an identification model to determine the overall type of the cavity from local features, obtaining key data such as cavity type, stress distribution, and stability of the surrounding soil and rock layers. Multiple factors are combined to determine the type of cavity and fissure.

[0044] S3. Grouting path planning: Based on the detection results, the area to be grouted is automatically identified, and the optimal grouting path and hole layout scheme are generated according to the adaptive prediction algorithm of karst fracture configuration.

[0045] By determining the rock strata composition based on the detection results and combining the distribution of cavities and fissures, the drilling path of the small drilling grouting mechanism is planned to minimize the damage to the stability of the rock strata during the drilling process. At the same time, the safe grouting point decision is made based on the type of cavity and the stress condition, providing a theoretical basis for the endpoint position of the small drilling grouting mechanism and minimizing the risk of accidental collapse caused by the destruction of the mechanical stability of cavities and fissures during the tunneling and grouting process.

[0046] S4. Parameter adaptive control: Matches grout type, concentration, grouting pressure and flow rate with the geological conditions of the target area to control the directional injection device;

[0047] Based on the exploration results, the geological type of the target area was identified, including but not limited to loose soil layers, fractured rock masses, and cavitary strata. Key geological parameters are as follows: loose soil layers have a porosity of 25%-40% and a permeability coefficient of 10. -3 -10 -4 The grout has a permeability coefficient of 10 m / s, meaning the grout tends to flow out along large pores during filling, requiring rapid setting and grouting under low pressure. -6 -10 -8 The permeability of this area is low, and ultrafine grout and appropriate pressurization are required to improve filling efficiency during filling; the permeability coefficient of fractured rock masses is typically 10 m / s. -4 -10 -6 m / s, the grout is prone to flow away along the dominant fractures during filling, so a medium viscosity grout is required and the filling pressure is controlled at the same time; cavitary strata require high flow rate and fast setting and compaction filling; the operation target is determined according to the permeability, pore structure and compressive strength of different geological areas, such as reinforcement, water plugging, filling, etc., and the system selects appropriate grout components for mixing according to the stratum characteristics and operation purpose;

[0048] S5. Directional grouting is carried out, and differentiated grouting is performed along the planned path. During this process, the sealing effect is judged by the changes in grouting reaction force, return flow rate or permeability, and the relevant grouting parameters are dynamically adjusted.

[0049] The system controls the directional injection device to perform differentiated grouting operations according to the planned path. Simultaneously, it inputs safe grouting points, plans precise routes, and controls adaptive parameters to achieve grouting in designated target areas for the small drilling machine. Specifically, it analyzes the stress conditions of the cavity area based on the output 3D image and determines safe grouting points. Based on the geological distribution detected by ground-penetrating radar, it automatically generates a planned path between the starting point and safe grouting points of the small drilling machine. The system controls the tunneling and grouting work efficiently and safely based on the optimal path determined by the system. Adaptive parameter control selects appropriate grouting materials based on the geological characteristics determined by the system, controls the appropriate grouting pressure, and adjusts the grouting pressure based on the feedback air pressure inside the cavity to ensure stable grouting operations.

[0050] S6. Sealing effect feedback: Magnetic tracer is added evenly during the grouting process. After detection, a three-dimensional grouting model is generated and compared with the hole model to determine the grouting quality.

[0051] The sealing effect is judged and parameters are dynamically adjusted by measuring changes in grouting reaction force, return flow rate, or permeability. Specifically, to address the problem that single evaluation indicators are prone to misjudgment and omission during the filling process, a multi-factor collaborative analysis system is proposed to judge the grouting completion degree. Based on the detection of changes in grouting reaction force, return flow rate, permeability, and cavity air pressure, the grouting completion degree of the cavity is analyzed. Based on different geological characteristics, thresholds for each factor are set for cross-analysis, and the grouting task is terminated after the preset index thresholds are met.

[0052] Post-sealing inspection involves uniformly adding a special testing agent to the grout. After inspection, a three-dimensional model of the grout is generated based on the volume occupied by the special agent and compared with the detected void model to determine the grouting quality. Specifically, the grouting distribution is visualized and quantitatively evaluated through special testing agent tracing, three-dimensional model reconstruction, and comparison of void and crack models. Taking a magnetic tracer as an example, the tracer is thoroughly mixed with the filling grout before filling. After the grout is filled, an electromagnetic gradient meter is used to scan and image the filled area, and the image is compared with the initial three-dimensional images of voids and cracks to further determine the filling quality and apply it to later maintenance.

[0053] Example 2, in addition to all the technical features included in Example 1, also includes:

[0054] When identifying the three-dimensional morphology and filling status of karst caves, multi-dimensional signal verification combined with three-dimensional modeling constraints is used to identify and output the three-dimensional spatial geometric center coordinates and boundary range of the caves, specifically including:

[0055] The system collects multi-dimensional detection information from ground-penetrating radar, and performs spatiotemporal matching of multi-dimensional signals through time-domain and frequency-domain feature fusion algorithms to generate a weighted cavity feature map. The weights are dynamically adjusted based on the amplitude of the reflected waves.

[0056] The geometric center of the cavity is determined by combining the minimum circumscribed sphere with the axis-aligned bounding box. The coordinate parameters of the minimum three-dimensional cube containing the cavity and the radius parameters of the sphere are output. A spatial coordinate system with the ground-penetrating radar as the origin is established and used as a reference to control the drilling trajectory of the drill bit.

[0057] The grouting path and borehole layout are optimized using an adaptive prediction algorithm based on fracture development patterns, extension characteristics, and three-dimensional cavity configurations. Specifically, this includes:

[0058] By reconstructing ground-penetrating radar images, the fracture orientation, dip angle, opening and extension length, as well as the three-dimensional coordinates, volume, wall thickness and connectivity topology of cavities are quantitatively extracted.

[0059] By combining ground stress data, the direction of crack propagation is predicted using a fracture finite element model, and the key support points of the cavity are identified through cavity stability analysis.

[0060] A hole position constraint model is constructed to avoid support points and stress concentration areas at the crack tips. The path follows the crack direction, and the hole spacing is dynamically adjusted according to density. After adaptive iterative optimization, with the goal of grout coverage ≥90% and disturbance stress increment ≤0.1MPa, the final path and hole position coordinates are output.

[0061] The parameter adaptive control is based on the improved FCM algorithm. According to the geological type discrimination system, different strata types and water-rich states are characterized. Then, according to the slurry decision system, the optimal slurry ratio and setting time parameters are recommended in combination with geological type and related factors to determine the optimal ratio.

[0062] During differentiated grouting, a small drilling rig is used for path drilling. The drilling is based on underground drilling path planning. The drilling direction of the small drilling rig is pre-planned and corrected in real time according to the acquired potential fractures, cavities and stratum attribute data to ensure the smooth progress of the drilling process. The endpoint is planned according to the safe grouting point given by the prediction algorithm to ensure the safety and reliability of the grouting work of the small drilling rig.

[0063] During directional grouting, by analyzing geological characteristics and geological composition conditions, a multi-factor decision-making approach is adopted to make reasonable matching decisions on the grouting component ratio, so as to ensure that the injected grout can play a role in reinforcement and filling without affecting or damaging the rock strata in the area. The multi-factor decision-making approach includes, but is limited to, the extraction of geological and physical characteristics of the target area, analysis of strata lithology and composition, control of engineering functional requirements, and consideration of construction constraints and environmental compatibility requirements.

[0064] The sealing effect feedback uses multiple sensors to monitor the grouting effect in real time and adjust parameters in a closed loop. The sensors used include, but are not limited to, pressure sensors, seepage rate sensors and backflow detectors installed at the front end of the small drilling machine. The small drilling machine is equipped with an exhaust pipe at the front end to facilitate the discharge of gas in the cracks and cavities during grouting, preventing the collapse of cavities and cracks due to pressure imbalance in the space.

[0065] The magnetic tracer is added under the following conditions: it is added uniformly without affecting the properties of the slurry itself, so that the geometric morphology of the filler is easier to detect after the slurry solidifies. This facilitates the subsequent generation of a three-dimensional model of the filler, and the similarity is compared with the three-dimensional model of the spatial geometry detected in the cavity and crack detection stage to determine the filling quality of the cavity area. At the same time, the magnetic tracer remaining inside is also beneficial to the subsequent maintenance work of cavities and cracks.

[0066] This method also includes fusing feedforward detection data with historical tunneling data to construct a geological-grouting response database, which is used to optimize grouting strategies for subsequent tunneling sections; storing fracture and cavity models in a cavity stress analysis and grouting point decision database, and retrieving models from the database for training and analysis when the processor is idle, so that the performance of the grouting decision model can be continuously improved as the amount of tunneling work increases.

[0067] Example 3, as Figures 1 to 4 As shown, a method for grouting and sealing karst fissures in front of a tunnel boring machine based on feedforward identification and directional control includes the following steps:

[0068] Step 1: Geological exploration at the working face. Specifically, a ground-penetrating radar is used to scan the strata in front of the tunnel boring machine to obtain potential fractures, cavities and strata properties. The scanning imaging purpose is achieved by using the underground propagation of high-frequency electromagnetic waves and the difference in electronic characteristics of the underground medium. This captures the potential fractures, cavities and geological feature data within the scanning range at the working face of the tunnel boring machine, providing a basis for subsequent trajectory planning and grouting mix design.

[0069] In practical applications, the performance of the ground-penetrating radar (GPR) is improved by installing a ground-penetrating radar, a high-frequency electromagnetic wave transmitter, and a receiver at the front end of the tunnel boring machine (TBM). The GPR continuously transmits and receives electromagnetic waves to obtain the two-dimensional morphology of cavities and fissures, as well as the geological features at the front end of the working face. After obtaining the information, the geological information is filtered by median filtering to remove noise and correct the geological information. Then, the reflection feature recognition technology is used to determine the primary outline of the target based on the "in-phase axis morphology" of the radar profile. Then, the abnormal reflection area is extracted by amplitude threshold interpolation. Combined with cross-section verification, the geometric features and coordinate positions of cavities and fissures are finally obtained. The volume of the target and the surrounding geological type are estimated by the reflection signal, providing a basis for subsequent material selection and material usage.

[0070] Step 2, Target Area Identification, specifically: Based on the detection results, automatically identify cavities and fissures requiring grouting; after standardizing and denoising the collected cavity information, extract the three-dimensional features of the cavity area, convert the detected raw signal into a quantifiable grouting target area, and perform three-dimensional imaging of the area. The identification model then judges the overall type from local features to obtain key data such as cavity type, stress distribution, and stability of surrounding soil and rock layers. Multiple factors are combined to determine the type of cavity and fissure. The data standardization process includes three parts: amplitude normalization, coordinate and scale unification, and fine noise suppression. Amplitude normalization uses min-max scaling to eliminate the energy difference between "strong near-surface reflection" and "weak deep signal," as shown in the following formula:

[0071] ;

[0072] Where x is the amplitude value of the original ground-penetrating radar reflection signal; xmin is the minimum value among all acquired amplitude data; and xmax is the maximum value among all acquired amplitude data.

[0073] Subsequently, an edge enhancement algorithm was used to highlight the reflection boundaries of strong reflections and shadowed areas within the voids, as well as the coaxial discontinuity of cracks. Specifically, the Laplacian operator was used to enhance details and highlight cracks, as shown in the following formula: ; ;

[0074] Among them, I enh (x, y) represents the pixel intensity value at coordinates (x, y) after enhancement by the Laplacian operator; k is the enhancement coefficient.

[0075] Geological information correction includes topographic correction and time-depth conversion. Topographic correction eliminates the influence of surface undulations on the imaging of subsurface interfaces, while time-depth conversion converts time-domain data into depth-domain data.

[0076] Step 3: Visualization. After data processing, further visualization is performed. Taking continuous fractures as an example, discrete profile data is interpolated, and three-dimensional features are extracted and quantified to provide a basis for subsequent type judgment and grouting necessity. These features need to cover three main categories: geometric features, physical features, and spatial distribution features. Geometric feature extraction is based on voxel counting (grid model) or triangulation area integration (surface model) of a three-dimensional mesh model; geometric parameters are automatically calculated using Python libraries. Physical feature extraction is based on statistical analysis of radar signals from "void regions" in the three-dimensional data volume; continuity is quantified using the signal continuity index. Spatial distribution feature extraction is based on fracture orientation or dip angle extracted using Hough transform; the distance between the fracture and engineering elements is calculated using spatial coordinates, with the relevant formulas as follows: Voxel counting: ; Where V is the total volume of the fracture (unit: m³); N 有效 The effective number of voxels within the fracture region; △ x , △ y , △ z Let be the side length of the voxel in the engineering coordinate system (x, y, z).

[0077] Key geometric parameters of the crack: length: ; Fractal dip angle: n=(A,B,C), ;

[0078] Radar signal statistics in the fracture region: Mean amplitude: ; Amplitude standard deviation: ;

[0079] The continuity of the phase axis is quantified by calculating the signal correlation between adjacent radar channels within the crack depth range:

[0080] Signal continuity index: ;

[0081] Where, ρ t The correlation coefficient between adjacent channels at depth point t is calculated using the following formula: ;

[0082] In a two-dimensional profile of a ground-penetrating radar (GPR), fracture reflection appears as a straight line. The Hough transform converts the rectangular coordinates to polar coordinates. The Hough transform formula is: ;

[0083] Furthermore, the formula for calculating the planar density of the fracture is as follows: ;

[0084] The three-dimensional features of holes and cracks are extracted using the above series of formulas.

[0085] Step 4: Stress analysis. A PINN combined with a GNN prediction system is used. The PINN physical information neural network is used to fuse physical constraints and data-driven approaches into the model. Then, the GNN is used to predict the topological characteristics and evolution of the fracture and cavity models, thereby completing the stress analysis of the three-dimensional models of the cavity and fracture.

[0086] Specifically, PINN directly embeds the residuals of partial differential equations into the neural network training process, forcing the model to follow physical laws, while simultaneously optimizing parameters using measured data. The model constructs a twin model based on the input parameters such as the three-dimensional geometric parameters of cavities and fissures, the physical parameters of the surrounding rock mass, and the initial stress field, and designs the neural network architecture to calculate the stress components of the output layer. A loss function is used to calculate the stress distribution, and finally, the results, including stress contour maps, safety factors, and displacement fields, are output. The relevant formulas are as follows: Stress balance equation: ; Stress concentration factor: ;

[0087] Considering the varying stability of rock masses under different geological characteristics, a safety factor for rock mass stability also needs to be taken into account. The calculation formula for determining whether a rock mass will become unbalanced based on the maximum shear stress and the shear strength of the surrounding rock mass is as follows: ,in ;

[0088] Where c is the rock mass cohesion, Φ is the internal friction angle of the rock mass, and σ1 and σ3 are the maximum and minimum resistance.

[0089] A GNN prediction system was used for feature extraction and topology analysis. Local and global features of voids and fractures were extracted through two layers of GCN. Then, the node-level features were transformed into image-level features through global average pooling. The stability index was calculated by combining the PINN results to obtain the stable grouting area and predict the evolution direction of voids and fractures.

[0090] Step 5: Grouting Path Planning. Based on the 3D geological model, target area characteristics, and obstacle data, dynamic path planning is performed. Specifically, DRL is used to model the dynamic path adjustment as a Markov decision process. By incorporating a scoring mechanism of states, actions, and rewards, the scoring mechanism is continuously optimized to encourage the model to continuously converge and optimize actions. The states include real-time geological parameters, grout status, and equipment status. The actions include hole location selection and borehole depth adjustment. The rewards include grouting effect score, cost consumption score, and safety risk score. Finally, combined with real-time manual monitoring data, the path decision-making capability is continuously optimized.

[0091] Step Six: Intelligent Grouting Selection. This step employs multi-parameter fusion, classifying geological types based on core physical parameters, retrieving electrical constants from ground-penetrating radar 3D data, and determining porosity and permeability coefficients through borehole sampling. It also uses the synergistic determination of water-bearing state based on physical parameters and radar signals, controlling the dynamic mapping of grout parameters according to geological conditions. When anomalies occur during actual grouting, such as excessive backflow or a sudden increase in grouting reaction force, the system automatically fine-tunes parameters to ensure suitability. Specifically, the intelligent grout selection decision-making process uses LightGBM+NSGA-II+k-NN. LightGBM handles inputs related to geological type, water-bearing state, porosity, permeability coefficient, and engineering requirements, and outputs basic grout parameters. NSGA-II optimizes the grout mix ratio to achieve optimal performance after mixing, ensuring rapid convergence of the mix ratio data to the optimal frontier. k-NN provides case-based verification of the optimized grout mix ratio, avoiding solutions exceeding the feasible range or failing to meet practical requirements due to insufficient multi-parameter fusion data.

[0092] Step 7: Grouting Feedback Monitoring. A closed-loop control system is formed through multi-sensor collaborative monitoring and dynamic parameter adjustment. Combined with the design of the exhaust pipe to balance the pressure in the grouting space, this is a core element in ensuring grouting quality and construction safety. Specifically, pressure sensors monitor the pressure within the grouting area in real time. When the pressure exceeds a threshold, a valve is opened to release pressure, preventing pressure instability within the filling area. Ultrasonic level sensors monitor the grouting process; when the return flow exceeds a set value, grouting is paused, and the pressure is appropriately increased to accelerate grout penetration based on the actual pressure. The control system employs an SVM algorithm to transform multi-source sensor data (pressure, seepage rate, return flow, etc.) into distinguishable features. Accurate classification of the grouting state is achieved by constructing an optimal classification hyperplane. A kernel function maps the multi-sensor features to a high-dimensional space, thereby finding the optimal hyperplane based on the high-dimensional feature space, obtaining the optimal solution. Finally, the grouting state of new samples is judged based on the hyperplane, realizing dynamic feedback of the system to multiple features, ultimately forming a closed-loop control. The relevant formulas are as follows: Kernel function formula: ;

[0093] Where K(x, x) i ) represents sample x and sample x i Similarity in high-dimensional space; γ is an adjustment parameter that controls the range of influence of each sample on its surroundings;

[0094] Step 8, Grouting Auxiliary Detection, specifically includes: during grouting preparation, uniformly adding a small amount of magnetic tracer and using an electromagnetic gradient meter to detect the filling status. Real-time monitoring of the filling area during the filling process displays the grout penetration and filling status, facilitating manual monitoring and supervision of the system's control effect. After filling, the filling area is scanned to obtain magnetic field strength distribution data. This distribution data is used to reconstruct a 3D model of the filling material, obtaining a 3D model of the filling material. This 3D model is then compared with 3D models of cavities and cracks to evaluate the filling quality and integrity. The 3D model of the filling material after initial solidification is numbered and stored in the system. During subsequent monitoring and maintenance of the filling area, the entered 3D model is compared with the 3D model of the filling material obtained during monitoring to determine the evolution of geological features and the safety of the filling.

[0095] Example 4: A method for grouting and sealing karst fissures in front of a tunnel boring machine (TBM) based on feedforward identification and directional control. This method employs a data collection system that includes an intelligent filling material selection model, a path planning model, a geological type judgment model, and a closed-loop control model for the filling process. Each model is used to collect data on the working process and form its own database. The system learns from the characteristics and measures taken during historical tasks and human intervention. When the server is idle under normal TBM operating conditions, the system performs self-training to continuously improve its self-judgment and decision-making capabilities, thereby reducing the workload required for human detection and intervention and enhancing the system's autonomy.

[0096] Mechanical and physical parameters of the strata ahead of the tunnel boring machine (TBM) are obtained through pilot borehole engineering parameter sensing, ground-penetrating radar, or ultrasonic scanning. This accurately captures the morphological characteristics, development patterns, and extension properties of karst fissures, thereby determining the three-dimensional morphology and filling status of the karst caves. Based on target area identification, an adaptive prediction algorithm is used to analyze fissure orientation, cavity configuration, and geostress data to generate the optimal grouting path and borehole layout scheme. This ensures that the path follows the fissure orientation and avoids stress concentration areas, and maximizes grouting coverage and minimizes disturbance by dynamically adjusting the borehole spacing. During the grouting parameter matching stage, an intelligent decision-making system is used to recommend grout ratio, concentration, pressure, and flow rate based on geological type, water-rich state, and engineering requirements, achieving adaptive parameter control. During directional grouting, the system controls a small drilling rig to perform differentiated grouting along the planned path. At the same time, pressure sensors and seepage rate sensors monitor changes in grouting reaction force, return flow rate, and permeability in real time, dynamically adjusting parameters to ensure grouting effectiveness. To assess the sealing quality, a magnetic tracer is uniformly added to the grout. After grouting, an electromagnetic gradient meter is used to scan and generate a three-dimensional model of the filling material, which is then compared with the initial cavity model to intuitively determine the integrity of the filling. In addition, the method integrates feedforward detection data with historical tunneling data to construct a database. Through machine learning, the grouting decision model is continuously trained and optimized to improve the system's adaptability and engineering reliability under complex geological conditions, ultimately forming a closed-loop intelligent control system from detection, planning, execution to feedback.

[0097] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

[0098] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A method for grouting and sealing karst fissures in front of a tunnel boring machine based on feedforward identification and directional control, characterized in that, The method for grouting and sealing karst fissures in front of the tunnel boring machine includes the following steps: S1. Obtaining geological information ahead: By sensing engineering parameters of the pilot borehole and scanning the strata ahead of the shield tunnel using ground-penetrating radar or ultrasonic waves, the mechanical and physical properties of the strata are obtained. S2. Target area identification: Based on the morphological characteristics of karst fissures, study the development characteristics and extension properties of karst fissures, and determine the three-dimensional morphology and filling status of karst caves. S3. Grouting path planning: Based on the detection results, the area to be grouted is automatically identified, and the optimal grouting path and hole layout scheme are generated according to the adaptive prediction algorithm of karst fracture configuration. S4. Parameter adaptive control: Matches grout type, concentration, grouting pressure and flow rate with the geological conditions of the target area to control the directional injection device; S5. Directional grouting is carried out, and differentiated grouting is performed along the planned path. During this process, the sealing effect is judged by the changes in grouting reaction force, return flow rate or permeability, and the relevant grouting parameters are dynamically adjusted. S6. Sealing effect feedback: Magnetic tracer is added evenly during the grouting process. After detection, a three-dimensional grouting model is generated and compared with the cavity model to determine the grouting quality.

2. The method for grouting and sealing karst fissures in front of a tunnel boring machine according to claim 1, characterized in that: In step S2, when identifying the three-dimensional morphology and filling status of the karst cave, the three-dimensional spatial geometric center coordinates and boundary range of the cave are identified and output through multi-dimensional signal verification combined with three-dimensional modeling constraints. Specifically, this includes: The system collects multi-dimensional detection information from ground-penetrating radar, and performs spatiotemporal matching of multi-dimensional signals through time-domain and frequency-domain feature fusion algorithms to generate a weighted cavity feature map. The weights are dynamically adjusted based on the amplitude of the reflected waves. The geometric center of the cavity is determined by combining the minimum circumscribed sphere with the axis-aligned bounding box. The coordinate parameters of the minimum three-dimensional cube containing the cavity and the radius parameters of the sphere are output. A spatial coordinate system with the ground-penetrating radar as the origin is established and used as a reference to control the drilling trajectory of the drill bit.

3. The method for grouting and sealing karst fissures in front of a tunnel boring machine according to claim 1, characterized in that: In step S3, the grouting path and hole arrangement are optimized using an adaptive prediction algorithm based on fracture development patterns, extension characteristics, and three-dimensional cavity configurations. Specifically, this includes: By reconstructing ground-penetrating radar images, the direction, dip angle, opening and extension length of fractures, as well as the three-dimensional coordinates, volume, wall thickness and connectivity topology of cavities, are quantitatively extracted. By combining geological stress data, the direction of fracture propagation is predicted using a fracture finite element model, and key support points of the cavity are identified through cavity stability analysis. A hole position constraint model is constructed to avoid support points and stress concentration areas at the crack tips. The path follows the crack direction, and the hole spacing is dynamically adjusted according to density. After adaptive iterative optimization, with the goal of grout coverage ≥90% and disturbance stress increment ≤0.1MPa, the final path and hole position coordinates are output.

4. The method for grouting and sealing karst fissures in front of a tunnel boring machine according to claim 1, characterized in that: In S4, the parameter adaptive control is based on the improved FCM algorithm. According to the geological type discrimination system, different strata types and water-rich states are characterized. Then, according to the slurry decision system, the optimal slurry ratio and setting time parameters are recommended in combination with geological type and related factors to determine the optimal ratio.

5. The method for grouting and sealing karst fissures in front of a tunnel boring machine according to claim 1, characterized in that: In S5, during differentiated grouting, a small drilling machine is used for path drilling. The drilling is based on underground drilling path planning. The drilling direction of the small drilling machine is pre-planned and corrected in real time according to the acquired potential fractures, cavities and stratum attribute data to ensure the smooth progress of the drilling process. The endpoint is planned according to the safe grouting point given by the prediction algorithm to ensure the safety and reliability of the grouting work of the small drilling machine.

6. The method for grouting and sealing karst fissures in front of a tunnel boring machine according to claim 1, characterized in that: In S5, during directional grouting, by analyzing geological characteristics and geological composition conditions, a multi-factor decision-making method is adopted to make a reasonable matching decision on the grouting composition ratio, so as to ensure that the injected grout can play a role in reinforcement and filling without affecting or damaging the rock strata in the area. The multi-factor decision-making method includes, but is not limited to, extraction of geological and physical characteristics of the target area, analysis of strata lithology and composition, control of engineering functional requirements, and combination of construction constraints and environmental compatibility requirements.

7. The method for grouting and sealing karst fissures in front of a tunnel boring machine according to claim 1, characterized in that: In S6, the sealing effect feedback adopts multiple sensors to monitor the grouting effect in real time and adjust the parameters in a closed loop. The sensors used include, but are not limited to, pressure sensors, seepage rate sensors and backflow detectors installed at the front end of the small drilling machine. The small drilling machine is equipped with an exhaust pipe at the front end to facilitate the discharge of gas in the cracks and cavities during grouting, and to prevent the collapse of cavities and cracks due to pressure imbalance in the space.

8. The method for grouting and sealing karst fissures in front of a tunnel boring machine according to claim 1, characterized in that: In step S6, the magnetic tracer is added uniformly without affecting the properties of the slurry itself, so that the geometric morphology of the filler is easier to detect after the slurry solidifies, which is conducive to the subsequent generation of a three-dimensional model of the filler and comparison with the three-dimensional model of the spatial geometry detected in the cavity and crack detection stage to determine the filling quality of the cavity area; at the same time, the magnetic tracer remaining inside is also conducive to the subsequent maintenance of cavities and cracks.

9. The method for grouting and sealing karst fissures in front of a tunnel boring machine according to any one of claims 1 to 8, characterized in that, It also includes fusing feedforward detection data with historical tunneling data to construct a geological-grouting response database, which is used to optimize grouting strategies for subsequent tunneling sections; storing fracture and cavity models in a cavity stress analysis and grouting point decision database, and retrieving models from the database for training and analysis when the processor is idle, so that the performance of the grouting decision model can be continuously improved as the amount of tunneling work increases.

Images