In-situ three-dimensional full-field transparent visualization evaluation method for rock mass grouting modification effect

Abstract

This invention discloses an in-situ three-dimensional full-field transparent visualization evaluation method for the effect of rock mass grouting modification, relating to the field of geotechnical engineering technology. The method includes S1: acquiring in-situ anisotropic acoustic CT data before and after grouting in the target grouting area using multi-directional acoustic ray clusters; obtaining a high-resolution three-dimensional anisotropic wave velocity field through anisotropic elastic wave inversion and physical constraint neural network super-resolution reconstruction; and automatically identifying high-value areas, low-value areas, gradient abrupt change areas, and stable areas based on wave velocity change and gradient change rate. This in-situ three-dimensional full-field transparent visualization evaluation method for the effect of rock mass grouting modification significantly improves the scientificity, precision, and visualization level of rock mass grouting modification effect evaluation, providing strong technical support for ensuring engineering quality and optimizing grouting design.

Description

Technical Field

[0001] This invention relates to the field of geotechnical engineering technology, specifically to an in-situ, three-dimensional, full-field transparent visualization evaluation method for the effect of rock mass grouting modification. Background Technology

[0002] Accurate and scientific evaluation of the effects of rock mass grouting modification is crucial for ensuring project quality. However, the grouting process occurs inside the rock mass, making it a typical "black box" hidden project. It has long faced the technical bottleneck of "invisibility," and existing technologies struggle to accurately measure the degree of grout filling in fissures and scientifically evaluate the rock mass improvement effect after grouting. There is an urgent need to develop transparent and visual testing and evaluation methods for the effects of rock mass grouting modification to achieve the goal of "visible rock mass modification" and ensure the quality of grouting projects.

[0003] Regarding visualization techniques for grouting modification effects, existing technologies are mainly limited to laboratory physical simulation tests (such as CN202510131784.X and CN202510004438.5), making it difficult to directly apply to in-situ visualization assessment of grouting modification effects in engineering projects. Currently, the main methods for in-situ testing and evaluation of grouting effects in engineering projects include two major categories: borehole methods and geophysical methods. Borehole methods (coring and water pressure tests) are essentially discrete point tests, making it difficult to comprehensively, continuously, and quantitatively reflect the modification effect of the entire grouting area. Geophysical methods (such as wave velocity testing) can provide continuous field information, but they lack universal and quantitative mapping models with key physical and mechanical parameters of the rock mass (grout filling rate, strength, and modulus), making it difficult to scientifically assess the actual improvement in the strength and mechanical properties of the rock mass after grouting modification. For example, Chinese patent CN202411718040.X discloses a method, system, and equipment for detecting grout vein distribution after grouting in rock and soil. Its core still relies on indirect parameters such as acoustic velocity for qualitative or semi-quantitative judgment, failing to obtain the distribution of mechanical strength of grout-modified rock mass. How to organically integrate and quantify the discrete point data (such as borehole test results) obtained from conventional grouting effect evaluation methods with full-field qualitative fuzzy data (such as geophysical exploration results), thereby achieving high-resolution, transparent visualization, and continuous full-field characterization of the physical and mechanical strength of rock mass in the grouting-modified area, is a key challenge that urgently needs to be overcome. Existing research on the processing of multi-source data such as geophysical exploration, borehole drilling, and engineering monitoring for grouting effect evaluation remains fragmented, lacking a unified spatiotemporal data fusion and analysis method, as well as an evaluation method system that is transparent and visualized for practical engineering applications. Summary of the Invention

[0004] The purpose of this invention is to provide an in-situ, three-dimensional, full-field transparent visualization evaluation method for the effect of rock mass grouting modification, thereby solving the problems existing in the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: an in-situ three-dimensional full-field transparent visualization evaluation method for the effect of rock mass grouting modification, the method comprising:

[0006] S1. In the target grouting area, in-situ anisotropic acoustic CT data before and after grouting are collected using multi-directional acoustic ray clusters. Through anisotropic elastic wave inversion and physical constraint neural network super-resolution reconstruction, a high-resolution three-dimensional anisotropic wave velocity field is obtained. Based on the wave velocity change and gradient change rate, high-value areas, low-value areas, gradient abrupt change areas and stable areas are automatically identified.

[0007] S2. Based on the target grouting area, target boreholes are drilled and core samples are taken. The grouting modified rock cores are characterized by micron-level three-dimensional structural features and tested for anisotropic mechanical parameters to obtain structural feature tensors and mechanical parameter tensors corresponding to spatial position and structural direction.

[0008] S3. Construct a graph neural network model with physical constraints such as orthogonal anisotropic wave equations. Using the high-resolution anisotropic wave velocity field as input, train the model by combining the mechanical parameter test results. Establish a global quantitative mapping relationship from the wave velocity field to the fracture / grout vein structure feature field and the anisotropic mechanical property field, and output a high-resolution three-dimensional mechanical property field covering the entire grouting area.

[0009] S4. Import the mechanical property field into numerical simulation software for engineering response prediction, and compare and analyze it with the field monitoring data. Based on the differences, dynamically optimize the parameters and update the model. As the project progresses, continuously acquire new data for learning, so as to achieve continuous improvement in dynamic evolution characterization and prediction accuracy.

[0010] S5 integrates geological survey data, merging wave velocity field, target grouting area distribution, mechanical parameter test results, mechanical property field, numerical simulation results, and field monitoring data into a three-dimensional transparent visualization platform. It also integrates an intelligent decision-making platform that can automatically extract key information, provide safety risk warnings, and assist in optimizing grouting reinforcement schemes.

[0011] Preferably, the multi-directional acoustic beam cluster acquisition in S1 includes: three or more non-collinear boreholes arranged in a triangular or tetrahedral configuration in the target grouting area; a programmable multi-channel acoustic wave transmitting transducer array arranged in the transmitting borehole to support directional acoustic beam transmission with full angular coverage from 0° to 180°, with a minimum step angle of no more than 5°; focusing acoustic wave energy using phase-controlled beamforming technology; recording direct wave and scattered wave field data using a high-sensitivity three-component accelerometer array in the receiving borehole; and ensuring time synchronization accuracy of no more than 100ns and spatial positioning accuracy of no more than 1cm by using GPS / PPS time synchronization and in-hole inertial navigation unit positioning.

[0012] Preferably, the inversion of the anisotropic wave velocity field in S1 involves discretizing the target region into a three-dimensional regular grid, selecting an orthogonal anisotropic elastic wave equation, using the first arrival wave travel time or full waveform information to invert the complete wave velocity tensor of each unit, generating a three-dimensional anisotropic wave velocity field with a resolution of 1m×1m×1m as input, and performing super-resolution reconstruction through a U-Net neural network with embedded physical constraints, outputting a three-dimensional anisotropic wave velocity tensor field with a resolution of 0.1m×0.1m×0.1m.

[0013] Preferably, the automatic identification based on wave velocity change and gradient change rate in S1 includes: calculating the wave velocity change field and its spatial gradient field before and after grouting, and automatically marking high-value areas, low-value areas, gradient abrupt change areas and stable areas based on the dual indicators ΔVp and |∇Vp| using the K-means clustering algorithm.

[0014] Preferably, the micron-level three-dimensional structural features in S2 are characterized by performing micro-focus X-ray μ-CT scanning on the grout core with a resolution of no more than 5 μm, and after three-dimensional reconstruction, the structural tensor method is used to calculate the main direction of the fracture network and the directional filling rate of the grout veins, and the roughness and contact tightness of the grout-rock interface are analyzed.

[0015] Preferably, the anisotropic mechanical parameter test in S2 is based on the main structural direction revealed by μ-CT, and standard specimens are prepared along the direction parallel and perpendicular to the dominant direction of cracks / veins to test the equivalent elastic modulus, uniaxial compressive strength, tensile strength and shear strength.

[0016] Preferably, the node features of the S3 graph neural network model include node center coordinates, cell size, anisotropic wave velocity tensor parameters, and anisotropic mechanical parameters. The edge features include the main direction of the fracture / mauldron, wave velocity gradient, and distance between adjacent cells. Furthermore, the physical constraint layer embeds the residual terms of the orthogonal anisotropic wave equation and the mass conservation equation.

[0017] Preferably, the parameter dynamic optimization in S4 is based on a Bayesian inversion framework or optimization algorithm, which uses the difference between numerical simulation prediction and field monitoring data to adjust the mechanical parameter field, and integrates newly added borehole core data, monitoring data and grouting parameter records through an incremental learning mechanism to update the model.

[0018] Preferably, the three-dimensional transparent visualization platform in S5 integrates geological exploration data, wave velocity field, key area distribution, core test results, mechanical property field, numerical simulation results and field monitoring data, supports arbitrary cutting, transparent display, attribute query and dynamic playback, and provides risk prediction and grouting reinforcement scheme optimization suggestions based on machine learning.

[0019] As can be seen from the above technical solution, the present invention has the following beneficial effects:

[0020] This in-situ, three-dimensional, full-field transparent visualization evaluation method for the effect of rock mass grouting modification utilizes multi-directional, in-situ anisotropic acoustic CT testing of the target area before and after grouting. Combined with anisotropic elastic wave inversion and super-resolution reconstruction using a neural network embedded with physical constraints, a high-resolution three-dimensional anisotropic wave velocity field is obtained. This accurately reflects the changes in the internal physical properties of the rock mass caused by grouting and automatically identifies high-value areas, low-value areas, gradient abrupt change areas, and stable areas, achieving precise positioning of key areas of modification effect. Targeted core drilling in key areas, combined with micron-level three-dimensional structural feature characterization and anisotropic mechanical parameter testing, yields high-precision physical and mechanical data corresponding one-to-one with spatial location and structural direction, providing real constraints for the quantitative mapping of the wave velocity field and mechanical field. Finally, a graph neural network model embedded with physical constraints such as orthogonal anisotropic wave equations is used to realize the transformation of the wave velocity field into the fracture / grout vein structural feature field and anisotropy. The comprehensive, continuous, and high-resolution prediction of mechanical property fields overcomes the limitations of traditional methods, such as borehole discretization and the lack of quantitative mapping models in geophysical exploration. The predicted mechanical property fields are imported into numerical simulation software for engineering response prediction, and compared with field monitoring data. Based on the differences, dynamic parameter optimization and model updates are carried out, constructing a closed-loop improvement mechanism of in-situ monitoring—inversion analysis—mechanical mapping—engineering prediction—feedback optimization. Finally, wave velocity field, key area distribution, core test results, mechanical property fields, numerical simulation results, and monitoring data are integrated into a three-dimensional transparent visualization platform. This achieves integrated, three-dimensional transparent expression of multi-source data from geological background, wave velocity changes, structural characteristics, mechanical enhancement to engineering response, supporting risk prediction and scheme optimization. It significantly improves the scientific rigor, precision, and visualization level of rock mass grouting modification effect evaluation, providing strong technical support for ensuring engineering quality and optimizing grouting design. Attached Figure Description

[0021] Figure 1 This is a flowchart of the method of the present invention;

[0022] Figure 2 A schematic diagram of a global quantitative mapping of wave velocity field to mechanical field driven by a physical constraint graph neural network;

[0023] Figure 3 A schematic diagram illustrating the construction of a 3D transparent visualization platform for multi-source data fusion. Detailed Implementation

[0024] 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.

[0025] like Figures 1-3 As shown, this invention provides a technical solution: an in-situ three-dimensional full-field transparent visualization evaluation method for the effect of rock mass grouting modification, the method comprising:

[0026] S1. In the target grouting area, in-situ anisotropic acoustic CT data before and after grouting are collected using multi-directional acoustic ray clusters. Through anisotropic elastic wave inversion and physical constraint neural network super-resolution reconstruction, a high-resolution three-dimensional anisotropic wave velocity field is obtained. Based on the wave velocity change and gradient change rate, high-value areas, low-value areas, gradient abrupt change areas and stable areas are automatically identified.

[0027] S2. Based on the target grouting area, target boreholes are drilled and core samples are taken. The grouting modified rock cores are characterized by micron-level three-dimensional structural features and tested for anisotropic mechanical parameters to obtain structural feature tensors and mechanical parameter tensors corresponding to spatial position and structural direction.

[0028] S3. Construct a graph neural network model with physical constraints such as orthogonal anisotropic wave equations. Using the high-resolution anisotropic wave velocity field as input, train the model by combining the mechanical parameter test results. Establish a global quantitative mapping relationship from the wave velocity field to the fracture / grout vein structure feature field and the anisotropic mechanical property field, and output a high-resolution three-dimensional mechanical property field covering the entire grouting area.

[0029] S4. Import the mechanical property field into numerical simulation software for engineering response prediction, and compare and analyze it with the field monitoring data. Based on the differences, dynamically optimize the parameters and update the model. As the project progresses, continuously acquire new data for learning, so as to achieve continuous improvement in dynamic evolution characterization and prediction accuracy.

[0030] S5 integrates geological survey data, merging wave velocity field, target grouting area distribution, mechanical parameter test results, mechanical property field, numerical simulation results, and field monitoring data into a three-dimensional transparent visualization platform. It also integrates an intelligent decision-making platform that can automatically extract key information, provide safety risk warnings, and assist in optimizing grouting reinforcement schemes.

[0031] This method first deploys a multi-directional acoustic beam cluster testing system in the target area before grouting to perform anisotropic acoustic CT in-situ detection of the rock mass and obtain wave velocity field data before grouting. After grouting, the exact same measurement setup and parameters are repeated to obtain wave velocity field data after grouting. The test data is used for anisotropic elastic wave inversion to obtain a preliminary distribution of the three-dimensional anisotropic wave velocity tensor. This tensor is then input into a neural network model with embedded physical constraints for super-resolution reconstruction, resulting in a three-dimensional anisotropic wave velocity field with a resolution improved to the 0.1m level.

[0032] Subsequently, by comparing the change in wave velocity (ΔVp) and gradient change rate (|∇Vp|) before and after grouting, the high-value areas, low-value areas, gradient abrupt change areas, and stable areas of grouting modification effect were automatically identified. Based on the location of these areas, targeted drilling and coring were performed to obtain high-quality, azimuth-traceable grout-modified rock core samples. These samples were then subjected to micron-level three-dimensional structural feature characterization and anisotropic mechanical parameter testing in the laboratory, forming structural feature tensors and mechanical parameter tensors that precisely correspond to the spatial location and structural orientation.

[0033] Using the aforementioned high-precision wave velocity field and core measurement data, a graph neural network model with physical constraints such as orthogonal anisotropic wave equations is constructed. The model is trained in a unified spatial coordinate system and a global quantitative mapping relationship is established from the wave velocity field to the fracture / vein structure feature field and the anisotropic mechanical property field. The global high-resolution three-dimensional mechanical property field is then predicted.

[0034] Mechanical property fields are imported into geotechnical engineering numerical simulation software (such as FLAC3D, 3DEC, and ABAQUS) to simulate the deformation and stability of engineering structures under load. The simulation results are compared and analyzed with field monitoring data, and the parameters are dynamically optimized and the model updated using the feedback results, forming a closed-loop iterative mechanism. Finally, wave velocity fields, key area distributions, core test results, mechanical property fields, numerical simulation results, and field monitoring data are integrated into a three-dimensional transparent visualization platform, achieving a full-chain, three-dimensional, full-field, interactive, and transparent expression from geological background, grouting structure changes, mechanical enhancement to engineering response.

[0035] By integrating multi-source data and intelligent mapping of physical constraints, the problems of the invisibility of grouting modification effects and the separation of discrete / continuous data were solved, realizing in-situ three-dimensional full-field transparent visualization and quantitative evaluation, thus improving the scientificity and accuracy of engineering evaluation.

[0036] The multi-directional acoustic beam cluster acquisition in S1 includes: three or more non-collinear boreholes arranged in a triangular or tetrahedral configuration in the target grouting area; a programmable multi-channel acoustic wave transmitting transducer array arranged in the transmitting borehole to support directional acoustic beam transmission with full angular coverage from 0° to 180°, with a minimum step angle of no more than 5°; focusing acoustic wave energy using phase-controlled beamforming technology; and arranging a high-sensitivity three-component accelerometer array in the receiving borehole to record direct wave and scattered wave field data. GPS / PPS time synchronization and in-hole inertial navigation unit positioning are used to ensure time synchronization accuracy of no more than 100ns and spatial positioning accuracy of no more than 1cm.

[0037] In the target grouting area, three or more non-collinear boreholes are preferably arranged in a triangular or tetrahedral configuration to ensure omnidirectional coverage of the test ray path in space. A programmable multi-channel acoustic wave transmitting transducer array is installed in the transmitting borehole, enabling omnidirectional acoustic beam transmission within the range of 0° to 180°, with a minimum step angle of no more than 5°. Phase-controlled beamforming technology is used to focus the acoustic energy towards the target direction, improving penetration and signal-to-noise ratio. A high-sensitivity three-component accelerometer array is arranged in the receiving borehole, capable of completely recording the three components of the direct and scattered wave fields. To ensure the temporal and spatial accuracy of the multi-channel, multi-bore arrangement, a high-precision GPS / PPS timing system (synchronization accuracy ≤100ns) and an in-bore high-precision inertial navigation unit (INS, positioning accuracy ≤1cm) are used to synchronize and position the acquisition equipment, ensuring strict spatiotemporal consistency of the ray data from different measuring points, thereby obtaining a high-density, multi-directional in-situ wave velocity dataset.

[0038] This arrangement improves the coverage and signal-to-noise ratio of the sound wave propagation direction, ensures the integrity and accuracy of the wave velocity field inversion data, and provides a reliable foundation for subsequent anisotropic analysis and super-resolution reconstruction.

[0039] The inversion of the anisotropic wave velocity field in S1 involves discretizing the target region into a three-dimensional regular grid, selecting an orthogonal anisotropic elastic wave equation, using the first arrival wave travel time or full waveform information to invert the complete wave velocity tensor of each unit, and then performing super-resolution reconstruction through a U-Net neural network with embedded physical constraints, outputting a three-dimensional anisotropic wave velocity tensor field with a resolution of 0.1m×0.1m×0.1m.

[0040] The target grouting area is discretized using a three-dimensional regular grid, and an orthogonal anisotropic elastic wave equation is introduced during the inversion stage to incorporate the differences in sound wave propagation speed in different directions. Using first-arrival travel time data or full waveform data, the complete wave velocity tensor of each grid cell is solved by inversion, including the three principal wave velocity components and their direction information. The resulting low-resolution wave velocity field (e.g., 1m×1m×1m) is used as the initial solution and input into a U-Net neural network with embedded physical constraints. Combined with the original travel time or waveform data as a supervision signal, super-resolution reconstruction is performed, ultimately obtaining a high-resolution three-dimensional anisotropic wave velocity tensor field with a resolution of 0.1m×0.1m×0.1m, significantly enhancing the analytical capability for small-scale cracks and heterogeneous bodies.

[0041] This method breaks through the traditional isotropic assumption and low resolution limitations, and can identify sub-meter-level cracks and magma vein structures, significantly improving the spatial precision of the wave velocity field.

[0042] The automatic identification based on wave velocity change and gradient change rate in S1 includes: calculating the wave velocity change field and its spatial gradient field before and after grouting, and automatically marking high-value areas, low-value areas, gradient abrupt change areas and stable areas based on the dual indicators ΔVp and |∇Vp| using the K-means clustering algorithm.

[0043] Using high-resolution three-dimensional anisotropic wave velocity fields measured before and after grouting, the wave velocity change ΔVp was calculated for the same grid cell, and the gradient change rate |∇Vp| was calculated based on three-dimensional spatial difference. Using ΔVp and |∇Vp| as clustering features, the K-means unsupervised clustering algorithm automatically divided the region into four categories: high-value regions (significant wave velocity increase), low-value regions (weak increase), gradient abrupt change regions (boundaries or pinch-out areas), and stable regions (minimal wave velocity change). This classification process was fully automated and output the spatial coordinates and extent of each category, providing precise location data for subsequent targeted coring and verification.

[0044] The micron-level three-dimensional structural features of S2 were characterized by performing micro-focus X-ray μ-CT scanning with a resolution of no more than 5 μm on the grouted core. After three-dimensional reconstruction, the structural tensor method was used to calculate the main direction of the fracture network and the directional filling rate of the grout veins, and the roughness and contact tightness of the grout-rock interface were analyzed.

[0045] In grout-modified rock cores drilled in key areas, a microfocus X-ray μ-CT scanning system with a resolution of no more than 5 μm was used to perform full-section scanning to acquire high-precision three-dimensional voxel data. Three-dimensional reconstruction technology was used to reconstruct the fracture network and magma vein distribution morphology within the rock cores. The principal directions (dip and dip angles) and their distribution probabilities of the fracture network were calculated using the structural tensor method. The volume fraction of magma veins in the corresponding fracture spaces was statistically analyzed according to directional intervals to obtain the magma vein directional filling rate. Simultaneously, the geometric morphology information of the magma-rock interface was extracted, and the interface roughness and contact tightness were calculated, thereby obtaining a complete set of structural feature tensor parameters.

[0046] The anisotropic mechanical parameters in S2 are tested based on the main structural directions revealed by μ-CT. Standard specimens are prepared along the directions parallel and perpendicular to the dominant directions of cracks / magma veins, and the equivalent elastic modulus, uniaxial compressive strength, tensile strength and shear strength are tested.

[0047] Based on the dominant orientation of fractures and magma veins revealed by μ-CT, standard specimens (meeting ISRM standard size requirements) were precisely cut from core samples along directions parallel and perpendicular to these orientations. Uniaxial compressive, tensile, and shear tests were then conducted, recording the directional differences in parameters such as elastic modulus and strength. Through pairwise data analysis, equivalent mechanical parameter tensors in different directions were established and correlated with the spatial coordinates of the sampling locations, forming a high-confidence directional mechanical database.

[0048] The node features of the S3 neural network include node center coordinates, cell size, anisotropic wave velocity tensor parameters, and anisotropic mechanical parameters. The edge features include the main direction of the fracture / mauldron, wave velocity gradient, and distance between adjacent cells. Furthermore, the physical constraint layer embeds the residual terms of the orthogonal anisotropic wave equation and the mass conservation equation.

[0049] The discretized grid cells of the grouting area are abstracted as nodes in a graph neural network, and edge connections are established between adjacent cells. Node features include node geometric parameters (center coordinates, cell size), anisotropic wave velocity tensor, and anisotropic mechanical parameters; edge features include the principal direction of fractures / grout veins, wave velocity gradients between adjacent cells, and the distance between the centers of two cells. A physical constraint layer is introduced into the network structure, embedding the residuals of the orthogonal anisotropic wave equation and the mass conservation equation as regularization terms to ensure that the structural feature field and mechanical property field predicted by the model are physically self-consistent and conform to the laws of rock mechanics.

[0050] In S4, parameter dynamic optimization is based on a Bayesian inversion framework or optimization algorithm. The difference between numerical simulation prediction and field monitoring data is used to adjust the mechanical parameter field. The model is updated by integrating newly added borehole core data, monitoring data and grouting parameter records through an incremental learning mechanism.

[0051] The three-dimensional mechanical property field obtained by PGNN mapping is input into numerical simulation software to establish a refined calculation model, predicting engineering response indicators such as displacement field, stress field, and safety factor under load conditions. Real-time data on multi-point displacement, stress, and convergence are acquired through a field monitoring system, and these measured results are compared with the simulation prediction results. The differences are input into a Bayesian inversion framework or optimization algorithm to dynamically adjust the local mechanical parameter field (such as elastic modulus E, cohesion c, friction angle φ, etc.), and new borehole test, monitoring data, and grouting parameters are introduced through an incremental learning mechanism to update the model in real time, achieving continuous improvement in prediction accuracy.

[0052] The S5 3D transparent visualization platform integrates geological exploration data, wave velocity field, key area distribution, core test results, mechanical property field, numerical simulation results, and field monitoring data. It supports arbitrary cross-sectioning, transparent display, attribute query, and dynamic playback, and provides risk prediction and grouting reinforcement scheme optimization suggestions based on machine learning.

[0053] A transparent visualization platform is built based on a 3D geological-engineering digital twin model. It integrates and renders multi-source data fields (including wave velocity fields, key area distributions, core test data, mechanical property fields, numerical simulation results, and field monitoring data), employing a GPU-parallel computing-driven volume rendering engine to achieve efficient 3D display of large-scale data. The platform supports arbitrary cross-sections, transparent display, attribute information querying, before-and-after grouting comparisons, and dynamic process playback. It can also integrate machine learning models to predict future risk trends. Furthermore, based on information such as low-value areas and gradient abrupt change areas, combined with empirical rules and simulation results, the system can automatically generate grouting reinforcement scheme suggestions.

[0054] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for in-situ three-dimensional full-field transparent visualization evaluation of the modification effect of rock mass grouting, characterized in that, The method comprises: S1, in the target grouting area, multi-directional acoustic ray cluster acquisition is used to collect in-situ anisotropic acoustic CT data before and after grouting, high-resolution three-dimensional anisotropic wave velocity field is obtained through anisotropic elastic wave inversion and physical constraint neural network super-resolution reconstruction, and high-value area, low-value area, gradient mutation area and stable area are automatically identified based on wave velocity variation and gradient variation rate; S2, according to the target grouting area, target drilling core is arranged, micron-level three-dimensional structural feature characterization and anisotropic mechanical parameter testing are carried out on the grouting modified core, and structural feature tensor and mechanical parameter tensor corresponding to spatial position and structural direction are obtained; S3, a graph neural network model embedded with physical constraints of orthogonal anisotropic wave equation is constructed, the high-resolution anisotropic wave velocity field is taken as input, the model is trained in combination with the mechanical parameter test results, global quantitative mapping relationship from the wave velocity field to the fracture / grout structure feature field and the anisotropic mechanical property field is established, and a high-resolution three-dimensional mechanical property field covering the entire grouting area is output; S4, the mechanical property field is introduced into a numerical simulation software to predict engineering response, and compared with field monitoring data, parameter dynamic optimization and model updating are carried out based on the difference, and the model is continuously learned with newly added data obtained as the engineering advances, so that dynamic evolution characterization and continuous improvement of prediction accuracy are realized; S5, geological survey data are integrated, the wave velocity field, target grouting area distribution, mechanical parameter test results, mechanical property field, numerical simulation results and field monitoring data are fused into a three-dimensional transparent visualization platform, and an intelligent decision-making platform for automatically extracting key information, safety risk early warning and assisting grouting reinforcement scheme optimization is integrated.

2. The method for in-situ three-dimensional full-field transparent visualization evaluation of rock mass grouting modification effect according to claim 1, characterized in that: The multi-directional acoustic ray cluster acquisition in S1 comprises: three or more non-collinear drill holes arranged in a triangular or tetrahedral configuration in the target grouting area, a programmable multi-channel acoustic emission transducer array is arranged in the emission hole, directional acoustic beam emission covering 0°-180° is supported, the minimum step angle is not greater than 5°, phase control beam forming technology is used to focus acoustic energy, a high-sensitivity three-component accelerometer array is arranged in the receiving hole to record direct wave and scattered wave field data, and GPS / PPS time service and hole inertial navigation unit positioning are used to ensure that the time synchronization accuracy is not greater than 100 ns and the spatial positioning accuracy is not greater than 1 cm.

3. The method for in-situ three-dimensional full-field transparent visualization evaluation of rock mass grouting modification effect according to claim 1, characterized in that: The anisotropic elastic wave inversion in S1 comprises: discretizing the target area into a three-dimensional regular grid, selecting an orthogonal anisotropic elastic wave equation, inverting the complete wave velocity tensor of each unit by using first arrival wave travel time or full waveform information, generating a resolution of 1m*1m*1m three-dimensional anisotropic wave velocity field as input, and performing super-resolution reconstruction by a U-Net neural network embedded with physical constraints to output a three-dimensional anisotropic wave velocity tensor field with a resolution of 0.1m*0.1m*0.1m.

4. The method for in-situ three-dimensional full-field transparent visualization evaluation of rock mass grouting modification effect according to claim 1, characterized in that: The automatic identification based on wave velocity variation and gradient variation rate in S1 comprises: calculating the wave velocity variation field and its spatial gradient field before and after grouting, and automatically marking the high-value area, low-value area, gradient mutation area and stable area based on the double indexes of ΔVp and |∇Vp|. The automatic identification based on wave velocity variation and gradient variation rate in S1 comprises: calculating the wave velocity variation field and its spatial gradient field before and after grouting, and automatically marking the high-value area, low-value area, gradient mutation area and stable area based on the double indexes of ΔVp and |∇Vp|.

5. The method for in-situ three-dimensional full-field transparent visualization evaluation of rock mass grouting modification effect according to claim 1, characterized in that: The micro three-dimensional structure feature in S2 is characterized by micro-focus X-ray mu-CT scanning of the core after grouting with a resolution not greater than 5 μm, and the main direction of the fracture network and the directional filling rate of the grout vein are calculated by using the structure tensor method after three-dimensional reconstruction, and the roughness and contact tightness of the grout-rock interface are analyzed.

6. The method for in-situ three-dimensional full-field transparent visualization evaluation of rock mass grouting modification effect according to claim 1, characterized in that: The anisotropic mechanical parameter test in S2 is based on the main direction revealed by mu-CT, and standard samples are prepared along the parallel and perpendicular directions of the dominant direction of the fracture / grout vein, and the equivalent elastic modulus, uniaxial compressive strength, tensile strength and shear strength are tested.

7. The method for in-situ three-dimensional full-field transparent visualization evaluation of rock mass grouting modification effect according to claim 1, characterized in that: The node features of the graph neural network model in S3 include node center coordinates, element size, anisotropic wave velocity tensor parameters and anisotropic mechanical parameters, and the edge features include the main direction of the fracture / grout vein, the wave velocity gradient and the distance between adjacent elements, and the physical constraint layer is embedded in the residual term of the orthogonal anisotropic wave equation and the mass conservation equation.

8. The method for in-situ three-dimensional full-field transparent visualization evaluation of rock mass grouting modification effect according to claim 1, characterized in that: The parameter dynamic optimization in S4 is based on the Bayesian inversion framework or optimization algorithm, and the difference between the numerical simulation prediction and the field monitoring data is used to adjust the mechanical parameter field, and the incremental learning mechanism is used to integrate the new drilling core data, monitoring data and grouting parameter records to update the model.

9. The method for in-situ three-dimensional full-field transparent visualization evaluation of rock mass grouting modification effect according to claim 1, characterized in that: The three-dimensional transparent visualization platform in S5 integrates geological survey data, wave velocity field, key area distribution, core test results, mechanical property field, numerical simulation results and field monitoring data, supports arbitrary cutting, transparent display, attribute query and dynamic playback, and provides risk prediction and grouting reinforcement scheme optimization suggestions based on machine learning.

Images