An experiment-based deep rock mass pore topology configuration numerical material fusion intelligent deduction method

Abstract

The application discloses a kind of based on test deep rock mass pore topological configuration number thing fusion intelligent deduction method, comprising: obtaining the standard rock sample of deep rock mass;The standard rock sample is marked, and test standard rock sample is obtained;Test standard rock sample is subjected to loading disaster test, and the mining process and test data of deep rock mass are obtained;According to the mining process and test data of deep rock mass, the point cloud information of loading process of test standard rock sample is obtained;According to point cloud data, test standard rock sample is three-dimensionally reconstructed, and the space dynamic evolution of deep rock mass loading behavior is obtained;According to space dynamic evolution, the feature extraction of deep loading rock mass pore topological configuration is carried out, and dynamic evolution feature is obtained.The application overcomes the dynamic observation problem of deep loading rock mass pore topological configuration based on number thing fusion technology, can reflect the real pore topological configuration evolution law of deep rock mass, and has important significance to reveal deep energy hosting rock mass mining disaster and stability evaluation.

Description

Technical Field

[0001] This invention belongs to the field of deep rock mass pore estimation technology, and particularly relates to an experimental-based intelligent estimation method for deep rock mass pore topology configuration through data and physical fusion. Background Technology

[0002] With increasing energy demand and the gradual depletion of shallow resources, energy development is increasingly moving towards the Earth's depths. Deep rock masses exist in complex environments under prolonged loading, and improper control during extraction can easily trigger deep engineering disasters. Pore topology is a crucial indicator describing the physical, mechanical, and permeability properties of rocks. Studying the evolution of pore topology can more accurately predict the mechanical response and failure mechanisms of rock masses during excavation, and assess the storage and migration conditions of deep energy resources, thus guiding deep energy exploration and development.

[0003] Due to the complex environment of deep rock masses, the lack of visibility within the rock mass, and the intricate pore topology, obtaining information on the pore distribution in deep rock masses is difficult. Current research on rock mass porosity measurement primarily focuses on laboratory experiments. Although various testing methods exist, most only measure pores in a single state of the rock mass, failing to capture the entire process from initial conditions to mining-induced disasters. The spatially random and complex pore topology of rock masses makes it impossible to precisely capture the dynamic evolution of pore topology within the rock mass. Furthermore, the difficulty and limited quantity of deep rock mass samples mean that the amount of data obtained through repeated experiments is small, and the limited experimental samples are insufficient to accurately reflect the true evolution of pore topology in deep, loaded rock masses. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention proposes an experimental-based intelligent inference method for deep rock mass pore topology configuration based on data and physical fusion, thereby resolving the issues present in the existing technologies.

[0005] To achieve the above objectives, in a first aspect, the present invention provides an experimental-based intelligent inference method for deep rock mass pore topology configuration based on data and physical fusion, comprising:

[0006] Obtain standard rock samples from deep rock masses;

[0007] The standard rock samples were marked to obtain the test standard rock samples;

[0008] The test standard rock samples were subjected to a load-bearing catastrophic test to obtain the deep rock mass mining process and test data;

[0009] Based on the deep rock mass mining process and test data, point cloud information of the loading process of the test standard rock sample is obtained through physical methods; based on the point cloud data, the test standard rock sample is reconstructed in three dimensions using mathematical methods to obtain the spatial dynamic evolution of the loading behavior of the deep rock mass;

[0010] Based on the aforementioned spatial dynamic evolution, the pore topology of deep loaded rock masses is characterized by feature extraction to obtain dynamic evolution features.

[0011] Preferably, marking the standard rock sample includes:

[0012] A standard rock sample was C-14 labeled using an isotope vacuum saturation tracer device; wherein the isotope vacuum saturation tracer device includes: a vacuum pump, valve A, a saturation cylinder, a standard rock sample, a pressure gauge, valve B, and a water injection tank.

[0013] Preferably, the loading-catastrophic test on the test standard rock sample includes:

[0014] A dynamic evolution tracking loading device was used to conduct a load-bearing catastrophic test on the test standard rock sample; wherein the dynamic evolution tracking loading device includes: a support base, a horizontal loading device-I, a horizontal loading device-II, a circular guide rail, a vertical loading device, and a vertical loading cylinder.

[0015] Preferably, obtaining the point cloud information of the loading process of the test standard rock sample includes:

[0016] Based on the deep rock mass mining process and test data, the spatial distance between the test standard rock sample and the isotope detection and tracking device was obtained.

[0017] Calculate the spatial coordinates of the test standard rock sample based on the spatial distance;

[0018] Based on the spatial coordinates, spatial point cloud data of the loading process of the test standard rock sample are obtained.

[0019] Preferably, feature extraction of the pore topology of deep loaded rock masses includes:

[0020] Based on the spatial dynamic evolution, a two-dimensional sample image of the test standard rock sample is obtained;

[0021] A fully convolutional neural network is used to extract pore topology features from the two-dimensional sample images to obtain dynamic evolution features.

[0022] Secondly, this invention also discloses an experimental-based intelligent inference system for deep rock mass pore topology configuration, comprising:

[0023] The rock sample acquisition module is used to acquire standard rock samples from deep rock masses;

[0024] A rock sample marking module is used to mark the standard rock sample to obtain the test standard rock sample;

[0025] The rock sample testing module is used to conduct load-bearing catastrophic tests on the test standard rock samples to obtain the deep rock mass mining process and test data.

[0026] The three-dimensional reconstruction module is used to obtain point cloud information of the loading process of the test standard rock sample based on the mining process and test data of the deep rock mass; and to perform three-dimensional reconstruction of the test standard rock sample based on the point cloud data to obtain the spatial dynamic evolution of the loading behavior of the deep rock mass.

[0027] The feature extraction module is used to extract features of the pore topology of deep loaded rock mass based on the spatial dynamic evolution, so as to obtain dynamic evolution features.

[0028] Thirdly, the present invention also discloses a computer device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method described in the first aspect.

[0029] Fourthly, the present invention also discloses a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described in the first aspect.

[0030] Fifthly, the present invention also discloses a computer program product, including a computer program that, when executed by a processor, implements the steps of the method described in the first aspect.

[0031] Compared with the prior art, the present invention has the following advantages and technical effects:

[0032] This invention provides an experimental-based intelligent inference method for deep rock mass pore topology configuration through data fusion, comprising: first, obtaining standard rock samples of the deep rock mass; second, marking the standard rock samples to obtain experimental standard rock samples; next, conducting a load-bearing catastrophic test on the experimental standard rock samples to obtain the deep rock mass mining process and test data; further, obtaining point cloud information of the loading process of the experimental standard rock samples based on the deep rock mass mining process and test data; third, performing three-dimensional reconstruction of the experimental standard rock samples based on the point cloud data to obtain the spatial dynamic evolution of the deep rock mass loading behavior; and finally, extracting features of the deep loaded rock mass pore topology configuration based on the spatial dynamic evolution to obtain dynamic evolution features.

[0033] This invention combines experimental testing techniques with intelligent algorithms, achieving real-time capture of the loading-induced catastrophic process and point cloud information of deep rock masses based on data-physical fusion technology. Through 3D reconstruction of the point cloud data, it recreates the spatial dynamic evolution of the loading behavior of deep rock masses, enabling the acquisition and intelligent prediction of the dynamic evolution characteristics of the pore topology of deep loaded rock masses. Compared with existing technologies, this invention overcomes the challenge of dynamically observing the pore topology of deep loaded rock masses. Through a limited number of indoor unit experiments, it reflects the true evolution law of the pore topology of deep rock masses, which is of great significance for revealing the mining-induced catastrophic events and stability assessment of deep energy-bearing rock masses. Attached Figure Description

[0034] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0035] Figure 1 This is a schematic diagram of the in-situ loading environment and sampling process of deep rock mass according to an embodiment of the present invention;

[0036] Figure 2 This is a schematic diagram of an isotope vacuum saturation tracer device according to an embodiment of the present invention;

[0037] Figure 3 This is a schematic diagram of the main structure of the loading device for tracking the dynamic evolution of pore topology under loading conditions in deep rock mass according to an embodiment of the present invention;

[0038] Figure 4 These are three views of a loading device for tracking the dynamic evolution of pore topology under loading conditions in deep rock masses, according to an embodiment of the present invention.

[0039] Figure 5 This is a schematic diagram of the structure of the loading device for tracking the dynamic evolution of pore topology under loading conditions in deep rock mass according to an embodiment of the present invention;

[0040] Figure 6 This is a schematic diagram of the composition and structure of the isotope detection and tracking device according to an embodiment of the present invention;

[0041] Figure 7 This is a schematic diagram of dynamic tracking of spatial deformation point cloud data of loaded rock mass according to an embodiment of the present invention;

[0042] Figure 8 This is a flowchart illustrating the three-dimensional reconstruction process of spatial point cloud information of a loaded rock mass according to an embodiment of the present invention.

[0043] Figure 9 This is a flowchart illustrating the intelligent inference process of spatial evolution for pore topology feature extraction based on a fully convolutional neural network, according to an embodiment of the present invention.

[0044] Figure 10This is a flowchart of a method according to an embodiment of the present invention;

[0045] The components include: 1. Vacuum pump; 2. Valve A; 3. Saturation cylinder; 4. Standard rock sample; 5. Pressure gauge; 6. Valve B; 7. Water tank; 8. Support base; 9. Horizontal loading device-I; 10. Horizontal loading device-II; 11. Circular guide rail; 12. Vertical loading device; 13. Vertical loading cylinder; 14. Vertical base; 15. Vertical moving support device; 16. Axial loading cylinder; 17. Vertical loading control system; 18. Horizontal loading cylinder; 19. Horizontal loading control system; 20. Graphite pad; 21. Copper pad; 22. Scintillation fluid; 23. Photomultiplier tube; 24. Pulse counter; 25. Isotope detection data tracking system; 26. Test standard rock sample; 27. Fluorescent particles; 28. Photocathode; 29. ​​Pulse signal; 30. β particles. Detailed Implementation

[0046] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0047] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0048] Pore ​​topology refers to a mathematical structure that studies the spatial shape and deformation of pore structures without considering specific measurements and distances. It focuses on properties such as connectivity, adjacency, and continuity of pores, rather than specific spatial measurements and distances.

[0049] Pore ​​topology has important applications in materials science, especially in the study of porous materials. Porous materials are essentially mesh-like structures composed of intersecting pillars and plates, with these small compartments packed and assembled together to fill space. The mechanical properties of porous materials can be determined by their composition, structure, and relative density, which is defined as the ratio of the density of the porous material to the density of a bulk material of the same composition. The topological configuration of porous materials has a significant impact on their mechanical properties. For example, the microstructure of open-cell foam materials consists of a mesh arrangement of interconnected pillars in three-dimensional space, while closed-cell foam materials contain plate-like surfaces of a certain thickness and length. These structures not only affect the mechanical properties of the materials but also determine their performance in practical applications.

[0050] Studying pore topology is crucial for understanding the fundamental physical meaning of unique effects observed in nanomaterials. By understanding these topologies, we can better design and fabricate materials with specific properties, thereby synthesizing metamaterials with tunable mechanical properties across various length scales.

[0051] This embodiment provides an experimental-based intelligent inference method for deep rock mass pore topology configuration based on data and physical fusion, including the following steps:

[0052] S1. Obtain standard rock samples from deep rock masses;

[0053] Determine the loading environment of deep rock masses and prepare standard rock samples through in-situ sampling. Specific implementation steps include: Figure 1 As shown, the rock mass in the deep engineering construction section is selected as the research object. The loading environment of the rock mass in the construction section is obtained by using the deep borehole hydraulic fracturing in-situ stress measurement method. The loading levels in the vertical and two horizontal directions are denoted as σ. V σ H σ h .

[0054] Furthermore, standard rock samples were obtained by drilling and processing in the laboratory. The dimensions of the standard rock samples were 100mm × 100mm × 200mm (length × width × height) to reflect the spatial distribution of pores in the rock mass.

[0055] S2. Mark the standard rock sample to obtain the test standard rock sample;

[0056] C-14 labeling of standard rock samples was performed using an isotope vacuum saturation tracer. Specific implementation steps included:

[0057] ① The standard rock sample was washed with deionized water to remove surface impurities before use.

[0058] ②Isotope labeling of standard rock samples was performed using an isotope vacuum saturation tracer.

[0059] As an additional implementation, this embodiment provides an isotope vacuum saturation tracer, such as... Figure 2 As shown. The experimental apparatus consists of a vacuum pump 1, valve A2, a saturation cylinder 3, a standard rock sample 4, a pressure gauge 5, a valve B6, and a water tank 7. The vacuum pump 1 is connected to valve A2, and valve A2 is connected to the saturation cylinder 3. The saturation cylinder 3 is used to hold the standard rock sample 4 and is connected to the pressure gauge 5 and valve B6 respectively. Valve B6 is connected to the water tank 7, which contains clean water labeled with C-14 isotope.

[0060] The specific operating procedure for the experiment is as follows: Open the saturation cylinder 3, place the standard rock sample 4 (after impurity treatment) into the saturation cylinder 3, leaving a certain gap between the standard rock samples 4 to ensure that the rock samples are fully saturated. After placement, close the saturation cylinder 3 and ensure it is sealed. Open valve A2 and close valve B6, turn on the vacuum pump 1 and observe the pressure gauge 5. Maintain the pressure gauge 5 reading at 0.4 MPa for 4 hours. After the requirement is met, open valve B6. At this time, the water containing C-14 isotope-labeled clean water in the water tank 7 will flow into the saturation cylinder 3 until the standard rock sample 4 is submerged. Then close valve B6 and continue to maintain the pressure gauge 5 reading at 0.4 MPa for 4 hours until no bubbles are observed escaping from the standard rock sample 4. Close the vacuum pump 1 and valve A2, and let it stand for another 4 hours before taking out the standard rock sample 4.

[0061] ③ The marked standard rock samples are solidified and dried to ensure that the C-14 isotopes are uniformly and stably distributed.

[0062] The test standard rock samples labeled with C-14 isotopes can be obtained by using the methods described in steps 1 to 3 of step 2.

[0063] S3. Conduct a load-bearing catastrophic test on the aforementioned standard rock sample to obtain the deep rock mass mining process and test data;

[0064] As an additional implementation, this embodiment incorporates a data-physical fusion method. The isotopic rock mass obtained in step S2 is used to capture point cloud data using physical techniques in step S3, and then coordinate positioning is performed based on geometric mathematical relationships.

[0065] Standard rock sample loading and disaster testing is conducted to simulate the mining process in deep rock masses and to receive target test data. Specific implementation steps include: Considering the loading environment and mining disaster characteristics of deep rock masses, this embodiment provides a dynamic evolution tracking loading device for pore topology configuration under loading conditions in deep rock masses, such as... Figures 3-6 As shown. The main structure of the test loading device is as follows. Figure 3 As shown, it includes: a support base 8, a horizontal loading device-Ⅰ9, a circular guide rail 11, a vertical loading device 12, and a vertical loading cylinder 13.

[0066] Specifically, the three views of the test loading device are as follows: Figure 4 As shown, it includes: a support base 8, a horizontal loading device-Ⅰ9, a circular guide rail 11, a vertical loading device 12, a vertical loading cylinder 13, a vertical base 14, and a vertical moving support device 15.

[0067] Specifically, the experimental device is structurally composed as follows: Figure 5As shown, it includes: a support base 8, a horizontal loading device-Ⅰ9, a circular guide rail 11, a vertical loading device 12, a vertical loading cylinder 13, a vertical base 14, a vertical moving support device 15, an axial loading cylinder 16, a vertical loading control system 17, a horizontal loading cylinder 18, a horizontal loading control system 19, a graphite pad 20, a copper sheet pad 21, a scintillation fluid 22, a photomultiplier tube 23, a pulse counter 24, an isotope detection data tracking system 25, and a test standard rock sample 26.

[0068] Furthermore, the test standard rock sample 26 has six faces, of which the left, front, and bottom faces are surrounded by graphite pads 20, and the right, top, and rear faces are surrounded by copper pads 21. During the test, the C-14 isotope in the test standard rock sample 26 will decay and release β particles 30. The graphite pads 20 block the propagation of β particles 30, while the β particles can pass through the copper pads 21.

[0069] Specifically, the graphite pads 20 and copper pads 21 corresponding to the six faces of the test standard rock sample 26 are connected to the axial loading cylinders 16 in the six directions, respectively. The axial loading cylinders 16 in the left and right directions are connected to the horizontal loading device-Ⅰ9, the axial loading cylinders 16 in the front and rear directions are connected to the horizontal loading device-Ⅱ10, and the axial loading cylinders 16 in the up and down directions are connected to the vertical loading device 12. The horizontal loading device-Ⅰ9, the horizontal loading device-Ⅱ10, and the vertical loading device 12 are all located on the support base 8.

[0070] Furthermore, the horizontal loading device-I9 and horizontal loading device-II10 contain horizontal loading cylinders 18, and the vertical loading device 12 contains a vertical loading cylinder 13. The horizontal loading control system 19 is connected to the horizontal loading device-I9 and horizontal loading device-II10, and the horizontal loading control system 19 can control the horizontal loading cylinders 18 inside the horizontal loading device-I9 and horizontal loading device-II10 to apply σ respectively. H With σ h The horizontal stress is controlled by the vertical loading control system 17, which is connected to the vertical loading device 12, which is located on the vertical base 14. The vertical loading control system 17 can control the vertical loading cylinder 13 inside the vertical loading device 12 to apply σ. V The vertical stress can be simulated through the operation to simulate the real loading environment of deep rock masses.

[0071] Specifically, the horizontal loading device-I9, horizontal loading device-II10, and vertical loading device-12 appear in pairs and act on the six faces of the standard rock sample 26 respectively. Circular guide rails 11 are set at the bottom of the left horizontal loading device-I9 and the front horizontal loading device-II10, so that the horizontal loading device-I9 and horizontal loading device-II10 can move on the circular guide rails 11. The left horizontal loading device-I9 is ​​equipped with a vertical moving support device 15 to support the vertical loading device 12 above. When the left horizontal loading device-I9 moves, the vertical moving support device 15 can also move, which facilitates the installation and removal of the test standard rock sample 26.

[0072] As an additional implementation, an isotope detection and tracking device is included inside the axial loading cylinder 16 of the right-side horizontal loading device-Ⅰ9, the rear horizontal loading device-Ⅱ10, and the upper vertical loading device 12, and its composition structure is as follows: Figure 6 As shown, it includes: scintillation fluid 22, photomultiplier tube 23, pulse counter 24, isotope detection data tracking system 25, fluorescent particles 27, photocathode 28, pulse signal 29, and β particles 30.

[0073] As an additional implementation method, the working principle of the isotope detection and tracking device is as follows: through step two, the interior of the test standard rock sample 26 is uniformly labeled with C-14 isotope. During the shear test, the C-14 isotope inside the test standard rock sample 26 decays to produce β particles 30. The β particles 30 pass through the copper pad 21 and enter the scintillation fluid 22 and interact to form fluorescent particles 27. The fluorescent particles 27 enter the photomultiplier tube 23 through the photocathode 28. After amplification and processing, they are converted into pulse signals 29. The radiation intensity can be recorded and analyzed in the pulse counter 24. The results are finally displayed in the isotope detection data tracking system 25.

[0074] Furthermore, the implementation procedure for tracking the dynamic evolution of pore topology under deep rock mass loading conditions is as follows: Before the test begins, the left horizontal loading device-Ⅰ9 and the front horizontal loading device-Ⅱ10 are moved to a distance via the circular guide rail 11, and the test standard rock sample 26 is placed at the instrument loading position. Then, the left horizontal loading device-Ⅰ9 and the front horizontal loading device-Ⅱ10 are pushed to seal the test standard rock sample 26. At the start of the test, the horizontal loading control system 19 controls the horizontal loading devices-Ⅰ9 and Ⅱ10 to simultaneously apply σ to the test standard rock sample 26. H With σ h The initial horizontal stress is applied to the test standard rock sample 26 by the vertical loading device 12 under the control of the vertical loading control system 17. VInitial vertical stress and initial horizontal stress are applied simultaneously. After the initial horizontal and vertical stresses are applied, the sample is left to stand for 30 minutes, and the initial isotope radiation data of the test standard rock sample 26 is recorded by the isotope detection data tracking system 25. The vertical loading control system 17 controls the vertical loading device 12 to gradually increase the vertical stress on the standard rock sample 26 until the standard rock sample 26 is damaged by the load. The isotope detection data tracking system 25 continuously collects the radiation data of the test standard rock sample 26 during the test. At the end of the test, the horizontal loading control system 19 controls the horizontal loading device-Ⅰ9 and the horizontal loading device-Ⅱ10 to unload the horizontal stress on the standard rock sample 26, and the vertical loading control system 17 controls the vertical loading device 12 to unload the vertical stress on the standard rock sample 26. The damaged test standard rock sample 26 is then removed and the test data is saved, thus ending the test.

[0075] S4. Based on the deep rock mass mining process and test data, obtain point cloud information of the loading process of the test standard rock sample; perform three-dimensional reconstruction of the test standard rock sample based on the point cloud data to obtain the spatial dynamic evolution of the loading behavior of the deep rock mass;

[0076] As an additional implementation method, point cloud information of the loading process of standard rock samples is acquired in real time, and the three-dimensional loaded rock mass is synchronously reconstructed and the evolution of pore topology is tracked in real time by solving the Poisson equation. Specific implementation steps include: uniformly distributing the same radiation source intensity onto the test standard rock sample obtained in step two; with the radiation source intensity remaining constant, the radiation intensity measured at any location within the test standard rock sample is different. Based on the radiation intensity measured at different locations in step three, the spatial point cloud coordinates of any location within the test standard rock sample are calculated. The principle is as follows: Figure 7 As shown. Based on the size data (100mm×100mm×200mm) of the test standard rock sample described in step three, the spatial coordinates of the three isotope detection and tracking devices can be obtained as (x1,y1,z1)=(50,0,100), (x2,y2,z2)=(0,50,100), and (x3,y2,z3)=(50,50,200).

[0077] As an additional implementation method, let the spatial coordinates of any point inside the test standard rock sample 26 be (x... i ,y i ,z i The intensity of the radioactive source at that point is denoted as I. i0 During the experiment, the radiation intensity at this point was captured by three isotope detection and tracking devices, denoted as I1, I2, and I3, respectively. Based on the relationship between the signal intensity of isotope decay and distance, the following can be obtained:

[0078]

[0079] Among them: I i0 The coordinates are (x i ,y i ,z i The intensity of the radioactive source at point d i1 d i2 d i3 These represent the distances between the three isotope detectors and the radioactive source, respectively. I1, I2, and I3 represent the distances between the radioactive source and the radioactive source at distances d, respectively. i1 d i2 d i3 The radiation intensity measured at the location.

[0080] Furthermore, the spatial coordinates (x, y) can be obtained using the above formula. i ,y i ,z i The distance between the detector and the three detectors is:

[0081]

[0082] Furthermore, based on spatial geometric relationships, the spatial coordinates (x... i ,y i ,z i The distance between the detector and the three detectors can also be expressed as:

[0083]

[0084] Where: (x1,y1,z1), (x2,y2,z2), and (x3,y2,z3) are the spatial coordinates of a known isotope detection and tracking device, (x i ,y i ,z i () represents the spatial coordinates of any point inside the test standard rock sample.

[0085] By combining the above formulas, the spatial coordinates of any point inside the test standard rock sample can be calculated as (x... i ,y i ,z i The test standard rock sample is considered to be composed of n spatial points. Based on the real-time radiation intensity data capture during the test, the spatial point cloud data information of the test standard rock sample can be obtained in real time.

[0086] As an additional implementation method, spatial point cloud data is used to reconstruct a three-dimensional model of the test standard rock sample through geometric mathematical relationships. The implementation process is as follows: Figure 8 As shown, the specific process is as follows:

[0087] (1) Based on the above method, obtain the coordinate data of n spatial point clouds inside the test rock mass sample, denoted as P(x i ,yi ,z i ), i = 1, 2, 3…n.

[0088] (2) Preprocessing of point cloud data of test rock mass samples, including classification, noise reduction and simplification of point cloud data. Based on the characteristics of point cloud density and arrangement, the three-dimensional point cloud data of rock mass samples during loading process is divided into scattered point clouds.

[0089] Furthermore, the original point cloud data was denoised using statistical methods. The coordinates P(x) of the three-dimensional point cloud data of the rock mass sample were standardized. i ,y i ,z i Standardizing the values ​​of i = 1, 2, 3…n yields P(x'). i ,y' i ,z' i ), i = 1, 2, 3…n, the calculation method is as follows:

[0090]

[0091] Where: (x i ,y i ,z i Let (x') be the spatial coordinates of any point inside the test standard rock sample. i ,y' i ,z' i ) represents the standardized coordinates, μ x μ y μ z These are the mean values ​​in the three coordinate directions, σ. x σ y σ z These represent the standard deviations in the three coordinate directions.

[0092] Furthermore, k-nearest neighbor search is used to calculate P(x') i ,y' i ,z' i The k nearest points M in the neighborhood k (P) = {P1, P2, ..., P} k The local density ρ(P) of each point is calculated using kernel density estimation (KDE), and a threshold ρ is set based on the local density ρ(P). th Points below the threshold are considered noise, and low-density points are removed from the point cloud to obtain the denoised point cloud data P. clean .

[0093] Furthermore, the denoised point cloud data P clean Simplification processes were implemented to improve the efficiency of three-dimensional point cloud reconstruction of rock mass samples. Based on the point cloud coordinate data P... cleanDetermine the coordinates of the 8 vertices of the minimum bounding box, divide the minimum bounding box into smaller grids, calculate the centroid of all contained points within each smaller grid region, and use spatial Euclidean distance to calculate the distance D from the midpoint of each point within each smaller grid region. i The centroid of a small grid region is determined and replaced with all points within that region to simplify the original rock mass sample point cloud data. The preprocessed 3D point cloud data is denoted as P. new (x' i ,y' i ,z' i ).

[0094] (3) Calculate point cloud data P new (x' i ,y' i The vector field z'i is used to construct the Poisson equation. By solving the Poisson equation and extracting isosurfaces, the three-dimensional model of the rock mass sample is finally reconstructed. The k-nearest neighbor search is used to calculate P. new (x' i ,y' i ,z' i The k nearest points M in the neighborhood k (P new )={P new1 ,P new2 …P newk The mean of the neighboring points is calculated based on their coordinate information. Given a covariance matrix C, perform eigenvalue decomposition on C to obtain eigenvalues ​​λ1, λ2, λ3 and their corresponding eigenvectors. Select the smallest eigenvector As the normal vector n i The normal vector is then normalized, and the calculation method is as follows:

[0095]

[0096] in: For the normalized normal vector, n i Let n be the normal vector. i | represents the magnitude of the normal vector.

[0097] Furthermore, the Poisson equation is constructed. The scalar field φ is obtained by discretizing the Poisson equation using the finite difference method. The discretized equation is then rewritten as a linear system of equations Aφ = b. The scalar field φ can be obtained by solving the system of equations.

[0098] in: To represent the divergence operation, φ is a scalar field. Let A be the normalized normal vector, A be the coefficient matrix, and b be the source term vector.

[0099] Furthermore, the Marching Cubes method is used to extract isosurfaces from the scalar field φ. The three-dimensional space is divided into an N×N×N cubic mesh, with each cube representing a voxel. Each cube has 8 vertices, and the scalar value S for each cube is calculated. i =φ(V i ), i = 0, 1, 2...7, according to the scalar value S i An index is constructed based on the relationship with the isosurface value T. Linear interpolation is performed between two vertices to determine the precise position of the triangle. Triangles generated in all voxels are combined into a complete 3D mesh.

[0100] Wherein: S i V is the scalar value for each vertex of the cube. i For each vertex of the cube, T is the isosurface value.

[0101] Since only the rock mass sample was isotopically labeled, while the pore portion inside the sample was not, the radiation intensity in the pores was relatively low during the experiment. Specifically, the radiation intensity at the pore location coordinates was 0, while the radiation intensity was present at the coordinates of the solid portion of the rock mass sample. Based on the radiation intensity values, the internal solid portion of the rock mass sample and its pore topology could be distinguished and observed in real time.

[0102] By using the method for obtaining three-dimensional point cloud data of rock mass described in step four and steps (1) to (3), real-time three-dimensional reconstruction of rock mass under load and dynamic tracking of pore topology can be achieved.

[0103] S5. Based on the spatial dynamic evolution, feature extraction is performed on the pore topology of the deep loaded rock mass to obtain dynamic evolution features.

[0104] This study utilizes image processing technology and a fully convolutional neural network method to refine and intelligently infer the pore topological configuration features of loaded rock masses. The specific implementation steps include: dividing the three-dimensional reconstruction model of the test rock mass sample into m 100×100mm two-dimensional image samples along the height direction, forming a two-dimensional sample image library {1, 2, 3, ..., m} for a single rock mass sample three-dimensional reconstruction model. The two-dimensional images contain both the solid portion and the pore structure portion of the rock mass sample. Binarization thresholding processing is performed on the two-dimensional sample images to obtain an updated two-dimensional sample image library {1', 2', 3', ..., m'}.

[0105] Furthermore, a fully convolutional neural network is used to extract pore topological configuration features and intelligently infer spatial evolution of the two-dimensional updated sample image library {1', 2', 3', ..., m'}. The implementation process is as follows: Figure 9As shown, the two-dimensional update sample image library {1', 2', 3', ..., m'} is divided into a two-dimensional training sample image library and a two-dimensional test sample image library. The two-dimensional training sample image library is used as the input layer to the convolutional layer of the fully convolutional neural network. Multiple convolutional kernels are used to extract the pore features of the rock mass. The calculation method for the convolution operation is as follows:

[0106]

[0107] Where: f is the input feature map, g is the convolution kernel, (i,j) is the position coordinate in the output feature map, m and n are the row and column indices of the convolution kernel as it slides on the input feature map, and M and N are the height and width of the convolution kernel.

[0108] Furthermore, the dimensionality of the feature map is reduced by a pooling layer to preserve the image's pore distribution features. The pooling operation formula is as follows:

[0109]

[0110] Where: f is the input feature map, R is the region of the pooling window, and m and n are the row and column indices of the convolution kernel as it slides across the input feature map.

[0111] Furthermore, the feature map is upsampled to the original input size using transposed convolution. The formula for the transposed convolution operation is:

[0112]

[0113] Where: f is the input feature map, g is the convolution kernel, (i,j) is the position coordinate in the output feature map, and m and n are the row and column indices of the convolution kernel as it slides on the input feature map.

[0114] Furthermore, the softmax activation function is used for classification, which is then used as the output of the last convolutional layer. The formula for the output layer operation is as follows:

[0115]

[0116] in: Here, f(x) represents the output probability, f(x) represents the network's original output score, and k represents the class index.

[0117] The above operations enable real-time extraction of the porosity distribution and spatial dynamic evolution characteristics of rock mass samples, ultimately outputting a convolved two-dimensional training sample image library {1”, 2”, 3”, ..., m”}. The pore topology and porosity values ​​of individual two-dimensional images of the test rock mass samples are obtained through {1”, 2”, 3”, ..., m”}. The porosity calculation method is as follows:

[0118]

[0119] Where: n rock To test the porosity of the rock mass, S pore S represents the pore area of ​​the rock mass. entity Area of ​​the solid portion of the rock mass.

[0120] Furthermore, the two-dimensional test sample image library is input into the fully convolutional neural network (WCNN), allowing the WCNN to undergo continuous iterative training and learning to improve the prediction level of the WCNN model. The rock mass sample parameters are then input into the trained WCNN model, including the basic physical and mechanical parameters of the rock mass test and the in-situ stress level σ of the rock mass sample obtained in step one. V σ H With σ h The trained model is used to intelligently deduce the evolution of pore topology configuration during the loading process of rock mass samples and output the pore topology configuration distribution.

[0121] The method implementation steps described in this embodiment are as follows: Figure 10 As shown, the above steps can be used to realize intelligent inference based on the pore topology evolution tracking test of deep loaded rock mass.

[0122] The above are merely preferred embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for intelligent inference of deep rock mass pore topology based on experimental data and physical data, characterized in that, Includes the following steps: Obtain standard rock samples from deep rock masses; The standard rock samples were marked to obtain the test standard rock samples; Marking the standard rock samples includes: A standard rock sample was C-14 labeled using an isotope vacuum saturation tracer device; wherein the isotope vacuum saturation tracer device includes: a vacuum pump, valve A, a saturation cylinder, a standard rock sample, a pressure gauge, valve B, and a water injection tank; The test standard rock samples were subjected to a load-bearing catastrophic test to obtain the deep rock mass mining process and test data; Based on the deep rock mass mining process and test data, point cloud information of the loading process of the test standard rock sample is obtained through physical methods; based on the point cloud information, the test standard rock sample is reconstructed in three dimensions using mathematical methods to obtain the spatial dynamic evolution of the loading behavior of the deep rock mass; Obtaining the point cloud information of the loading process of the test standard rock sample includes: Based on the deep rock mass mining process and test data, the spatial distance between the test standard rock sample and the isotope detection and tracking device was obtained. Calculate the spatial coordinates of the test standard rock sample based on the spatial distance; Based on the spatial coordinates, the spatial point cloud information of the loading process of the test standard rock sample is obtained; Based on the aforementioned spatial dynamic evolution, the pore topology of deep loaded rock masses is feature-extracted to obtain dynamic evolution features; Feature extraction of pore topology in deep loaded rock masses includes: Based on the spatial dynamic evolution, a two-dimensional sample image of the test standard rock sample is obtained; A fully convolutional neural network is used to extract pore topology features from the two-dimensional sample images to obtain dynamic evolution features.

2. The experimental-based intelligent inference method for deep rock mass pore topology configuration based on data and physical fusion according to claim 1, characterized in that, The load-bearing catastrophic test on the aforementioned standard rock samples includes: A dynamic evolution tracking loading device was used to conduct a load-bearing catastrophic test on the test standard rock sample; wherein the dynamic evolution tracking loading device includes: a support base, a horizontal loading device-I, a horizontal loading device-II, a circular guide rail, a vertical loading device, and a vertical loading cylinder.

3. A data-physical fusion intelligent inference system for deep rock mass pore topology based on experiments, characterized in that, The system includes: The rock sample acquisition module is used to acquire standard rock samples from deep rock masses; A rock sample marking module is used to mark the standard rock sample to obtain the test standard rock sample; an isotope vacuum saturation tracer is used to mark the standard rock sample with C-14; wherein, the isotope vacuum saturation tracer includes: a vacuum pump, valve A, a saturation cylinder, a standard rock sample, a pressure gauge, valve B and a water injection tank; The rock sample testing module is used to conduct load-bearing catastrophic tests on the test standard rock samples to obtain the deep rock mass mining process and test data. The three-dimensional reconstruction module is used to obtain point cloud information of the loading process of the test standard rock sample based on the deep rock mass mining process and test data; obtain the spatial distance between the test standard rock sample and the isotope detection and tracking device based on the deep rock mass mining process and test data; calculate the spatial coordinates of the test standard rock sample based on the spatial distance; obtain the spatial point cloud information of the loading process of the test standard rock sample based on the spatial coordinates; and perform three-dimensional reconstruction of the test standard rock sample based on the point cloud information to obtain the spatial dynamic evolution of the loading behavior of the deep rock mass. The feature extraction module is used to extract features of the pore topology of the deep loaded rock mass based on the spatial dynamic evolution to obtain dynamic evolution features; to obtain two-dimensional sample images of the test standard rock samples based on the spatial dynamic evolution; and to extract pore topology features from the two-dimensional sample images using a fully convolutional neural network to obtain dynamic evolution features.

4. A computer device comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the steps of the method according to any one of claims 1-2.

5. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the steps of the method according to any one of claims 1-2.

6. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the steps of the method according to any one of claims 1-2.

Images