Layered rock mass numerical simulation method based on mixed progressive failure model

Abstract

The invention relates to the technical field of stratified rock mass numerical simulation, and discloses a stratified rock mass numerical simulation method based on a hybrid progressive failure model, and the method comprises the steps: obtaining an engineering rock mass structural plane of a research area, and grading geological structural planes such as faults, rock veins and fractures of surrounding rock of an underground cavity based on structural plane features; through three-dimensional laser scanning, obtaining the spatial three-dimensional coordinates of each sampling point on the surface of the footrill, obtaining the point cloud of a measured object, and connecting adjacent discrete points; carrying out secondary sampling on the spliced point cloud data, carrying out plane fitting to obtain a structural plane occurrence, and constructing a footrill simulation model; analyzing the occurrence of the structural surface by the footrill simulation model to obtain a stratified rock mass value, and dividing the structural surface in the rock mass into a rigid structural surface and a weak structural surface according to the character, weathering degree, tightness degree and filler condition; and outputting the data to a cloud platform. According to the invention, manual intervention is reduced, and the safety, accuracy and efficiency of measurement are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of layered rock mass numerical simulation, and in particular to a layered rock mass numerical simulation method based on a mixed progressive failure model. Background Art

[0002] Due to its anisotropy and heterogeneity, the failure modes of layered rock masses are complex and diverse. Studies have shown that the failure modes of layered rock masses include failure along joint planes, rotational block failure, step failure, and mixed failure. These failure modes are closely related to parameters such as joint inclination, joint step angle, layer spacing, and rock bridge length. Therefore, numerical simulation needs to be able to capture these complex failure mechanisms and provide accurate mechanical response analysis. At present, numerical simulation methods mainly include finite element method (FEM), discrete element method (DEM), and their hybrid forms (FEM / DEM). RFPA (Real Failure Process Analysis) is a numerical simulation method based on statistical damage constitutive relations, which can simulate the nonlinear, heterogeneous, and anisotropic failure process of rocks. In addition, software such as FLAC3D and 3DEC are also widely used in the numerical simulation of layered rock masses. By introducing joint parameters and meshing technology, the failure modes of rock masses can be effectively simulated.

[0003] Prior art 1, a Chinese patent with the patent number 202311435002.9, discloses a rock mass change prediction system based on rock mass fracture seepage information. This system relates to the field of geological engineering and is used to improve the problem of current rock mass change prediction systems being significantly influenced by subjective factors, resulting in high uncertainty in prediction results. The prediction system includes a data acquisition module, a data processing module, a model building module, and a change prediction module. The data acquisition module acquires basic prediction data, the data processing module acquires prediction model data based on the basic prediction data, the model building module receives the prediction model data and establishes a rock mass change prediction model. The change prediction module acquires rock mass change prediction classification data based on the rock mass change prediction model and formulates a response strategy. By acquiring the basic prediction data and the prediction model data and establishing a rock mass change prediction model, rock mass changes are predicted. Although this improves the certainty of rock mass change prediction results, numerical simulation methods still have shortcomings in multi-field coupling analysis, resulting in inaccurate numerical values.

[0004] Prior art two, a Chinese patent with patent number 202510187700.4, relates to the field of image data processing technology and discloses a large-scale dangerous rock mass identification method based on digital image deformation processing. This method aims to address the low efficiency and accuracy of existing methods. The solution mainly includes: obtaining two images of the target rock mass collected within two preprocessing cycles, and generating two-dimensional grayscale matrices of the two images in the spatial domain; performing a two-dimensional discrete Fourier transform on the two-dimensional grayscale matrices of the two images, calculating the cross-correlation function of the two images, calculating the first phase difference corresponding to the frequency domain based on the cross-correlation function, and determining the second phase difference corresponding to the spatial domain based on the first phase difference; obtaining a phase-demodulated displacement field based on the second phase difference, and calculating the displacement gradient and strain field based on the displacement field; and determining the hazard level of each area of ​​the target rock mass based on a pre-trained dangerous rock mass identification model. Although this method improves the efficiency and accuracy of dangerous rock mass identification and is particularly suitable for large-scale dangerous rock mass identification, the numerical simulation method still has shortcomings in multi-field coupling analysis, resulting in inaccurate numerical values.

[0005] Prior art three, Chinese patent, patent number: 202411650810.1 relates to the field of rock mass analysis technology, in particular, a method and system for analyzing the degradation mechanism of the mechanical properties of limestone in karst areas. The method comprises the following steps: obtaining limestone from the karst area to construct a limestone test sample; obtaining test results based on testing the limestone test sample; observing the limestone test sample to obtain microscopic characteristic observation results of the limestone test sample; obtaining microscopic characteristic analysis results of the limestone test sample based on analysis of the microscopic characteristic observation results; and analyzing the degradation mechanism of the mechanical properties of limestone by combining the test results and the microscopic characteristic analysis results. Although the dynamic dissolution test is constructed to more accurately fit the rock mass degradation in the natural environment, the variation law of the mechanical property parameters of limestone is studied based on the dynamic dissolution test, and the degradation mechanism of the mechanical properties of limestone in karst areas is analyzed, providing reliable theoretical and technical support for engineering applications; however, the numerical simulation method still has deficiencies in multi-field coupling analysis, resulting in inaccurate numerical values.

[0006] Currently, the numerical simulation methods of the prior art 1, prior art 2, and prior art 3 still have deficiencies in multi-field coupling analysis, resulting in inaccurate numerical values. To address the above problem, the present invention provides a numerical simulation method for layered rock mass based on a hybrid progressive failure model. Summary of the Invention

[0007] The main purpose of the present invention is to provide a numerical simulation method for layered rock mass based on a hybrid progressive failure model to solve the problem that the prior art still has deficiencies in multi-field coupling analysis, resulting in inaccurate numerical values.

[0008] To achieve the above object, the present invention provides the following technical solutions:

[0009] A layered rock mass numerical simulation method based on a hybrid progressive failure model, comprising:

[0010] Obtain the structural surfaces of the engineering rock mass in the study area and classify them into primary structural surfaces, tectonic structural surfaces, and superficial structural surfaces according to their scale and surface characteristics; classify the fault, dyke, and fissure geological structural surfaces of the surrounding rock of the underground caverns based on the structural surface characteristics;

[0011] Through 3D laser scanning, the spatial coordinates of each sampling point on the tunnel surface are obtained to obtain the point cloud of the measured object, and the adjacent discrete points are connected; the spliced ​​point cloud data is resampled and plane fitting is performed to obtain the structural surface occurrence and construct the tunnel simulation model;

[0012] The adit simulation model analyzes the dip of the structural surface and obtains the numerical value of the layered rock mass. The structural surfaces in the rock mass are divided into rigid structural surfaces and weak structural surfaces according to their properties, degree of weathering, density and filling conditions; and the data is output to the cloud platform.

[0013] As a further improvement of the present invention, the process of classifying the fault, vein and fissure geological structural surfaces of the surrounding rock of the underground cavern based on the structural surface characteristics includes the following steps:

[0014] Obtain the extension length of the engineering rock mass structural surface and the width of the fracture zone in the study area, as well as the scale and engineering geological impact degree, and classify them into level I structural surface, level II structural surface, level III structural surface, level IV structural surface and level V structural surface;

[0015] According to geological composition, they are divided into primary structural surface, tectonic structural surface and superficial structural surface; according to mechanical properties, they are divided into rigid structural surface and weak structural surface;

[0016] Among them, primary structural surfaces include bedding; tectonic structural surfaces include faults and joints; superficial structural surfaces include weathering cracks; rigid structural surfaces include closed and unfilled; weak structural surfaces include weak interlayers or fillings;

[0017] The parameters of the level I and level II structural surfaces are obtained by measurement to obtain the boundary conditions of the project stability; the level III and below structural surfaces are statistically processed to comprehensively analyze the development degree, bonding degree and control effect of the structural surfaces on the project stability.

[0018] As a further improvement of the present invention, the process of obtaining the occurrence of the structural surface includes the following steps:

[0019] Through 3D laser scanning, the original 3D point cloud data is cut using a cutting tool to remove invalid point cloud data, and the 0-6 meter point cloud scanning data of the horizontal tunnel is obtained. Then, the 3-9 meter point cloud cutting data of the horizontal tunnel is obtained through iteration.

[0020] Select multiple feature points from the overlapping parts of two point cloud data respectively, and automatically stitch them together based on the previous point cloud data. After the stitching is completed, compare it with the preset error threshold. If the comparison is greater than the preset error threshold, stitching is performed again.

[0021] The spliced ​​point cloud data is resampled to adjust the density between point clouds; the density between the point clouds after hanging is plane fitted, and three points are selected to fit the structural surface; the normal vector of the selected plane is calculated to obtain the occurrence of the first structural surface; and a horizontal tunnel simulation model is constructed.

[0022] As a further improvement of the present invention, the process of removing invalid point cloud data from the original three-dimensional point cloud data by using a cutting tool includes the following steps:

[0023] Obtain original point cloud data through 3D laser scanning, and use the cutting tool to directly cut out invalid areas; intercept valid point cloud data within the range of 0-6 meters, and eliminate redundant point cloud data outside the range; redundant point cloud data includes surrounding environment and noise points;

[0024] A dynamic threshold is set to automatically identify and remove invalid points. By traversing the laser points, the distance between the current point and adjacent points is compared. If the distance exceeds the preset threshold, it is determined to be an invalid noise point and removed. The preset threshold includes discrete points caused by noise or abnormal reflections.

[0025] After each scan, the range of the cutting tool is adjusted to three meters. By adjusting the coverage area of ​​the cutting tool multiple times, continuous point cloud data is finally spliced ​​together to form 3-9 meter point cloud cutting data of the horizontal tunnel.

[0026] As a further improvement of the present invention, the process of automatic splicing includes the following steps:

[0027] Select multiple feature points in the overlapping area of ​​two adjacent point cloud data as matching benchmarks; wherein the multiple feature points include local point sets of geometric features such as corner points and edge points;

[0028] Using the coordinate system of the previous point cloud as a reference, calculate the cylinder transformation matrix from the current point cloud to the reference point; minimize the sum of the spatial values ​​of the corresponding feature points in the two point clouds; after the stitching is completed, calculate the matching error between the transformed point clouds;

[0029] If the error exceeds the preset threshold, the feature point selection is readjusted; when the error exceeds the limit, re-alignment is performed by dynamically adjusting the number or position of feature points.

[0030] As a further improvement of the present invention, the process of obtaining the occurrence of the first structural surface includes the following steps:

[0031] The spliced ​​point cloud data is subsampled and the data scale is optimized by adjusting the density between point clouds. The point cloud data is segmented and the structural surface is segmented by modifying the parameters using the normal vector difference and eigenvalue.

[0032] Select three points to fit the structural surface plane, and calculate the normal force of the three-point fitting mechanism surface plane to obtain the inclination and inclination angle of the structural surface; dynamically segment different structural surfaces and eliminate segmentation phenomena by plane merging;

[0033] The fitting results are monitored for flatness, and the root mean square of the distance between the point planes and the root mean square of the normal vector angle are calculated. A three-dimensional grid model of the tunnel is generated based on the segmented structural surface data, and the three-dimensional grid model is processed to obtain a tunnel simulation model.

[0034] As a further improvement of the present invention, the process of generating a three-dimensional grid model of the tunnel based on the segmented structural surface data includes the following steps:

[0035] The flatness of the structural surface is evaluated by calculating the root mean square of the distance from the point cloud to the fitting plane and the root mean square of the normal vector angle; if the error exceeds the threshold, the plane fitting or segmentation parameters are readjusted;

[0036] Based on the segmented structural surface data, the normal vector difference and eigenvalue correction parameters are used to refine the segmentation and eliminate redundant point clouds; the structural surface is meshed and the discrete points are connected into triangular patches to form a continuous surface;

[0037] The local grids of each structural surface are spliced ​​together to process the spatial relationship between different structural surfaces; a three-dimensional grid model including elements such as the terrain surface and the rock layer interface is constructed, and material properties of different lithologic units are assigned. Based on the three-dimensional grid model, a tunnel simulation model is constructed.

[0038] As a further improvement of the present invention, the process of constructing a tunnel simulation model based on a three-dimensional grid model includes the following steps:

[0039] The point cloud data in the 3D mesh model is subjected to point cloud denoising, discrete point removal, and data redundancy reduction preprocessing operations to form a data set; the data set is divided into a training set, a validation set, and a test set; and a tunnel simulation model is constructed based on the 3D network model;

[0040] Process the 3D point cloud and use the pre-processed point cloud data as a tunnel simulation model, outputting the predicted structure of the structural surface parameters. By inputting small batches of data, the difference between the forward propagation results and the labels is calculated and the model parameters are updated.

[0041] The point cloud data is input into the adit simulation model and simulated; the simulation results are analyzed to obtain the degree of structural surface opening, groundwater development on the structural surface, structural surface filling conditions, and structural surface filling shape classification.

[0042] As a further improvement of the present invention, the process of obtaining the structural surface opening degree, the structural surface groundwater development situation, the structural surface filling situation and the structural surface filling shape classification includes the following steps:

[0043] The pre-processed point cloud data is input into the tunnel simulation model to generate a continuous triangulated network model and form a three-dimensional geometric expression of the structural surface; the degree of opening is quantified by calculating the spatial distance or displacement between adjacent structural surfaces;

[0044] Identify the edges of structural surfaces and areas of curvature change. High areas correspond to cracks with openings greater than a threshold. If the point cloud contains reflectivity or color information, analyze the abnormal reflectivity values ​​in specific areas to infer the presence of groundwater. If there are depressions or holes in the grid model, it is determined to be groundwater storage space.

[0045] Identify the point cloud density mutation or texture continuity terminal, and judge it as the filling area, and separate the filling material from the bedrock in the filling area; parameterize the shape of the filling area and classify the filling material by region; and identify regular or irregular filling shapes by analyzing the grid connectivity of the filling area.

[0046] As a further improvement of the present invention, the process of obtaining the numerical value of the layered rock mass includes the following steps:

[0047] Based on the 3D point cloud data in the adit simulation model, the occurrence information of each structural surface is calculated through the normal vector, and the normal vector of the point cloud is converted into actual geographic coordinates; the structural surfaces are integrated through proximity search, and a 2D projection surface is established to generate projection lines;

[0048] The distances between the intersections of the orthogonal survey lines and the projection lines are calculated, and the average is taken as the spacing between the structural surfaces. An independent coordinate system is assigned to the structural surface, and the extension length in the main direction is calculated to quantify the extension degree of the structural surface. The structural surface is intelligently segmented and divided into different groups based on the normal vector difference and characteristic final value.

[0049] The occurrence distribution of each group is counted, and the average inclination angle and spacing interval values ​​are calculated; compared with the actual measurement results, if the occurrence angle error is greater than the preset value, the segmentation is re-performed.

[0050] This method classifies engineering rock mass structural surfaces into primary, tectonic, and superficial structural surfaces based on their scale and surface characteristics. It also classifies geological structural surfaces, such as faults, dykes, and fissures, in the surrounding rock of underground caverns. 3D laser scanning technology is used to obtain high-precision point cloud data of the adit surface, and a simulation model of the adit is constructed through plane fitting and secondary sampling. The adit simulation model analyzes the occurrence of the structural surfaces, classifying them into rigid and weak structural surfaces based on their properties, degree of weathering, density, and filling material. The data is then exported to a cloud platform. The entire process utilizes 3D laser scanning technology and automated algorithms, reducing manual intervention and improving measurement safety, accuracy, and efficiency. Furthermore, the non-contact measurement method avoids the safety hazards associated with traditional methods. 3D laser scanning technology enables rapid acquisition of high-precision point cloud data, significantly improving measurement efficiency and accuracy. Combining structural surface classification and grading with simulation model analysis can more comprehensively reveal the geological characteristics of the rock mass and their impact on engineering projects. Once the data is exported to the cloud platform, it can be easily shared and analyzed, providing real-time support for the design, construction, and monitoring of rock mass engineering projects. The non-contact measurement method avoids the safety hazards associated with traditional methods, effectively ensuring personnel safety, especially when conducting measurements in complex geological environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 1. A schematic flow chart of steps of an embodiment of a method for numerical simulation of layered rock mass based on a hybrid progressive failure model of the present invention;

[0052] Figure 2 This is a schematic flow chart of the steps for classifying geological structural surfaces such as faults, veins, and fissures in the surrounding rock of an underground cavern based on structural surface characteristics in one embodiment of a layered rock mass numerical simulation method based on a hybrid progressive failure model of the present invention;

[0053] Figure 3 This is a schematic flow chart of the steps for obtaining the occurrence of a structural plane in an embodiment of a numerical simulation method for layered rock mass based on a hybrid progressive failure model of the present invention;

[0054] Figure 4 This is a schematic flow chart of the steps of removing invalid point cloud data from original three-dimensional point cloud data using a shearing tool according to an embodiment of the layered rock mass numerical simulation method based on the hybrid progressive failure model of the present invention;

[0055] Figure 5 A schematic flow chart of the steps of automatically splicing a layered rock mass according to an embodiment of the present invention's numerical simulation method based on a hybrid progressive failure model;

[0056] Figure 6This is a schematic flow chart of the steps for obtaining the attitude of the first structural plane in an embodiment of the layered rock mass numerical simulation method based on the hybrid progressive failure model of the present invention;

[0057] Figure 7 This is a schematic flow chart of the steps of generating a three-dimensional mesh model of a tunnel based on segmented structural surface data in one embodiment of a layered rock mass numerical simulation method based on a hybrid progressive failure model of the present invention;

[0058] Figure 8 This is a schematic flow chart of the steps of constructing a tunnel simulation model based on a three-dimensional grid model in one embodiment of a layered rock mass numerical simulation method based on a hybrid progressive failure model of the present invention;

[0059] Figure 9 A schematic flow chart of the steps for obtaining the degree of structural plane opening, groundwater development on the structural plane, structural plane filling conditions, and structural plane filling shape classification according to an embodiment of the present invention's layered rock mass numerical simulation method based on a hybrid progressive failure model;

[0060] Figure 10 A schematic flow chart of the steps for obtaining layered rock mass numerical values ​​according to an embodiment of the layered rock mass numerical simulation method based on a hybrid progressive failure model of the present invention;

[0061] Figure 11 This is a schematic structural diagram of an embodiment of an electronic device of the present invention;

[0062] Figure 12 This is a schematic structural diagram of an embodiment of a storage medium of the present invention. DETAILED DESCRIPTION

[0063] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0064] The terms "first", "second" and "third" in the present invention are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first", "second" and "third" may explicitly or implicitly include at least one of the features. In the description of the present invention, the meaning of "multiple" is at least two, for example, two, three, etc., unless otherwise clearly and specifically defined. All directional indications in the embodiments of the present invention (such as up, down, left, right, front, back...) are only used to explain the relative position relationship, movement, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units that are not listed, or may optionally include other steps or units that are inherent to these processes, methods, products or devices.

[0065] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present invention. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute a separate or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0066] like Figure 1 As shown, this embodiment provides an embodiment of a layered rock mass numerical simulation method based on a hybrid progressive failure model. In this embodiment, the layered rock mass numerical simulation method based on a hybrid progressive failure model specifically includes the following steps:

[0067] Step S1: Obtain the structural surfaces of the engineering rock mass in the study area and classify them into primary structural surfaces, tectonic structural surfaces, and superficial structural surfaces according to their scale and surface characteristics; classify the geological structural surfaces such as faults, veins, and fissures in the surrounding rock of the underground caverns based on the structural surface characteristics;

[0068] Step S2: Obtain the spatial coordinates of each sampling point on the tunnel surface through 3D laser scanning to obtain the point cloud of the measured object, and connect adjacent discrete points; perform secondary sampling on the spliced ​​point cloud data, perform plane fitting, obtain the structural surface occurrence, and construct a tunnel simulation model;

[0069] Step S3: The adit simulation model analyzes the occurrence of the structural surface, obtains the layered rock mass value, and divides the structural surfaces in the rock mass into rigid structural surfaces and weak structural surfaces according to the properties, degree of weathering, density and filling conditions; and outputs the data to the cloud platform.

[0070] Preferably, in this embodiment, the structural surfaces of the engineering rock mass are classified into primary, tectonic, and superficial structural surfaces based on their scale and surface characteristics. Geological structural surfaces such as faults, dykes, and fissures in the surrounding rock of the underground cavern are also classified. 3D laser scanning technology is used to obtain high-precision point cloud data of the adit surface, and a simulation model of the adit is constructed through plane fitting and secondary sampling. The adit simulation model is used to analyze the occurrence of the structural surfaces, classifying the structural surfaces in the rock mass into rigid and weak surfaces based on their properties, degree of weathering, density, and filling material. The data is then exported to a cloud platform. The entire process utilizes 3D laser scanning technology and automated algorithms, reducing manual intervention and improving measurement safety, accuracy, and efficiency. Furthermore, the non-contact measurement method avoids the safety hazards associated with traditional methods. 3D laser scanning technology enables rapid acquisition of high-precision point cloud data, significantly improving measurement efficiency and accuracy. Combining structural surface classification and grading with simulation model analysis can more comprehensively reveal the geological characteristics of the rock mass and their impact on the project. Once the data is exported to the cloud platform, it can be easily shared and analyzed, providing real-time support for the design, construction, and monitoring of rock mass engineering projects. The non-contact measurement method avoids the safety hazards associated with traditional methods, effectively ensuring personnel safety, especially when conducting measurements in complex geological environments.

[0071] Furthermore, if Figure 2 As shown, the process of classifying geological structural surfaces such as faults, veins and fissures of the surrounding rock of the underground cavern based on structural surface characteristics in step S1 specifically includes the following steps:

[0072] Step S11: Obtain the extension length and fracture zone width of the engineering rock mass structural surface in the study area and the scale and engineering geological impact degree for classification; divide into level I structural surface, level II structural surface, level III structural surface, level IV structural surface and level V structural surface;

[0073] Step S12: Divide into primary structural surfaces, tectonic structural surfaces, and superficial structural surfaces according to geological composition; and divide into rigid structural surfaces and weak structural surfaces according to mechanical properties;

[0074] Among them, primary structural surfaces include bedding; tectonic structural surfaces include faults and joints; superficial structural surfaces include weathering cracks; rigid structural surfaces include closed and unfilled; weak structural surfaces include weak interlayers or fillings;

[0075] Step S13: Parameters of the level I and level II structural surfaces are obtained by measurement to obtain the boundary conditions of the engineering stability; level III and below structural surfaces are statistically processed; and the development degree, bonding degree and control effect of the structural surfaces on the engineering stability are comprehensively analyzed.

[0076] Preferably, in this embodiment, the structural surface is divided into five levels from I to V according to the extension length of the structural surface of the engineering rock mass, the width of the fracture zone and the degree of influence on the engineering geology. The structural surface is classified according to the geological composition (primary, structural, superficial) and mechanical properties (rigid, weak). For the structural surfaces of levels I and II, the parameters are obtained by direct measurement; for the structural surfaces of levels III and below, statistical processing is performed. By comprehensively analyzing the development degree, the degree of integration and the control effect on the stability of the project, the impact of the structural surface is comprehensively evaluated. Through grading and classification, the impact of structural surfaces on engineering stability can be more accurately identified and evaluated, especially in the direct measurement of Class I and Class II structural surfaces, which provides reliable boundary conditions for engineering design; through the classification and analysis of structural surfaces of different levels, corresponding engineering measures can be taken in a targeted manner, such as avoiding or treating weak structural surfaces, thereby reducing engineering risks and improving construction safety; for Class IV and Class V structural surfaces, although their impact on engineering stability is relatively small, their distribution and characteristics still need to be paid attention to in order to avoid potential geological disaster risks; this method combines macro- and micro-scale analysis, which can better reveal the distribution patterns and mechanical properties of structural surfaces, and provide technical support for the refined research of rock engineering.

[0077] Furthermore, if Figure 3 As shown, the process of obtaining the structural surface occurrence in step S2 specifically includes the following steps:

[0078] Step S21: Using a 3D laser scan, the original 3D point cloud data is cut using a shearing tool to remove invalid point cloud data, thereby obtaining 0-6 meter point cloud scan data of the horizontal tunnel; and iterating sequentially to obtain 3-9 meter point cloud shearing data of the horizontal tunnel;

[0079] Step S22: Select multiple feature points in the overlapping parts of the two point cloud data respectively, and automatically stitch them together based on the previous point cloud data; after the stitching is completed, compare it with the preset error threshold, and if the comparison is greater than the preset error threshold, stitch again;

[0080] Step S23: perform secondary sampling on the spliced ​​point cloud data to adjust the density between point clouds; perform plane fitting on the density between the point clouds after hanging the certificate, and select three points to fit the structural surface; calculate the normal vector of the selected plane to obtain the first structural surface attitude; and construct a horizontal tunnel simulation model.

[0081] Preferably, this embodiment obtains original point cloud data through three-dimensional laser scanning, and uses a cutting tool to remove invalid point cloud data, retaining the data of the target area (such as the 0-6 meter and 3-9 meter ranges of the tunnel); in the point cloud stitching process, multiple feature points of the overlapping parts of the two point cloud data are selected, and automatic stitching is performed based on the previous point cloud data; after the stitching is completed, a comparison is performed through a preset error threshold. If the error exceeds the threshold, the stitching is performed again, which embodies the idea of ​​iterative optimization; the stitched point cloud data is resampled and the point cloud density is adjusted to optimize the computational efficiency of subsequent processing; the structural surface is fitted by three points, and the normal vector of the selected plane is calculated to extract the occurrence of the first structural surface; finally, the tunnel simulation model is constructed using the stitched and fitted point cloud data; other technologies (such as holographic image fusion) may be combined in the model construction process to further improve the authenticity and usability of the model. Through point cloud stitching and iterative optimization, the continuity and consistency of the data are ensured, and the impact of stitching errors on the overall model is reduced; the extraction and matching of feature points improve the accuracy of stitching and avoid errors caused by manual operations; secondary sampling reduces the amount of point cloud data and reduces the computational burden of subsequent processing; plane fitting and normal vector calculation simplify the modeling process of complex terrain and improve the efficiency of model construction; the constructed tunnel simulation model can truly reflect the terrain characteristics and provide a reliable basis for subsequent engineering analysis, monitoring and decision-making; it is suitable for 3D modeling of various complex terrains, such as tunnels, mines and tunnels, and has strong versatility and scalability.

[0082] Furthermore, if Figure 4 As shown, the process of removing invalid point cloud data from the original three-dimensional point cloud data by using a cutting tool in step S21 specifically includes the following steps:

[0083] Step S211: Obtaining original point cloud data through 3D laser scanning, using a cutting tool to directly cut out invalid areas; intercepting valid point cloud data within a range of 0-6 meters, and removing redundant point cloud data outside the range; wherein, redundant point cloud data includes surrounding environment and noise points;

[0084] Step S212: Dynamic thresholds are set to automatically identify and remove invalid points. By traversing the laser points, the distance between the current point and adjacent points is compared. If the distance exceeds a preset threshold, the point is identified as an invalid noise point and removed. The preset threshold includes discrete points caused by noise or abnormal reflections.

[0085] Step S213: After each scan, the range of the shearing tool is adjusted to three meters. By adjusting the coverage area of ​​the shearing tool multiple times, continuous point cloud data is finally spliced ​​together to form 3-9 meter point cloud shearing data of the horizontal tunnel.

[0086] Preferably, this embodiment acquires raw point cloud data through 3D laser scanning and uses a cutting tool to directly crop out invalid areas, retaining only valid point cloud data within a range of 0-6 meters. This step eliminates redundant point cloud data that exceeds the range, including surrounding environment and noise points. During the traversal of the laser points, the distance between the current point and adjacent points is compared. If the distance exceeds a preset threshold, it is determined to be an invalid noise point and is eliminated. This method can automatically identify and remove discrete points caused by noise or abnormal reflections. Dynamic adjustment of the preset threshold can classify noise points based on the Euclidean distance between noise points and non-noise points, further optimizing the denoising effect. The range of the cutting tool is adjusted after each scan, and the coverage area is adjusted multiple times to ultimately form continuous point cloud data. By eliminating redundant point cloud data and noise points, interference factors in data processing are reduced, and the accuracy and reliability of the point cloud data are improved. The dynamic threshold denoising method can effectively identify and eliminate noise points, further improving the quality of the point cloud data. By repeatedly adjusting the coverage area of ​​the cutting tool and splicing the data, continuous point cloud data within a range of 3-9 meters of the adit is ultimately formed. By dynamically adjusting the preset threshold and stitching multiple scans, manual intervention is reduced and the automation and efficiency of data processing are improved; the dynamic threshold denoising method can reduce the time and computational complexity of subsequent processing while maintaining geometric features.

[0087] Furthermore, if Figure 5 As shown, the process of automatic splicing in step S22 specifically includes the following steps:

[0088] Step S221: selecting multiple feature points in the overlapping area of ​​two adjacent point cloud data as matching benchmarks; wherein the multiple feature points include local point sets of geometric features such as corner points and edge points;

[0089] Step S222: Using the coordinate system of the previous point cloud as a reference, calculate the cylinder transformation matrix from the current point cloud to the reference point; minimize the sum of the spatial values ​​of the corresponding feature points in the two point clouds; after the stitching is completed, calculate the matching error between the transformed point clouds;

[0090] Step S223: If the error exceeds the preset threshold, the feature point selection is readjusted; when the error exceeds the limit, the number or position of the feature points is dynamically adjusted to perform re-registration.

[0091] Among them, the calculation formula of the cylinder transformation matrix of the point cloud splicing in step S222 is:

[0092]

[0093] Where, P iand Pj represent the coordinate matrices of the i-th and j-th point clouds, respectively; T represents the cylinder transformation matrix, the parameter to be solved; λ represents the regularization coefficient, which adjusts the weight of the objective function; fi,k and fj,k represent the eigenvectors corresponding to the k-th feature point in the i-th and j-th point clouds, respectively; Rk(T) represents the rotation transformation acting on the k-th feature point under the transformation matrix T. In the least squares optimization framework, the first term |P i -T·Pj|2 2 Essentially, this is a Frobenius norm minimization problem in Euclidean space, derived from Gaussian least squares theory. Point cloud registration is transformed into a numerical optimization problem of finding the optimal rigid body transformation, ensuring that the overall deviation of the transformed point cloud is minimized. The rigid body transformation matrix T∈SE(3) belongs to a special Euclidean group, and its 6 degrees of freedom (3 rotations + 3 translations) ensures the physical feasibility of the transformation. The formula parameterizes the nonlinear rotation matrix through Lie algebra, making optimization methods such as gradient descent feasible. The Tikhonov regularization term introduced by the λ coefficient originates from the optimization theory of ill-posed problems. It prevents overfitting and improves the robustness of the algorithm by constraining the degree of alignment of the feature points after transformation (the second term |Rk(T)·fi,k-fj,k|). While traditional ICP only considers point distances, this formulation introduces geometric constraints through the eigenvectors fi,k. Corner features maintain local curvature invariance, while edge features maintain topological continuity. Eigenvectors can include descriptors such as normals and curvature. The regularization coefficient λ implements dual control: increasing λ to enhance feature matching when initial registration deviations are large, while decreasing λ during fine registration, prioritizing overall point cloud alignment, has been shown to achieve optimal results when λ = 0.1 to 0.5. Error assessment not only incorporates point distance (global accuracy) but also employs a dual check using eigenvector residuals (local consistency): translation error is typically required to be <10% of the point cloud spacing, and the rotation error threshold is often set to <0.5 degrees. Compared to traditional ICP, this solution is more robust in the following scenarios, offering optimized computational efficiency and multimodal adaptability.

[0094] Preferably, this embodiment selects multiple feature points as matching benchmarks in the overlapping area of ​​two adjacent point cloud data, including local point sets of geometric features such as corner points and edge points; uses the coordinate system of the previous point cloud as a reference to calculate the rigid body transformation matrix from the current point cloud to the benchmark point; after the stitching is completed, calculates the matching error between the transformed point clouds. If the error exceeds the preset threshold, the feature point selection is readjusted, and re-registration is performed by dynamically adjusting the number or position of feature points; the error evaluation may involve distance error, rotation error, and translation error between matching point pairs to ensure the accuracy of the final registration result. Through the precise selection of feature points and the optimized calculation of the rigid body transformation matrix, the displacement and rotation misalignment between point cloud data can be effectively reduced, thereby improving the registration accuracy; the mechanism of dynamically adjusting the number or position of feature points further enhances the robustness of the algorithm to noise and outliers, and avoids registration failures caused by improper initial feature point selection; by limiting the number and range of feature points, unnecessary redundant calculations and matching times are reduced, thereby improving the efficiency of the algorithm.

[0095] Furthermore, if Figure 6 As shown, the process of obtaining the occurrence of the first structural surface in step S23 specifically includes the following steps:

[0096] Step S231: performing secondary sampling on the spliced ​​point cloud data, optimizing the data scale by adjusting the density between point clouds; segmenting the point cloud data, and segmenting the structural surface by modifying the parameters using the normal vector difference and eigenvalue;

[0097] Step S232: Selecting a three-point fitting structural surface plane, and calculating the normal force of the three-point fitting structural surface plane to obtain the inclination and inclination angle of the structural surface; dynamically segmenting different structural surfaces, and eliminating segmentation phenomena by plane merging;

[0098] Step S233: monitor the flatness of the fitting results, calculate the root mean square of the distance between the point planes and the root mean square of the normal vector angle; generate a three-dimensional grid model of the tunnel based on the segmented structural surface data, process the three-dimensional grid model, and obtain a tunnel simulation model.

[0099] Among them, the structural surface plane normal force fitting formula in step S232 is:

[0100]

[0101] Where p1, p2, p3 represent the coordinates of three spatial points in the three-point fitting plane; n represents the normal force vector of the fitting plane;

[0102] The root mean square calculation formula for the flatness monitoring in step S233 is:

[0103]

[0104] Where, Indicates the i ′ The distance from the point to the fitting plane; θ j′ Indicates the jth ′ The angle between the normal vector of a point and the normal vector of the fitted plane; N represents the total number of points; M represents the total number of normal vectors. The mathematical essence of the normal vector calculation formula is the cross product property: the cross product of two in-plane vectors is used to obtain an orthogonal vector; normalization processing: the standard normal vector is obtained by L2 norm normalization; the right-hand rule: the relationship between the upper and lower plates of the structural surface is implicitly determined; structural geological mapping realizes the mathematical conversion of geological occurrence elements: normal vector n → dip, normal vector n → inclination, with an accuracy of ±1° (when the point spacing is <3 times the point cloud accuracy). The statistical connotation of flatness monitoring, the root mean square of distance, the Hausdorff distance metric from the point cloud to the fitted plane, reflects the degree of weathering of the structural surface; the root mean square of angle, the eigenvalue analysis of the normal vector covariance matrix.

[0105] Dynamic segmentation algorithm, segmentation threshold based on normal vector difference, eigenvalue correction to prevent over-segmentation; plane merging optimization to eliminate false segmentation surfaces caused by point cloud noise.

[0106] Preferably, this embodiment performs secondary sampling on the spliced ​​point cloud data, optimizes the data scale by adjusting the density between point clouds; segments the point cloud data, and segments the structural surface through the normal vector difference and eigenvalue correction parameters; selects three points to fit the structural surface plane, and calculates the normal force of the three-point fitting structural surface plane to obtain the inclination and inclination of the structural surface; dynamically segments different structural surfaces, and eliminates the segmentation phenomenon by merging planes; monitors the flatness of the fitting results, and generates a three-dimensional grid model of the tunnel based on the segmented structural surface data. Through secondary sampling and density optimization, the redundant information of point cloud data is reduced and the efficiency of subsequent processing is improved; by segmenting point cloud data through normal vector difference and eigenvalue correction parameters, the structural surface can be extracted more accurately, avoiding the errors that may be caused by traditional segmentation methods; dynamic segmentation and plane merging can effectively eliminate the segmentation phenomenon and ensure the continuity and consistency of the segmentation results; the three-dimensional mesh model generated based on the segmented structural surface data has higher accuracy and integrity, providing a reliable foundation for subsequent simulation and analysis; the overall process significantly improves modeling accuracy and simulation effects by optimizing point cloud data, accurately segmenting structural surfaces and generating high-quality three-dimensional mesh models, providing technical support for the generation of horizontal tunnel simulation models.

[0107] Furthermore, if Figure 7 As shown, the process of generating a three-dimensional mesh model of the tunnel based on the segmented structural surface data in step S233 specifically includes the following steps:

[0108] Step S2331: Evaluate the flatness of the structural surface by calculating the root mean square of the distance from the point cloud to the fitting plane and the root mean square of the normal vector angle; if the error exceeds the threshold, readjust the plane fitting or segmentation parameters;

[0109] Step S2332: Based on the segmented structural surface data, refine the segmentation by using normal vector difference and eigenvalue correction parameters to eliminate redundant point clouds; mesh the structural surface, connect the discrete points into triangular facets, and form a continuous surface;

[0110] Step S2333: Splice the local grids of each structural surface and process the spatial relationship between different structural surfaces; construct a three-dimensional grid model containing elements such as the terrain surface and the rock layer interface, assign material properties to different lithologic units, and build a tunnel simulation model based on the three-dimensional grid model.

[0111] Preferably, this embodiment evaluates the flatness of the structural surface by calculating the root mean square error (RMSE) of the distance from the point cloud to the fitting plane and the root mean square of the normal vector angle; if the error exceeds the threshold, the plane fitting or segmentation parameters are readjusted to optimize the fitting effect; based on the segmented structural surface data, the segmentation is refined by the normal vector difference and eigenvalue correction parameters to eliminate redundant point clouds; the structural surface is meshed, and discrete points are connected into triangular facets to form a continuous surface; the local meshes of each structural surface are spliced ​​to process the spatial relationship between different structural surfaces; a three-dimensional mesh model containing elements such as terrain surfaces and rock layer interfaces is constructed, and material properties of different lithologic units are assigned. By evaluating the root mean square error (RMSE) and the normal vector angle, the deviation of the plane fitting can be effectively detected, and the parameters can be dynamically adjusted to optimize the results; the refined segmentation and feature point optimization can eliminate redundant point clouds, reduce the computational burden, and generate a high-quality triangular mesh model. By splicing local meshes and assigning material properties, a three-dimensional mesh model containing elements such as terrain surfaces and rock layer interfaces is constructed. The tunnel simulation model is constructed based on the three-dimensional grid model, which can simulate the spatial layout and mechanical behavior of underground projects.

[0112] Furthermore, if Figure 8 As shown, the process of constructing the tunnel simulation model based on the three-dimensional grid model in step S2333 specifically includes the following steps:

[0113] Step S23331: performing preprocessing operations such as point cloud denoising, discrete point removal, and data redundancy reduction on the point cloud data in the three-dimensional mesh model to form a data set; dividing the data set into a training set, a validation set, and a test set; and constructing a tunnel simulation model based on the three-dimensional network model;

[0114] Step S23332: Process the 3D point cloud, use the pre-processed point cloud data as the tunnel simulation model, and output it as the predicted structure of the structural surface parameters; calculate the difference between the forward propagation result and the label through small batch data input, and update the model parameters;

[0115] Step S23333: input the point cloud data into the adit simulation model and simulate the point cloud data; analyze the simulation results to obtain the degree of structural surface opening, groundwater development on the structural surface, structural surface filling conditions, and structural surface filling shape classification.

[0116] Among them, the forward propagation and loss function update formula in step S23332 are:

[0117]

[0118] Where W and b represent the weight matrix and bias vector of the model respectively; and y i″ Denote the predicted value and true label of the i″th sample, respectively; η is the L2 regularization coefficient; B is the batch size; α is the learning rate; and t is the number of iterations. The mathematical principle of the loss function design and the construction basis of the least squares error term are based on the maximum likelihood estimation assuming that the error follows a Gaussian distribution. The Euclidean distance metric between the predicted value and the label is constructed in the output space and is sensitive to outliers. The L2 regularization term constrains the weight matrix by the Frobenius norm, which is equivalent to a Gaussian prior distribution of the weights, preventing overfitting.

[0119] The dynamics of gradient descent, the parameter update formula, and the theoretical basis of Lyapunov stability theory; learning rate α selection: adaptive methods (such as Adam) are better than fixed values, and the initial recommendation is α = 0.001 (point cloud data characteristics); mini-batch mechanism, and the trade-offs of batch size B

[0120] Variance reduction techniques for stochastic gradient descent.

[0121] Special processing of geological simulation modeling, structural surface parameter prediction, multi-task learning framework: shared underlying point cloud feature extraction, branch prediction of different parameters (such as filling shape using Softmax output); feature engineering: local normal vector histogram (for openness estimation), point density distribution (reflecting groundwater development);

[0122] Improvements to 3D convolution, using PointNet++ instead of traditional convolution and setting the hierarchical feature aggregation radius, have achieved groundbreaking applications in deep mines (such as gold mines below 3,000 meters), water conservancy project dam foundations (predicting structural surface shear strength), and shale gas reservoirs (modeling natural fracture networks).

[0123] Preferably, operations such as point cloud denoising, discrete point removal, and data redundancy reduction in this embodiment are important links in point cloud data processing; by inputting small batches of data, the difference between the forward propagation result and the label is calculated, and the model parameters are updated to optimize the model performance; the output of the horizontal tunnel simulation model is the structural surface parameter prediction structure, which can extract key features from the point cloud data and make predictions, which is of great significance in geological modeling and engineering analysis; through the analysis of the simulation results, information such as the degree of structural surface opening, groundwater development, filling conditions, and filling shape classification can be obtained, which has practical application value for geological research and engineering design. Through preprocessing operations such as point cloud denoising, discrete point removal, and data redundancy reduction, the quality of point cloud data was significantly improved and the impact of noise on model training was reduced; data set partitioning ensured the stability and generalization ability of model training and avoided the overfitting problem; the tunnel simulation model based on the three-dimensional network model can efficiently extract features from point cloud data and make predictions, improving the model's computational efficiency and prediction accuracy; small batch data input and parameter update mechanism enable the model to quickly adapt to new data distribution, improving the model's robustness and adaptability; simulation and result analysis provide important technical support for geological research.

[0124] Furthermore, if Figure 9 As shown, the process of obtaining the structural surface opening degree, structural surface groundwater development, structural surface filling condition, and structural surface filling shape classification in step S23333 specifically includes the following steps:

[0125] Step S233331: input the pre-processed point cloud data into the tunnel simulation model to generate a continuous triangulated network model and form a three-dimensional geometric expression of the structural surface; quantify the degree of opening by calculating the spatial distance or displacement between adjacent structural surfaces;

[0126] Step S233332: Identify the edges of the structural surface and areas of curvature change. High areas correspond to cracks with an opening greater than a threshold. If the point cloud contains reflectivity or color information, analyze the abnormal reflectivity values ​​in specific areas to infer the presence of groundwater. If there are depressions or holes in the grid model, it is determined to be a groundwater storage space.

[0127] Step S233333: If a sudden change in point cloud density or a terminal of texture continuity is identified, it is judged as a filling area, and the filling area is separated from the bedrock; the shape of the filling area is parameterized, and the filling is regionally classified; by analyzing the grid connectivity of the filling area, regular or irregular filling shapes are identified.

[0128] Among them, the quantitative formula of the structural surface opening degree in step S233331 is:

[0129]

[0130] Where, vl represents the vector of the lth line segment; projn(v l ) means that the vector v l Projection of the normal vector on the structural surface; u l represents the average vector of the neighboring cells of the lth line segment; γ represents the function smoothing coefficient, which adjusts the nonlinear effect of the opening degree; L represents the total number of total segments. The mathematical geological principle of the opening degree formula, the projection distance normalization term, is theoretically based on the surface deviation metric in differential geometry; its geological significance reflects the relative degree of separation between the fracture walls; the normalization process eliminates the influence of absolute size (accommodating millimeter- to meter-scale fractures); the directional consistency modulation term, cosine similarity, assesses the local consistency of fracture orientation, and the regulatory effect of the exponent γ suppresses directional jitter caused by noise and highlights the directional characteristics of the main controlling structural surface.

[0131] Preferably, this embodiment improves the quality of point cloud data through preprocessing (such as filtering, denoising, etc.) to ensure the accuracy of subsequent modeling; uses triangular mesh generation technology to convert discrete point cloud data into a continuous three-dimensional surface model to form a geometric expression of the structural surface; quantifies the degree of opening by calculating the spatial distance or displacement between adjacent structural surfaces, reflecting the ability to accurately analyze geometric features; based on curvature analysis and edge detection of point cloud data, identifies high areas corresponding to crack areas with an opening greater than a threshold; combines reflection degree or color information to infer the possibility of groundwater existence, reflecting the ability of multimodal data fusion; judges the filling area through sudden changes in point cloud density or interruptions in texture continuity, and separates the filling from the bedrock; performs shape parameterization and regional classification on the filling area, analyzes grid connectivity to identify regular or irregular filling shapes, reflecting the ability to model complex geometric shapes; judges the groundwater storage space through the identification of depressions or hole areas in the grid model, reflecting the ability to accurately extract underground space features. By generating a continuous triangulated network model, the three-dimensional geometric expression of the structural surface is achieved, providing an accurate basis for subsequent analysis; the degree of opening is quantified, providing an important basis for geological stability assessment; the presence of groundwater is inferred by combining the degree of reflection or color information, improving the accuracy of underground resource detection; the identification of depressions or holes in the grid model provides technical support for the judgment of groundwater storage space; the shape parameterization and regional classification of the filling area enhance the understanding and modeling capabilities of complex geometric forms; by analyzing the grid connectivity of the filling area, regular or irregular filling shapes are identified, providing a reference for engineering design and construction; and the preprocessing and optimization of point cloud data improves the accuracy and efficiency of modeling.

[0132] Furthermore, if Figure 10 As shown, the process of obtaining the layered rock mass value in step S3 specifically includes the following steps:

[0133] Step S31: Based on the three-dimensional point cloud data in the adit simulation model, the occurrence information of each structural surface is calculated through the normal vector, and the normal vector of the point cloud is converted into actual geographic coordinates; the structural surfaces are integrated through proximity search, and a two-dimensional projection surface is established to generate projection lines;

[0134] Step S32: Count the distances between the intersections of the orthogonal survey lines and the projection lines, and take the average as the distance between the structural surfaces; assign an independent coordinate system to the structural surface, calculate the extension length in the main direction, and quantify the extension degree of the structural surface; perform intelligent segmentation on the structural surface, and divide the structural surface into different groups by combining the normal vector difference and the characteristic final value;

[0135] Step S33: Count the occurrence distribution of each group and calculate the average inclination angle and spacing interval; compare with the actual measurement results and display. If the occurrence angle error is greater than the preset value, re-segmentation is performed.

[0136] Preferably, this embodiment is based on the three-dimensional point cloud data in the horizontal tunnel simulation model, calculates the attitude information of each structural surface through the normal vector, and converts the normal vector of the point cloud into actual geographic coordinates; integrates the structural surface through neighboring search, establishes a two-dimensional projection surface to generate projection lines; counts the distance between the intersection of the orthogonal survey line and the projection line, takes the average as the structural surface spacing, and assigns the structural surface an independent coordinate system to calculate the extension length in the main direction; counts the attitude distribution of each group, and calculates the average inclination and spacing interval and other values; realizes the automation of the whole process from data collection to analysis; if the attitude angle error is greater than the preset value, it is re-segmented; through three-dimensional point cloud data and normal vector calculation, the attitude information of the structural surface (such as strike, dip and inclination) is quickly extracted, solving the problem of low efficiency and large error of manual measurement in traditional methods. Combined with the normal vector difference and characteristic final value, the extension degree and spacing distribution of the structural surface are quantified, providing accurate data support for subsequent geological analysis and engineering design. Through the intelligent segmentation algorithm, the complex structural surface is divided into different groups, and the error correction is performed in combination with the actual measurement results, which improves the accuracy and reliability of the analysis. The entire process is highly automated, reducing manual intervention and significantly improving work efficiency. Error correction and optimization algorithms also ensure the accuracy of the final results. This system is suitable for structural surface analysis under complex geological conditions and can effectively handle complex field conditions, such as areas with steep slopes and complex terrain.

[0137] like Figure 11 As shown, this embodiment provides an embodiment of an electronic device. In this embodiment, the electronic device 1 includes a processor 11 and a memory 12 coupled to the processor 11.

[0138] The memory 12 stores program instructions for implementing the layout method of the layered rock mass numerical simulation method based on the hybrid progressive failure model according to any of the above embodiments.

[0139] The processor 11 is configured to execute program instructions stored in the memory 12 to perform a layout of a layered rock mass numerical simulation method based on a hybrid progressive failure model.

[0140] The processor 11 may also be referred to as a CPU (Central Processing Unit). The processor 11 may be an integrated circuit chip having signal processing capabilities. The processor 11 may also be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic device, a discrete gate or transistor logic device, or a discrete hardware component. The general-purpose processor may be a microprocessor or any conventional processor.

[0141] Furthermore, Figure 12 This is a schematic diagram of the structure of the storage medium of an embodiment of the present application. The storage medium 2 of the embodiment of the present application stores program instructions 21 that can implement all the above methods, wherein the program instructions 21 can be stored in the above storage medium in the form of a software product, including a number of instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) or a processor to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, or terminal devices such as a computer, server, mobile phone, and tablet.

[0142] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interface, device or unit, which can be electrical, mechanical or other forms.

[0143] In addition, the functional units in the various embodiments of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated units may be implemented in the form of hardware or in the form of software functional units. The above is only an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

[0144] The above detailed description of the specific embodiments of the invention is intended to be illustrative only, and the present invention is not limited to the specific embodiments described above. For those skilled in the art, any equivalent modifications or substitutions to the invention are also within the scope of the present invention. Therefore, equivalent changes, modifications, and improvements made without departing from the spirit and scope of the present invention are also encompassed within the scope of the present invention.

Claims

1. A numerical simulation method for layered rock mass based on a hybrid progressive failure model, characterized in that: The layered rock mass numerical simulation method based on the hybrid progressive failure model includes: Obtain the structural surfaces of the engineering rock mass in the study area and classify them into primary structural surfaces, tectonic structural surfaces, and superficial structural surfaces according to their scale and surface characteristics; classify the fault, dyke, and fissure geological structural surfaces of the surrounding rock of the underground caverns based on the structural surface characteristics; Through 3D laser scanning, the spatial coordinates of each sampling point on the tunnel surface are obtained to obtain the point cloud of the measured object, and the adjacent discrete points are connected; the spliced ​​point cloud data is resampled and plane fitting is performed to obtain the structural surface occurrence and construct the tunnel simulation model; The adit simulation model analyzes the dip of the structural surface and obtains the numerical value of the layered rock mass. The structural surfaces in the rock mass are divided into rigid structural surfaces and weak structural surfaces according to their properties, degree of weathering, density and filling conditions; and the data is output to the cloud platform.

2. The layered rock mass numerical simulation method based on the hybrid progressive failure model according to claim 1 is characterized in that: The process of classifying the fault, dyke and fracture geological structural surfaces of the surrounding rock of underground caverns based on structural surface characteristics includes the following steps: Obtain the extension length of the engineering rock mass structural surface and the width of the fracture zone in the study area, as well as the scale and engineering geological impact degree, and classify them into level I structural surface, level II structural surface, level III structural surface, level IV structural surface and level V structural surface; According to geological composition, they are divided into primary structural surface, tectonic structural surface and superficial structural surface; according to mechanical properties, they are divided into rigid structural surface and weak structural surface; Among them, primary structural surfaces include bedding; tectonic structural surfaces include faults and joints; superficial structural surfaces include weathering cracks; rigid structural surfaces include closed and unfilled; weak structural surfaces include weak interlayers or fillings; The parameters of the level I and level II structural surfaces are obtained by measurement to obtain the boundary conditions of the project stability; the level III and below structural surfaces are statistically processed to comprehensively analyze the development degree, bonding degree and control effect of the structural surfaces on the project stability.

3. The method for numerical simulation of layered rock mass based on a hybrid progressive failure model according to claim 1, characterized in that: The process of obtaining the occurrence of the structural surface includes the following steps: Through 3D laser scanning, the original 3D point cloud data is cut using a cutting tool to remove invalid point cloud data, and the 0-6 meter point cloud scanning data of the horizontal tunnel is obtained. Then, the 3-9 meter point cloud cutting data of the horizontal tunnel is obtained through iteration. Select multiple feature points from the overlapping parts of two point cloud data respectively, and automatically stitch them together based on the previous point cloud data. After the stitching is completed, compare it with the preset error threshold. If the comparison is greater than the preset error threshold, stitching is performed again. The spliced ​​point cloud data is resampled to adjust the density between point clouds; the density between the point clouds after hanging is plane fitted, and three points are selected to fit the structural surface; the normal vector of the selected plane is calculated to obtain the occurrence of the first structural surface; and a horizontal tunnel simulation model is constructed.

4. The method for numerical simulation of layered rock mass based on a hybrid progressive failure model according to claim 3, characterized in that: The process of removing invalid point cloud data from the original 3D point cloud data using the cutting tool includes the following steps: Obtain original point cloud data through 3D laser scanning, and use the cutting tool to directly cut out invalid areas; intercept valid point cloud data within the range of 0-6 meters, and eliminate redundant point cloud data outside the range; redundant point cloud data includes surrounding environment and noise points; A dynamic threshold is set to automatically identify and remove invalid points. By traversing the laser points, the distance between the current point and adjacent points is compared. If the distance exceeds the preset threshold, it is determined to be an invalid noise point and removed. The preset threshold includes discrete points caused by noise or abnormal reflections. After each scan, the range of the cutting tool is adjusted to three meters. By adjusting the coverage area of ​​the cutting tool multiple times, continuous point cloud data is finally spliced ​​together to form 3-9 meter point cloud cutting data of the horizontal tunnel.

5. The method for numerical simulation of layered rock mass based on a hybrid progressive failure model according to claim 3, characterized in that: The process of automatic splicing includes the following steps: Select multiple feature points in the overlapping area of ​​two adjacent point cloud data as matching benchmarks; wherein the multiple feature points include local point sets of geometric features of corner points and edge points; Using the coordinate system of the previous point cloud as a reference, calculate the cylinder transformation matrix from the current point cloud to the reference point; minimize the sum of the spatial values ​​of the corresponding feature points in the two point clouds; after the stitching is completed, calculate the matching error between the transformed point clouds; If the error exceeds the preset threshold, the feature point selection is readjusted; when the error exceeds the limit, re-alignment is performed by dynamically adjusting the number or position of feature points.

6. The method for numerical simulation of layered rock mass based on a hybrid progressive failure model according to claim 3, characterized in that: The process of obtaining the occurrence of the first structural surface includes the following steps: The spliced ​​point cloud data is subsampled and the data scale is optimized by adjusting the density between point clouds. The point cloud data is segmented and the structural surface is segmented by modifying the parameters using the normal vector difference and eigenvalue. Select three points to fit the structural surface plane, and calculate the normal force of the three-point fitting mechanism surface plane to obtain the inclination and inclination angle of the structural surface; dynamically segment different structural surfaces and eliminate segmentation phenomena by plane merging; The fitting results are monitored for flatness, and the root mean square of the distance between the point planes and the root mean square of the normal vector angle are calculated. A three-dimensional grid model of the tunnel is generated based on the segmented structural surface data, and the three-dimensional grid model is processed to obtain a tunnel simulation model.

7. The method for numerical simulation of layered rock mass based on a hybrid progressive failure model according to claim 6, characterized in that: The process of generating a 3D mesh model of the tunnel based on the segmented structural surface data includes the following steps: The flatness of the structural surface is evaluated by calculating the root mean square of the distance from the point cloud to the fitting plane and the root mean square of the normal vector angle; if the error exceeds the threshold, the plane fitting or segmentation parameters are readjusted; Based on the segmented structural surface data, the normal vector difference and eigenvalue correction parameters are used to refine the segmentation and eliminate redundant point clouds; the structural surface is meshed and the discrete points are connected into triangular patches to form a continuous surface; The local grids of each structural surface are spliced ​​together to process the spatial relationship between different structural surfaces; a three-dimensional grid model including terrain surface and rock layer interface elements is constructed, and material properties of different lithologic units are assigned. A horizontal tunnel simulation model is constructed based on the three-dimensional grid model.

8. The method for numerical simulation of layered rock mass based on a hybrid progressive failure model according to claim 7, characterized in that: The process of building a tunnel simulation model based on a 3D mesh model includes the following steps: The point cloud data in the 3D mesh model is subjected to point cloud denoising, discrete point removal, and data redundancy reduction preprocessing operations to form a data set; the data set is divided into a training set, a validation set, and a test set; and a tunnel simulation model is constructed based on the 3D network model; Process the 3D point cloud and use the pre-processed point cloud data as a tunnel simulation model, outputting the predicted structure of the structural surface parameters. By inputting small batches of data, the difference between the forward propagation results and the labels is calculated and the model parameters are updated. The point cloud data is input into the adit simulation model and simulated; the simulation results are analyzed to obtain the degree of structural surface opening, groundwater development on the structural surface, structural surface filling conditions, and structural surface filling shape classification.

9. The method for numerical simulation of layered rock mass based on a hybrid progressive failure model according to claim 8, characterized in that: The process of obtaining the degree of structural surface opening, the development of groundwater on the structural surface, the filling condition of the structural surface, and the classification of the filling shape of the structural surface includes the following steps: The pre-processed point cloud data is input into the tunnel simulation model to generate a continuous triangulated network model and form a three-dimensional geometric expression of the structural surface; the degree of opening is quantified by calculating the spatial distance or displacement between adjacent structural surfaces; Identify the edges of structural surfaces and areas of curvature change. High areas correspond to cracks with openings greater than a threshold. If the point cloud contains reflectivity or color information, analyze the abnormal reflectivity values ​​in specific areas to infer the presence of groundwater. If there are depressions or holes in the grid model, it is determined to be groundwater storage space. Identify the point cloud density mutation or texture continuity terminal, and judge it as the filling area, and separate the filling material from the bedrock in the filling area; parameterize the shape of the filling area and classify the filling material by region; and identify regular or irregular filling shapes by analyzing the grid connectivity of the filling area.

10. The method for numerical simulation of layered rock mass based on a hybrid progressive failure model according to claim 1, characterized in that: The process of deriving numerical values ​​for layered rock masses includes the following steps: Based on the 3D point cloud data in the adit simulation model, the occurrence information of each structural surface is calculated through the normal vector, and the normal vector of the point cloud is converted into actual geographic coordinates; the structural surfaces are integrated through proximity search, and a 2D projection surface is established to generate projection lines; The distances between the intersections of the orthogonal survey lines and the projection lines are calculated, and the average is taken as the spacing between the structural surfaces. An independent coordinate system is assigned to the structural surface, and the extension length in the main direction is calculated to quantify the extension degree of the structural surface. The structural surface is intelligently segmented and divided into different groups based on the normal vector difference and characteristic final value. The occurrence distribution of each group is counted, and the average dip angle and spacing interval values ​​are calculated; compared with the actual measurement results, if the occurrence angle error is greater than the preset value, the segmentation is re-performed.

Images