Information extraction method for underground engineering tunnel face rock mass structural surface

Abstract

The invention relates to the technical field of underground engineering tunnel face information extraction, and discloses an information extraction method for an underground engineering tunnel face rock mass structural plane, and the method comprises the steps: extracting the normal vector of each point from the original point cloud data of the underground engineering tunnel face rock mass structural plane, inputting the converted normal vector into a BIRCH algorithm for clustering to obtain a dominant structural plane category; taking the types of the dominant structural surfaces as input, executing a DBSCAN algorithm on the point cloud of each type of dominant structural surfaces, and segmenting the dominant structural surfaces into independent structural surfaces; performing plane fitting on the independent structural plane by using an RANSAC algorithm to obtain a plane fitting equation, and calculating an inclination angle and an inclination direction of the structural plane according to the plane fitting equation; according to the method, point cloud data provides a rich information source for rock mass structure analysis, an independent structural plane is accurately recognized, and tunnel face rock mass structural plane information is reasonably extracted.

Description

Technical Field

[0001] The invention relates to the technical field of underground engineering tunnel face information extraction, and in particular to an information extraction method for a rock mass structural surface of an underground engineering tunnel face. Background Art

[0002] At present, the New Austrian Tunneling Method (NATM) is commonly used in mountain underground projects for design and construction. Its core idea is to make full use of the bearing capacity of the surrounding rock, use anchor spraying support and other means to constrain the deformation of the surrounding rock, and dynamically adjust the design and construction through monitoring and measurement. However, the geological conditions of actual projects are often complex and changeable, and the untimely and imprecise collection of geological information brings great difficulties to dynamic design and safe and efficient construction. Therefore, in the actual construction process, the three aspects of geological information collection, dynamic design and construction of underground projects are easily disconnected. It is urgent to use intelligent means to "empower" these three aspects, and connect "collection-design-construction" into a seamless whole to realize the intelligent transformation and upgrading of underground project construction. Therefore, the determination of the surrounding rock level of the face is the prerequisite for the intelligent design and construction of underground projects, and in-depth research is needed in this regard to help the intelligent construction of underground projects.

[0003] Rock mass is one of the important environments for civil engineering represented by underground engineering. Since rock masses with different construction processes have undergone multiple periods of geological tectonic movements and the transformation of shallow geological actions in the long geological history, there are widely discontinuous surfaces with different occurrences, scales and properties in the rock mass, which are called structural surfaces. The existence of structural surfaces gives the rock mass significant structural characteristics, the main characteristics of which are complexity, discontinuity, anisotropy, and nonlinear characteristics of mechanical response. These structural characteristics constitute the weak parts in the rock mass and become one of the important factors controlling the stability of the rock mass.

[0004] In order to achieve a standardized description of structural surface characteristics, an index system for quantitatively describing rock mass structural surfaces has been proposed, including occurrence, spacing, continuity, trace length, number of groups, roughness, opening, filling, gap wall strength and groundwater, of which seven are related to the geometric characteristics of the structural surface. The traditional method of collecting rock mass structural surface information requires geologists to carry out manual contact measurement with handheld measuring instruments (such as geological compass, tape measure, feeler gauge, etc.). This method has the disadvantages of being time-consuming, incomplete measurement, measurement accuracy affected by manual experience, and the engineering site environment (such as rock collapse and rock burst during underground engineering face measurement, high and steep slope measurement) posing a danger to the safety of surveyors. In the process of rock mass underground engineering construction, backward artificial geological sketching technology often cannot meet the actual needs of rapid engineering construction, resulting in the structural surface information of the face rock mass not being fully extracted, analyzed and utilized, and unable to form a timely and effective feedback effect. Therefore, it is crucial to quickly, reliably and objectively obtain the structural surface information of the face rock mass of underground engineering. Summary of the invention

[0005] In view of this, the present invention provides a method for extracting information on the structural surface of the rock mass at the face of an underground engineering project, so as to solve the problem that the artificial geological sketching technology in the prior art is difficult to meet the actual needs of rapid engineering construction, resulting in the structural surface information of the rock mass at the face not being fully extracted, analyzed and utilized.

[0006] In a first aspect, the present invention provides a method for extracting information from a rock mass structure surface of an underground engineering tunnel face, the method comprising:

[0007] Extract the normal vector of each point from the original point cloud data of the rock mass structural surface of the underground engineering face, input the converted normal vector into the BIRCH algorithm for clustering, and obtain the dominant structural surface category;

[0008] Taking the dominant structural surface category as input, executing the DBSCAN algorithm on the point cloud of each category of dominant structural surface, and segmenting the dominant structural surface into independent structural surfaces;

[0009] Using the RANSAC algorithm to perform plane fitting on the independent structural surface, obtain a plane fitting equation, and calculate the inclination and inclination of the structural surface according to the plane fitting equation;

[0010] The PCA weighted iterative algorithm is used to shrink the rock mass trace.

[0011] The present invention processes the point cloud data of the rock structure surface of the underground engineering face. The point cloud data, with its high density and high precision, provides a rich information source for rock structure analysis, so as to classify the dominant structural surfaces, use the DBSCAN algorithm to achieve more refined independent structural surface segmentation, so as to accurately identify independent structural surfaces in complex geological environments, use the RANSAC algorithm to perform robust plane fitting on the structural surfaces to reduce the interference of abnormal points, and use the PCA weighted iterative algorithm to shrink the rock traces, so as to reasonably extract the rock structure surface information of the face.

[0012] In an optional implementation, the converted normal vector is input into the BIRCH algorithm for clustering to obtain the dominant structural surface category, including:

[0013] The converted normal vector is input into the BIRCH algorithm to construct the CF Tree and cluster the normal vectors of the structural surface;

[0014] Based on the clustering results, identify the dominant structural surface categories that are significantly representative.

[0015] The present invention utilizes the BIRCH algorithm to efficiently and automatically divide and identify dominant structural surface categories, so as to identify dominant structural surface categories with significant representativeness, and provide data support for subsequent geomechanical analysis and stability evaluation.

[0016] In an optional implementation, executing the DBSCAN algorithm on the point cloud of each type of dominant structural surface to segment the dominant structural surface into independent structural surfaces includes:

[0017] The neighborhood radius and the minimum number of points are set, and the DBSCAN algorithm is executed on the point cloud of each type of dominant structural surface. The dominant structural surface is divided into independent structural surfaces according to the local density of the point cloud.

[0018] The present invention reasonably sets the neighborhood radius and the minimum number of points. The reasonably set neighborhood radius takes into account the actual spacing between structural surfaces and the resolution of the point cloud. The reasonably set minimum number of points ensures that the effective outline of the structural surface can be captured without incorrectly connecting adjacent different structural surfaces, so as to facilitate the segmentation of the dominant structural surface into independent structural surfaces.

[0019] In an optional implementation, the plane fitting of the independent structural surface using the RANSAC algorithm to obtain the plane fitting equation includes:

[0020] From the point cloud constituting the independent structural surface, three points are randomly selected to calculate the plane fitting equation, which is as follows:

[0021] AX+BY+CZ+D=0.

[0022] The present invention utilizes the RANSAC algorithm to perform robust plane fitting on the structural surface to reduce the interference of abnormal points.

[0023] In an optional implementation, the calculating the inclination and the dip of the structural surface according to the plane fitting equation includes:

[0024] Replace [A,B,C] in the plane fitting equation T The normal vector to the XOY plane is [0,0,1] T The included angle is determined as the inclination of the structural surface;

[0025] The angle between the positive X-axis and the inclination line clockwise is determined as the inclination of the structural surface.

[0026] The present invention calculates the dip angle and inclination of the structural surface respectively according to the plane fitting equation, so as to realize the calculation of the occurrence of the structural surface.

[0027] In an optional implementation, the rock mass trace shrinkage using the PCA weighted iterative algorithm includes:

[0028] Use PCA to calculate the point cloud vector, and use the boundary point calculation formula to calculate the boundary point set corresponding to the point cloud vector;

[0029] Calculating the shrinkage weight of each point in the boundary point set, and performing shrinkage processing on the boundary points;

[0030] The displacement direction of the contracted boundary is corrected to obtain the rock mass trace.

[0031] The present invention utilizes the PCA weighted iterative algorithm to shrink the rock mass trace, taking into account the eigenvalue, eigenvector and the shrinkage weight of each point, and correcting the displacement direction so that the shrunken points are evenly distributed along the length direction of the trace, so that the trace identification result is close to the occurrence grouping.

[0032] In a second aspect, the present invention provides a device for extracting information from a rock mass structure surface of an underground engineering tunnel face, the device comprising:

[0033] The clustering module is used to extract the normal vector of each point from the original point cloud data of the rock mass structural surface of the underground engineering face, and input the converted normal vector into the BIRCH algorithm for clustering to obtain the dominant structural surface category;

[0034] An independent structural surface segmentation module is used to take the dominant structural surface category as input, execute the DBSCAN algorithm on the point cloud of each category of dominant structural surface, and segment the dominant structural surface into independent structural surfaces;

[0035] A calculation module, used to perform plane fitting on the independent structural surface using the RANSAC algorithm to obtain a plane fitting equation, and calculate the inclination and inclination of the structural surface according to the plane fitting equation;

[0036] The trace shrinkage module is used to shrink the rock trace using the PCA weighted iterative algorithm.

[0037] In a third aspect, the present invention provides a computer device, comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the information extraction method for the rock structure surface of an underground engineering face according to the above-mentioned first aspect or any corresponding embodiment thereof by executing the computer instructions.

[0038] In a fourth aspect, the present invention provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to enable a computer to execute the method for extracting information from a rock structure surface of an underground engineering face according to the first aspect or any corresponding embodiment thereof.

[0039] In a fifth aspect, the present invention provides a computer program product, comprising computer instructions for causing a computer to execute the method for extracting information from a rock mass structure surface at a tunnel face of an underground engineering project according to the first aspect or any corresponding embodiment thereof. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0041] Figure 1 It is a schematic flow chart of a method for extracting information of a rock mass structural surface of an underground engineering face according to an embodiment of the present invention;

[0042] Figure 2 is a schematic diagram of occurrence grouping and independent structural surface identification results according to an embodiment of the present invention;

[0043] Figure 3 is a schematic diagram of eigenvalues ​​and eigenvectors according to an embodiment of the present invention;

[0044] Figure 4 is a schematic diagram of trace directional shrinkage according to an embodiment of the present invention;

[0045] Figure 5 is a schematic diagram of trace recognition and shrinkage according to an embodiment of the present invention;

[0046] Figure 6 is a structural block diagram of a device for extracting information of a rock mass structure surface of an underground engineering face according to an embodiment of the present invention;

[0047] Figure 7 It is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0049] According to an embodiment of the present invention, an embodiment of a method for extracting information from a rock mass structure surface at an underground engineering face is provided. It should be noted that the steps shown in the flowchart of 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 can be executed in an order different from that shown here.

[0050] In this embodiment, a method for extracting information of rock mass structure surface of underground engineering tunnel face is provided. Figure 1 is a flow chart of a method for extracting information of rock mass structure surface of an underground engineering face according to an embodiment of the present invention. Figure 1 As shown, the process includes the following steps:

[0051] Step S101, extracting the normal vector of each point from the original point cloud data of the rock mass structural surface of the underground engineering face, inputting the converted normal vector into the BIRCH algorithm for clustering, and obtaining the dominant structural surface category.

[0052] In an embodiment of the present invention, point cloud data, as an important achievement of modern geological surveying and mapping, provides a rich source of information for rock structure analysis with its high density and high precision. The original point cloud data of the rock structure surface of the underground engineering face is obtained, and the normal vector of each point is extracted from the original point cloud data and used as a feature vector. The feature vector is converted into a representation suitable for clustering, and the converted normal vector is input into the BIRCH algorithm for clustering to obtain the dominant structural surface category.

[0053] Step S102, taking the dominant structural surface category as input, executing the DBSCAN algorithm on the point cloud of each category of dominant structural surface, and segmenting the dominant structural surface into independent structural surfaces.

[0054] In the embodiment of the present invention, after the preliminary classification of the dominant structural surfaces is completed, in order to more finely distinguish and define each structural unit, the DBSCAN (Density-Based Spatial Clustering of Applications with Noise) density clustering algorithm is used to achieve the segmentation of independent structural surfaces.

[0055] Specifically, DBSCAN is a density-based clustering method that is particularly suitable for noisy data sets and can effectively identify clusters of arbitrary shapes, which is particularly important for the fine segmentation of complex rock structure surfaces.

[0056] The dominant structural surface categories obtained by clustering the BIRCH algorithm are used as input, and the DBSCAN algorithm is performed on the point cloud of each type of dominant structural surface. The dominant structural surface is further subdivided into independent structural units according to the local density of the point cloud.

[0057] Step S103, using the RANSAC algorithm to perform plane fitting on the independent structural surface to obtain a plane fitting equation, and according to the plane fitting equation, calculate the inclination and inclination of the structural surface.

[0058] In the embodiment of the present invention, after obtaining the independent structural surface, the occurrence of the structural surface, that is, the inclination and dip of the structural surface is calculated. Since the result of the structural surface generation is a collection of triangular facets belonging to the same structural surface. Considering that the surface of the structural surface is usually not an idealized plane and there are abnormal points, the RANSAC algorithm is used to perform robust plane fitting on the structural surface, reduce the interference of abnormal points, and calculate the inclination and dip of the structural surface.

[0059] Step S104, using the PCA weighted iterative algorithm to shrink the rock mass trace.

[0060] In the embodiment of the present invention, a PCA weighted iterative algorithm is used to perform directional shrinkage on the rock mass trace to obtain the rock mass trace, so as to realize the extraction of information on the rock mass structure surface of the underground engineering face.

[0061] The present embodiment provides a method for extracting information on rock mass structural surfaces at underground engineering faces. By processing point cloud data of rock mass structural surfaces at underground engineering faces, the point cloud data, with its high density and high precision, provides a rich source of information for rock mass structure analysis, so as to classify dominant structural surfaces. The DBSCAN algorithm is used to achieve more refined independent structural surface segmentation, so as to accurately identify independent structural surfaces in complex geological environments. The RANSAC algorithm is used to perform robust plane fitting on structural surfaces to reduce interference from abnormal points. The PCA weighted iterative algorithm is used to shrink rock mass traces, so as to reasonably extract information on rock mass structural surfaces at the face.

[0062] In this embodiment, a method for extracting information of rock mass structure surface of underground engineering tunnel face is provided, and the process includes the following steps:

[0063] Step S201, extracting the normal vector of each point from the original point cloud data of the rock mass structural surface of the underground engineering face, inputting the converted normal vector into the BIRCH algorithm for clustering, and obtaining the dominant structural surface category.

[0064] Specifically, the above step S201 includes:

[0065] Step S2011, input the converted normal vector into the BIRCH algorithm, construct a CF Tree, and cluster the normal vectors of the structural surface.

[0066] Step S2012, identifying dominant structural surface categories with significant representativeness based on the clustering results.

[0067] In the embodiment of the present invention, the rock mass dominant structural surface grouping includes the following two parts:

[0068] (1) Normal vector calculation

[0069] In rock engineering, the occurrence is usually expressed by inclination and dip. The traditional manual contact measurement method can directly read the occurrence information from the geological compass. For the 3D point cloud model obtained by laser scanning or stereo photogrammetry, it is necessary to first calculate the normal vector of the point cloud. The least squares method and the principal component analysis (PCA) method are two commonly used methods for normal vector calculation. Considering the robustness to noise, this section uses the PCA algorithm to calculate the normal vector. Suppose there is a point p0 in the three-dimensional point cloud, and the normal vector of p0 needs to calculate the covariance matrix:

[0070]

[0071] Among them, p0 is the initial point, k nn is the number of neighboring points of p0, p i is the i-th k of p0 nn Neighboring points, T is the transposed symbol, is the eigenvector, λ i is the eigenvalue corresponding to the i-th eigenvector.

[0072] Considering the small k nn (k nn <15) will cause a lot of noise in the generated normal vector, and a large k nn (k nn >30) will make the local curvature too smooth, so k is chosen nn =20 is the default value. Let λ1>λ2>λ3 be the covariance matrix The characteristic value of is the corresponding eigenvector, then the normal vector corresponding to point p0 is the eigenvector corresponding to the minimum eigenvalue

[0073] (2) Structural surface classification based on improved BIRCH clustering algorithm based on cosine similarity

[0074] In the field of geological engineering and underground engineering construction, accurate identification and classification of rock mass structural surfaces are the key to evaluating the stability of surrounding rocks and optimizing the design and construction plan. The dominant structural surface classification method based on normal vector calculation in point cloud model and BIRCH (Balanced Iterative Reducing and Clustering using Hierarchies) clustering algorithm aims to efficiently and automatically divide and identify the main structural units in the rock mass, providing support for subsequent geomechanical analysis and stability evaluation.

[0075] The BIRCH algorithm is an efficient online clustering algorithm suitable for large-scale data sets. It achieves data compression and clustering by constructing a hierarchical clustering feature tree (CF Tree). The terminology is explained as follows:

[0076] CF (Cluster Feature): represents the summary information of a cluster, including centroid, radius, and number of points.

[0077] CF Tree: Contains leaf nodes and non-leaf nodes. Leaf nodes store CF, and non-leaf nodes summarize the CF information of their child nodes.

[0078] Incremental learning: The algorithm continuously updates the CF Tree while traversing the data and controls memory usage by merging similar CFs.

[0079] Hierarchical clustering: CF Tree naturally forms a hierarchical clustering structure of data and supports dynamic insertion and deletion of data points.

[0080] First, the normal vector of each point is extracted from the original point cloud and used as the feature vector. The three-dimensional normal vector is converted into a representation suitable for clustering, such as a normalized vector or an encoding based on the normal vector angle. This is only an example and not a limitation. The converted normal vector is input into the BIRCH algorithm, and a CF Tree is constructed by setting appropriate parameters (such as the maximum height of the tree, the branching factor, etc.) to achieve efficient clustering of the structural surface normal vectors. According to the clustering results, a small number of but significantly representative structural surface categories are identified. These categories correspond to the dominant structural surfaces in the rock mass and have a decisive influence on the stability of the surrounding rock and the safety of the project. The clustering quality is evaluated by the silhouette coefficient, Calinski-Harabasz index, etc., and the BIRCH parameters are adjusted to optimize the clustering effect.

[0081] In the traditional BIRCH algorithm, Euclidean distance is usually used as the similarity metric. However, in high-dimensional space, Euclidean distance may not be the best choice because the increase in dimension causes the distance metric to become less effective. In contrast, cosine similarity can better reflect the angular difference between two vectors, which is more stable in high-dimensional space.

[0082] Define new clustering features: Since cosine similarity involves vector direction rather than distance, we need to redefine the clustering features in the CF tree to support cosine similarity calculation.

[0083] Cosine similarity calculation: When inserting new data points, use cosine similarity instead of Euclidean distance to determine the closest cluster.

[0084] Adjust the splitting strategy: When splitting a node, considering the characteristics of cosine similarity, you may need to adjust the splitting criteria or threshold to ensure that the child nodes after the split still maintain a high similarity.

[0085] Optimize the CF tree: In order to improve efficiency, it may be necessary to modify the structure of the CF tree, such as adjusting the height, width, or node capacity limit of the CF tree.

[0086] Step S202 , taking the dominant structural surface category as input, executing the DBSCAN algorithm on the point cloud of each category of dominant structural surface, and segmenting the dominant structural surface into independent structural surfaces.

[0087] Specifically, the above step S202 includes:

[0088] Step S2021, setting the neighborhood radius and the minimum number of points, executing the DBSCAN algorithm on the point cloud of each type of dominant structural surface, and segmenting the dominant structural surface into independent structural surfaces according to the local density of the point cloud.

[0089] In the embodiment of the present invention, the DBSCAN (Density-Based Spatial Clustering of Applications with Noise) density clustering algorithm is used to achieve the segmentation of independent structural surfaces.

[0090] The DBSCAN algorithm defines clustering through two core parameters: ε (epsilon, neighborhood radius) and MinPts (minimum number of points). Its basic idea is as follows:

[0091] Density direct access: If the distance from point P to point Q does not exceed ε (distance threshold), then point Q is called a point in the ε-neighborhood of point P. If there are at least MinPts points (including point Q itself) in the ε-neighborhood of point Q, then point Q is called a density direct access point of point P.

[0092] Density reachability: Density reachability between points is constructed by transferring the density direct relationship. If there is a point sequence P1, P2, ..., Pn, where every two points are density directly reachable, then point P1 to point Pn is called density reachable.

[0093] Density connected: If two points are density reachable from each other, they belong to the same density connected set.

[0094] Cluster generation: All density-connected point sets are considered as a cluster. Points that do not belong to any density-connected set are marked as noise points.

[0095] The dominant structural surface categories obtained by BIRCH clustering are used as input, and the point cloud of each type of dominant structural surface is processed separately, using the calculated normal vector and spatial position information. According to the characteristics of the rock mass structure, ε and MinPts are set reasonably. The ε value needs to take into account the actual spacing between the structural surfaces and the resolution of the point cloud, while `MinPts` should ensure that the effective outline of the structural surface can be captured without incorrectly connecting adjacent different structural surfaces. The DBSCAN algorithm is executed for each type of dominant structural surface, and the dominant structural surface is further subdivided into independent structural units according to the local density of the point cloud. The results of DBSCAN clustering will naturally reveal the boundaries of the structural surface, that is, those areas marked as noise points, which are usually transition areas between different structural surfaces. To evaluate the segmentation results, it may be necessary to adjust the parameters of DBSCAN according to the actual situation to optimize the segmentation effect. The rationality of the structural surface segmentation can be checked through visualization tools and verified with the knowledge of geological experts.

[0096] The DBSCAN density clustering algorithm provides an effective way to independently segment the dominant structural planes and can accurately identify independent structural units in complex geological environments. By flexibly adjusting parameters and combining prior knowledge, this method can significantly enhance the understanding of rock mass structure and provide more accurate basic data for subsequent stability analysis, support design, and safety assessment. In the future, we can explore integrated learning strategies that combine the advantages of multiple clustering algorithms to cope with more diverse geological challenges.

[0097] Step S203, using the RANSAC algorithm to perform plane fitting on the independent structural surface, obtain a plane fitting equation, and calculate the inclination and inclination of the structural surface according to the plane fitting equation.

[0098] Specifically, the above step S203 includes:

[0099] Step S2031, randomly select three points from the point cloud constituting the independent structural surface to calculate the plane fitting equation.

[0100] Step S2032: replace [A, B, C] in the plane fitting equation T The normal vector to the XOY plane is [0,0,1] T The included angle is determined as the inclination of the structural surface.

[0101] Step S2033, determining the angle between the positive axis of the X-axis and the inclination line in the clockwise direction as the inclination of the structural surface.

[0102] In the embodiment of the present invention, the RANSAC algorithm is used to perform robust plane fitting on the structural surface to reduce the interference of abnormal points. The specific process is as follows:

[0103] Step (1): For a point cloud constituting a structural surface, three points are randomly selected to calculate its plane fitting equation: AX+BY+CZ+D=0.

[0104] Step (2): Calculate the vertical distance from each point in the point cloud of each structural surface to the fitting plane of the structural surface

[0105] The formula is as follows:

[0106]

[0107] Among them, d i It is the vertical distance from the point in the structural surface point cloud to the structural surface fitting plane.

[0108] Step (3): Set the distance threshold t rsc , then d t >t rsc The point is regarded as an outlier point, and d t <t rsc The points are regarded as interior points, and the number of interior points N is counted in ;

[0109] Iterate the above steps (1) to (3) N times rsc times, and then select the fitting plane with the largest number of internal points;

[0110] The plane with the most internal points is then fitted to obtain the final plane fitting equation.

[0111] In the plane fitting equation [A,B,C] T The normal vector to the XOY plane is [0,0,1] T The angle between the two is the dip angle (DA) of the structural surface, with a value range of 0 to 90, and is defined as:

[0112]

[0113] Let vec t ' is the normal vector vec t Projection on the XOY plane, θ is the positive direction of the X axis and the projection vec t ', then:

[0114]

[0115] The dip direction (DD) of the structural surface is the direction between the positive axis of the X-axis and the dip line vec t ', the value range is 0~360. θ∈[0,90], the conversion relationship between θ and DD is shown in Table 1.

[0116] Table 1

[0117] -D / A -D / B -D / C relation + + + DD=θ - + + DD=180-0 - - + DD=180+0 + - + DD=360-0 + + - DD=180+0 - + - DD=360-0 - - - DD=θ + - - DD=180-0

[0118] This embodiment provides an example. The underground project is excavated by drilling and blasting. The blasting disturbance and construction disturbance joints increase the difficulty of identifying independent structural surfaces. After the blasting and slag removal, in the construction gap before the installation of the steel arch, a camera is used to take 8 images from left to right in front of the face. The total shooting time is about 30 seconds, and the image resolution is 4032×3024. Then the above-mentioned normal vector calculation method is used to reconstruct the three-dimensional point cloud model from the image sequence. The original point cloud includes 475112 points, of which the face ROI area includes 324827 points. This example is used as the excavation face data of a rock underground project including blasting disturbance and excavation disturbance joints to verify the adaptability and robustness of this method to the identification of dominant structural surfaces and independent structural surfaces in different complex scenarios.

[0119] The results of occurrence grouping and independent structural surface identification are as follows: Figure 2 The comparison of occurrence identification results is shown in Table 3.

[0120] Table 3

[0121] Joint number Real production This method error J1 45° / 68° 46° / 69° Δ1° / Δ1° J2 18° / 35° 17° / 39° Δ1° / Δ4°

[0122] Step S204: shrinking the rock mass trace using the PCA weighted iterative algorithm.

[0123] Specifically, the above step S204 includes:

[0124] Step S2041, using PCA to calculate the point cloud vector, and using the boundary point calculation formula to calculate the boundary point set corresponding to the point cloud vector.

[0125] Step S2042, calculating the shrinkage weight of each point in the boundary point set, and performing shrinkage processing on the boundary points.

[0126] Step S2043, performing displacement direction correction on the shrunk boundary to obtain the rock mass trace.

[0127] In the embodiment of the present invention, after obtaining the three-dimensional point cloud, the point cloud vector is first calculated using the above PCA method. By solving the covariance matrix in formula (1), its eigenvalue is obtained as λ1>λ2>λ3, and the corresponding eigenvector is like Figure 3 As shown, represents the longest extension direction of point p0 and its neighboring points, and therefore represents the direction vector of point p0, It represents the minimum extension direction of point p0 and its neighboring points, and therefore represents the normal vector of point p0.

[0128] Basic sharp points are defined as edge points and corner points with obvious geometric curvature in the 3D point cloud model. Since the geometric shapes of the sharp point neighborhood vary greatly, the sharp points are identified based on the angle change of the neighborhood normal vector. Suppose the point cloud is P = {p1, p2, ..., p n}, the corresponding normal vector is Vec = {vec1,vec2,...,vec n}, then the neighborhood angle change value of the i-th normal vector,δ ang,i Defined as:

[0129]

[0130] In the formula, the function Ang is used to calculate the acute angle between two vectors, vec i Represents point p i The normal vector, vec j Represents point p j K nn The normal vector of the jth neighbor point among the neighbor points. Then, the sharp point is defined as the neighborhood angle change value δ ang,i Greater than the neighborhood angle change value δ of all points in the point cloud ang The average point, δ ang Substitute into formula (5) to calculate the sharp point set Shp:

[0131]

[0132] After obtaining the boundary points of the occurrence grouping, the PCA weighted iterative algorithm is used for directional contraction. Considering that in the PCA analysis, the eigenvector corresponding to the maximum eigenvalue λ1 (i.e., direction vector) represents the longest extension direction of the neighborhood point set, and the neighborhood points are shrunk using the proportion of λ1 in the eigenvalue as the weight. Before shrinking, Update to the adaptive direction vector in the previous section to improve the directional accuracy of directional contraction. The specific steps are as follows:

[0133] (1) Assume that the boundary point coordinate set Bp is generated by formula (5), and re-search the k points within Bp for Bp. nn Nearest points, and calculate the direction vector of each point;

[0134] (2) For each point bp i ∈Bp, the eigenvalue calculated by formula (1) is λ1>λ2>λ3, then its shrinkage weight wb i Defined as

[0135]

[0136] (3) Set point bp i K nnThe index of the neighbor point in Bp is Neig_bp i ={nb i1 ,nb i2 ,...,nb iknn}, then point bp i Point after contraction bp i ' is defined as:

[0137]

[0138] (4) Next, in order to make the shrunken points evenly distributed along the length of the trace, it is necessary to adjust bp i ' is corrected in the displacement direction so that bp i Finally, the vertical movement relative to the center axis of the trace occurs instead of parallel movement, so the parallel displacement relative to the center axis of the trace needs to be offset. Figure 4 shown.

[0139] The algorithm is also verified based on the underground project mentioned in the above example.

[0140] like Figure 5 As shown, Figure 5 The various steps of trace extraction are performed for this example. The optimal number of occurrence groupings is 3. The shrunk point cloud skeleton can accurately reflect the occurrence grouping boundary. In addition, the consistency between the trace segment and the occurrence grouping results is quantified by the average minimum distance between the trace segment point cloud and the occurrence grouping boundary point cloud. The average minimum distance of this method is 0.0138, which proves that the traces identified by this method are closer to the occurrence grouping boundary line, and the trace recognition results are more consistent with the occurrence grouping.

[0141] This example shows that this method can be effectively applied to the trace identification of the excavation surface of rock underground engineering, and its resistance to overlapping interference and consistency with the occurrence grouping are better than traditional methods.

[0142] The information extraction method for the rock structure surface of the underground engineering face provided in this embodiment uses the PCA weighted iterative algorithm to shrink the rock trace, taking into account the eigenvalues, eigenvectors and the shrinkage weights of each point, and performing displacement direction correction so that the shrunk points are evenly distributed along the length direction of the trace, so that the trace recognition result is close to the occurrence grouping.

[0143] In this embodiment, a device for extracting information from a rock mass structure surface of an underground engineering face is also provided. The device is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made are omitted. As used below, the term "module" can be a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0144] This embodiment provides a device for extracting information from rock mass structure surfaces of underground engineering tunnel faces. Figure 6 As shown, including:

[0145] The clustering module 601 is used to extract the normal vector of each point from the original point cloud data of the rock mass structural surface of the underground engineering face, input the converted normal vector into the BIRCH algorithm for clustering, and obtain the dominant structural surface category.

[0146] The independent structural surface segmentation module 602 is used to take the dominant structural surface category as input, execute the DBSCAN algorithm on the point cloud of each category of dominant structural surface, and segment the dominant structural surface into independent structural surfaces.

[0147] The calculation module 603 is used to perform plane fitting on the independent structural surface using the RANSAC algorithm to obtain a plane fitting equation, and calculate the inclination and tilt of the structural surface according to the plane fitting equation.

[0148] The trace shrinkage module 604 is used to shrink the rock mass trace using the PCA weighted iterative algorithm.

[0149] In some optional implementations, the clustering module 601 includes:

[0150] The clustering unit is used to input the converted normal vector into the BIRCH algorithm, construct the CF Tree, and cluster the normal vectors of the structural surface.

[0151] The category recognition unit is used to identify the dominant structural surface categories with significant representativeness according to the clustering results.

[0152] In some optional implementations, the independent structural surface segmentation module 602 includes:

[0153] The segmentation unit is used to set the neighborhood radius and the minimum number of points, execute the DBSCAN algorithm on the point cloud of each type of dominant structural surface, and segment the dominant structural surface into independent structural surfaces according to the local density of the point cloud.

[0154] In some optional implementations, the calculation module 603 includes:

[0155] The fitting unit is used to randomly select three points from the point cloud that constitutes the independent structural surface to calculate the plane fitting equation. The plane fitting equation is as follows:

[0156] AX+BY+CZ+D=0.

[0157] In some optional implementations, the calculation module 603 further includes:

[0158] The first calculation unit is used to convert [A, B, C] in the plane fitting equation T The normal vector to the XOY plane is [0,0,1] T The included angle is determined as the inclination of the structural surface.

[0159] The second calculation unit is used to determine the angle between the positive axis of the X-axis and the inclination line in the clockwise direction as the inclination of the structural surface.

[0160] In some optional implementations, the trace shrinking module 604 includes:

[0161] The third calculation unit is used to calculate the point cloud vector by using PCA, and calculate the boundary point set corresponding to the point cloud vector by using the boundary point calculation formula.

[0162] The shrinkage unit is used to calculate the shrinkage weight of each point in the boundary point set and perform shrinkage processing on the boundary points.

[0163] The correction unit is used to correct the displacement direction of the contracted boundary to obtain the rock mass trace.

[0164] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0165] The information extraction device for the rock structure surface of the underground engineering face in this embodiment is presented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0166] The embodiment of the present invention also provides a computer device having the above Figure 6 The device shown is used for extracting information of rock structure surface at underground engineering face.

[0167] See also Figure 7 , Figure 7 is a schematic diagram of the structure of a computer device provided by an optional embodiment of the present invention, such as Figure 7As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components are connected to each other using different buses for communication, and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer device, including instructions stored in or on the memory to display the graphical information of the GUI on an external input / output device (such as, a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 7 A processor 10 is taken as an example.

[0168] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.

[0169] The memory 20 stores instructions executable by at least one processor 10, so that at least one processor 10 executes the method shown in the above embodiment.

[0170] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0171] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.

[0172] The computer device also includes an input device 30 and an output device 40. The processor 10, the memory 20, the input device 30 and the output device 40 may be connected via a bus or other means. Figure 7 The example of connecting through bus is taken in the following.

[0173] The input device 30 can receive input digital or character information and generate key signal input related to user settings and function control of the computer device, such as a touch screen, etc. The output device 40 can include a display device, etc.

[0174] The embodiment of the present invention also provides a computer-readable storage medium. The method according to the embodiment of the present invention can be implemented in hardware, firmware, or can be implemented as a computer code that can be recorded in a storage medium, or can be implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium through a network download, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor, or hardware, the method shown in the above embodiment is implemented.

[0175] A part of the present invention may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present invention through the operation of the computer. Those skilled in the art should understand that the existence of the computer program instruction in a computer-readable medium includes, but is not limited to, a source file, an executable file, an installation package file, etc., and accordingly, the way in which the computer program instruction is executed by the computer includes, but is not limited to: the computer directly executes the instruction, or the computer compiles the instruction and then executes the corresponding compiled program, or the computer reads and executes the instruction, or the computer reads and installs the instruction and then executes the corresponding installed program. Here, the computer-readable medium may be any available computer-readable storage medium or communication medium accessible to the computer.

[0176] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations all fall within the scope of the present application.

Claims

1. A method for extracting information from rock mass structural surfaces at underground engineering faces, characterized in that: The method comprises: Extract the normal vector of each point from the original point cloud data of the rock mass structural surface of the underground engineering face, input the converted normal vector into the BIRCH algorithm for clustering, and obtain the dominant structural surface category; Taking the dominant structural surface category as input, executing the DBSCAN algorithm on the point cloud of each category of dominant structural surface, and segmenting the dominant structural surface into independent structural surfaces; Using the RANSAC algorithm to perform plane fitting on the independent structural surface, obtain a plane fitting equation, and calculate the inclination and inclination of the structural surface according to the plane fitting equation; The PCA weighted iterative algorithm is used to shrink the rock mass trace.

2. The method according to claim 1, characterized in that The converted normal vector is input into the BIRCH algorithm for clustering to obtain the dominant structural surface categories, including: The converted normal vector is input into the BIRCH algorithm to construct the CF Tree and cluster the normal vectors of the structural surface; Based on the clustering results, identify the dominant structural surface categories that are significantly representative.

3. The method according to claim 1, characterized in that The DBSCAN algorithm is executed on the point cloud of each type of dominant structural surface to segment the dominant structural surface into independent structural surfaces, including: The neighborhood radius and the minimum number of points are set, and the DBSCAN algorithm is executed on the point cloud of each type of dominant structural surface. The dominant structural surface is divided into independent structural surfaces according to the local density of the point cloud.

4. The method according to claim 1, characterized in that: The RANSAC algorithm is used to perform plane fitting on the independent structural surface to obtain a plane fitting equation, including: From the point cloud constituting the independent structural surface, three points are randomly selected to calculate the plane fitting equation, which is as follows: AX+BY+CZ+D=0.

5. The method according to claim 4, characterized in that The step of calculating the inclination and dip of the structural surface according to the plane fitting equation comprises: Replace [A,B,C] in the plane fitting equation T The normal vector to the XOY plane is [0,0,1] T The included angle is determined as the inclination of the structural surface; The angle between the positive X-axis and the inclination line clockwise is determined as the inclination of the structural surface.

6. The method according to claim 1, characterized in that The method of using the PCA weighted iterative algorithm to shrink the rock mass trace includes: Use PCA to calculate the point cloud vector, and use the boundary point calculation formula to calculate the boundary point set corresponding to the point cloud vector; Calculating the shrinkage weight of each point in the boundary point set, and performing shrinkage processing on the boundary points; The displacement direction of the contracted boundary is corrected to obtain the rock mass trace.

7. An information extraction device for rock mass structure surface of underground engineering face, characterized in that: The device comprises: The clustering module is used to extract the normal vector of each point from the original point cloud data of the rock mass structural surface of the underground engineering face, and input the converted normal vector into the BIRCH algorithm for clustering to obtain the dominant structural surface category; An independent structural surface segmentation module is used to take the dominant structural surface category as input, execute the DBSCAN algorithm on the point cloud of each category of dominant structural surface, and segment the dominant structural surface into independent structural surfaces; A calculation module, used to perform plane fitting on the independent structural surface using the RANSAC algorithm to obtain a plane fitting equation, and calculate the inclination and inclination of the structural surface according to the plane fitting equation; The trace shrinkage module is used to shrink the rock trace using the PCA weighted iterative algorithm.

8. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the information extraction method for the rock structure surface of an underground engineering face according to any one of claims 1 to 6 by executing the computer instructions.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the method for extracting information from a rock mass structure surface of an underground engineering face according to any one of claims 1 to 6.

10. A computer program product, characterized in that The method comprises computer instructions for causing a computer to execute the method for extracting information of a rock mass structure surface of an underground engineering face according to any one of claims 1 to 6.

Images