Automatic cutting and accurate model construction method of dangerous rock mass based on point cloud data

Abstract

The present invention relates to a method for automated cutting and precise model building of dangerous rock masses based on point cloud data, and belongs to the field of dangerous rock mass segmentation and three-dimensional model building. The present invention proposes an automated method capable of accurately segmenting the joint surfaces of dangerous rock masses and generating precise models by combining the point cloud data acquired by an unmanned aerial vehicle and the inclination angle information of the joint surfaces. The real joint surfaces in the precise model of the dangerous rock masses are curved meshes. The present invention aims to improve the modeling efficiency and accuracy of three-dimensional models of dangerous rock masses, and avoid the complexity and errors caused by manual segmentation; and the present invention can overcome the problem of inaccurate dangerous rock mass models caused by traditional planar structural surface cutting. The precise model can make the static analysis and dynamic research of dangerous rock masses more accurate, thereby providing protection for the risks to human life and property safety that dangerous rock masses may bring.

Description

Technical Field

[0001] The invention belongs to the field of dangerous rock mass segmentation and three-dimensional model construction, and specifically relates to a dangerous rock mass automatic cutting and precise model construction method based on point cloud data. Background Art

[0002] In the engineering simulation and analysis of dangerous rock mass, it is usually necessary to segment the rock mass mesh model to construct a three-dimensional model that conforms to the actual situation for analysis and simulation. In previous modeling methods, since non-planar joint surface meshes are difficult to construct, planar joint surface mesh models are usually used to directly obtain the joint plane through the dip angle for analysis.

[0003] However, since the joint surfaces of dangerous rocks in nature are usually not completely flat and the boundaries may have irregular fluctuations, simple plane cutting will lead to inaccurate block segmentation. This error may cause the joint surface to be enlarged and the block model of dangerous rock to be abnormal and wrong, thus affecting the authenticity of the model, causing the subsequent analysis and simulation results to be overly conservative or even wrong, and unable to accurately evaluate the stability of the rock mass.

[0004] Patent application CN202211371060.5 provides a method for identifying and monitoring dangerous rocks on slopes based on drone inspection, including the following steps: Step 1: Use drone inspection to collect slope image information to obtain a first point cloud model, and perform vegetation removal preprocessing on the first point cloud model; Step 2: Extract the normal vector direction of the main joints of the slope based on the first point cloud model; Step 3: Determine the location of dangerous rocks on the slope based on the normal vector direction of the main joints of the slope and using the stereographic projection analysis method and the normal vector matching method; Step 4: Repeat steps 1 to 3 for the same slope according to a preset time interval to obtain a second point cloud model, and perform point cloud model registration and relative distance calculation on the second point cloud model and the first point cloud model to obtain the result of slope change. Its purpose is to solve the problem of insufficient research on non-contact rapid identification of dangerous rocks on slopes and achieve the technical effect of subsequent monitoring of slope changes. Patent application CN202410068954.X provides a method for intelligent identification of structural surfaces based on three-dimensional point clouds, including the following steps: selecting an area where structural surfaces are developed, and obtaining three-dimensional point cloud data of the area through drone remote sensing; calculating the normal vector of each point; using the normal vector of each point as an input parameter, clustering using the DBSCAN algorithm to obtain the number of joint groups; for points in each joint group, using the point coordinates as input parameters, clustering using the DBSCAN algorithm again to obtain separate structural surfaces; calculating the normal vector and occurrence of each structural surface, and using area-weighted averaging to obtain the occurrence of each joint group. This method does not require filtering, denoising, etc. for the point cloud data obtained by the drone, and can automatically remove noise points during the clustering process. Only one algorithm is used in the process of structural surface identification, which is more convenient and concise. It is not necessary to know the number of joint groups before identification, and the obtained results are more objective, and can provide technical support for the subsequent establishment of calculation models and engineering stability analysis.

[0005] Both patent applications focus on the identification of the occurrence of joint surfaces, and what is obtained are plane joint surfaces. Although plane joint surfaces are simple and easy to use, when performing numerical simulations, using planes as joint surfaces cannot well reflect engineering reality.

[0006] At the same time, manual segmentation of fine mesh models is not only cumbersome and error-prone, difficult to adjust and modify, but the resulting mesh is also usually of poor quality. Therefore, there is an urgent need for an automated segmentation method to achieve accurate model construction and change the smoothness of the joints as needed.

[0007] Therefore, there is a need in the art for a method for automated cutting of dangerous rock masses and construction of precise models based on point cloud data. Summary of the invention

[0008] The present invention provides a method for automatically cutting and accurately modeling a dangerous rock mass based on point cloud data, the method comprising the following steps: step S1, converting point cloud data into a grid model: generating a three-dimensional grid model of the dangerous rock mass and inclination and dip angle information of a rough joint surface through point cloud processing of precise point cloud data of the dangerous rock mass acquired by an unmanned aerial vehicle; the grid model comprises outer contour information of the dangerous rock mass; step S2, constructing a reference joint plane P': determining the approximate direction of the reference joint plane P' according to the inclination and dip angle information of the rough joint surface in step S1; selecting any point on the dangerous rock boundary as the center, constructing an original plane P, performing two three-dimensional rotations through the inclination and dip angle, thereby constructing a reference joint plane P' parallel to the rough joint surface; step S3, obtaining the relationship between the reference joint plane P' and the grid model True intersection line: first obtain the intersection line of the reference joint plane P' and the grid model in step S1, and then eliminate the false intersection lines to obtain the true intersection line; step S4, form the initial joint polygonal surface S; step S5, obtain the intersection point screening spatial point set U'; step S6, obtain the intersection control points of the real joint surface S' and the grid model; step S7, generate the reference plane Ss of the real joint surface; step S8, grid cutting and screening the real joint surface S'; use the grid model of step S1 to cut the reference plane Ss of the real joint surface obtained in step S7, and the curved grid structure within the screened grid model is the real joint surface S'; step S9, construct an accurate model of the dangerous rock mass: one or more real joint surfaces S' obtained in step S8 and the grid model in step S1 together constitute an accurate model of the dangerous rock mass.

[0009] In the present invention, the two three-dimensional rotations in step S2 are one for aligning the inclination and the other for aligning the dip angle.

[0010] In a specific embodiment, step S4 includes the following steps: first, equidistant points are taken on the true intersection line in step S3, and then adjacent points of the equidistant points are connected with straight lines to obtain a simplified joint surface plane intersection line, and the range within the simplified joint surface plane intersection line is the initial joint polygon surface S.

[0011] In a specific implementation, the step S5 includes the following steps: S5.1, obtaining the midpoint O and normal vector normal of the initial joint polygon surface S, with O as the center and normal as the normal direction, generating a circle beyond the range of the grid model and equidistantly dividing the points on the circle {os|o1, o2, o3, ...}, the number of division points obtained by the equidistant division is n; S5.2, with os as the center and the distance d between adjacent circle centers as the maximum diameter, generating n groups of m concentric circles in each group; S5.3, taking points equidistantly in each concentric circle of the m×n concentric circles, and obtaining the search point set u0 of the screening space; S5.4, expansion of the search point set: offset the points on these concentric circles up and down along the normal vector normal to obtain the point set U, convert the data structure of the point set U, and obtain the intersection screening space point set U'.

[0012] In a specific implementation, in step S5, the number of segmentation points, that is, the number of groups of concentric circles n, ranges from 30 to 120, each group of concentric circles includes m concentric circles, and the value range of m is 5 to 20.

[0013] In a specific implementation, the step S6 includes the following steps: S6.1, taking the midpoint O of the initial joint polygonal surface S as the starting point, and drawing a line segment with each point on the n groups of point sets in the intersection point screening spatial point set U' as the end point; S6.2, finding the intersection of the line segment in step S6.1 and the dangerous rock body grid model in step S1 to obtain an intersection set; S6.3, selecting the optimal intersection from the intersection set of step S6.2 to form an optimal intersection point set; S6.4, using line segments to connect adjacent optimal intersection points in step S6.3 to form a three-dimensional non-planar curve l ; S6.5, on three-dimensional non-planar curves l The upper interpolation forms more interpolation points than the optimal number of intersection points; S6.6, searching the grid model described in step S1 for the points closest to each interpolation point in step S6.5, which are the intersection control points of the real joint surface S' and the grid model.

[0014] In a specific implementation, in step S6.5, the number of interpolation points is 2.5 to 5 times the number of optimal intersection points.

[0015] In a specific implementation, step S7 includes first constructing a NURBS surface master formula; then selecting parameters to generate an interpolation surface; and then adjusting the parameters to optimize the surface fitting to obtain a reference surface Ss of the real joint surface.

[0016] When searching for the approximate occurrence of the joint surface, the present invention also first uses a simple clustering method, that is, first finds the approximate plane occurrence structure, and then uses this as a reference to search for the occurrence of the real joint surface, so that it is more in line with the real geological structure and the weathering of dangerous rocks. The present invention aims to improve the modeling efficiency and accuracy of the three-dimensional model of dangerous rock bodies, avoiding the complexity and errors caused by manual segmentation; and the present invention can overcome the problem of inaccurate dangerous rock body models caused by traditional plane structural surface cutting. The precise model can make the static analysis and dynamic research of dangerous rock bodies more accurate, thereby providing protection for the risks to human life and property safety that dangerous rock bodies may bring. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is the precise point cloud data map of the dangerous rock mass obtained by the drone in step S1.

[0018] Figure 2 is the grid model diagram obtained in step S1.

[0019] Figure 3 Schematic diagram of the original plane P and the constructed reference joint plane P' in step S2.

[0020] Figure 4 This is the process of obtaining the intersection line between the reference joint plane P' and the grid model in step S3.

[0021] Figure 5 is the intersection line obtained in step S3.

[0022] Figure 6 is the value obtained in step S3. Figure 5 The true intersection lines remaining after removing the false intersection lines.

[0023] Figure 7 is the Figure 6 The result obtained by taking equidistant points on the true intersection line.

[0024] Figure 8 is the Figure 7 The simplified plane intersection line of the joint surface obtained by connecting the points in .

[0025] Fig. 9 is the initial joint polygonal surface S obtained in step S4.

[0026] Fig.10 This is a schematic diagram of generating m×n concentric circles in step S5.

[0027] Fig.11 This is a schematic diagram of equidistant point selection within each of the m×n concentric circles generated in step S5.

[0028] Fig.12It is the result of shifting the points on the circumference of the concentric circles up and down along the normal vector normal in step S5.

[0029] Fig.13 The schematic diagram of drawing a line segment in step S6 is as follows: the line segment starts from the midpoint O of the initial joint polygonal surface S and ends at each point on the n point sets.

[0030] Fig.14 is the value in step S6 Fig.13 Schematic diagram of the set of intersection points obtained by finding the intersection points of the line segments in and the dangerous rock mass grid, with the top view on the top and the front view on the bottom.

[0031] Fig.15 In step S6, Fig.14 Schematic diagram of the point set formed by selecting the optimal intersection point from the intersection points in , with the top view on the top and the front view on the bottom.

[0032] Fig.16 is the value in step S6 Fig.15 The three-dimensional non-planar curve formed by connecting the optimal intersection points in l Schematic diagram of a device, wherein the top view is on the top and the front view is on the bottom.

[0033] Fig.17 is the value obtained in step S6. Fig.15 The 60 points of the center view are Fig.16 Three-dimensional non-planar curves in l Schematic diagram of the 198 points obtained after interpolation.

[0034] Fig.18 is the value of Fig.17 The 198 points in the grid are closest to the 198 points on the grid.

[0035] Fig.19 The results of different real joint surface reference surfaces Ss obtained by adjusting different parameters Span and Flexibility in step S7, wherein the parameter Span refers to the span, and the parameter Flexibility refers to the flexibility.

[0036] Fig. 20 This is a schematic diagram of the reference plane Ss of the real joint surface S' in step S8 being cut by the mesh model in step S1.

[0037] Fig.21 is a real joint surface S' obtained in step S8.

[0038] Fig. 22 These are the results of the four real joint surfaces S' obtained in step S8.

[0039] Fig.23It is the result of comparing the four real joint surfaces S' obtained in step S8 with the initial joint polygonal surface S in step S4.

[0040] Fig.24 It is a mesh schematic diagram of the real joint surface S' in step S9.

[0041] Fig.25 It is a schematic diagram of the accurate model of the dangerous rock mass obtained in step S9.

[0042] Fig.26 This is a photograph of the Mushroom Rock from angle 1 in Example 2.

[0043] Fig. 27 This is a photo of the mushroom rock from angle 2 in Example 2.

[0044] Fig.28 This is a comparison diagram of the initial joint polygonal surface S (planar joint surface) and the real joint surface S' (curved joint surface) of the mushroom rock in Example 2, including a front view and a top view.

[0045] Fig.29 It is a schematic flow chart of the method for automatically cutting dangerous rock masses and constructing accurate models based on point cloud data of the present invention. DETAILED DESCRIPTION

[0046] The present invention is further described by the following examples and drawings, but the protection scope of the present invention is not limited to the following examples and drawings.

[0047] Example 1

[0048] 1. Step S1: Convert point cloud data into a mesh model.

[0049] First, the precise point cloud data of the dangerous rock mass is obtained by using drones, such as Figure 1 As shown in Figure 2, a three-dimensional grid model of the dangerous rock mass and the inclination angle information of the approximate joint surface are generated by point cloud processing. Figure 2 As shown in Figure 2. The mesh model provides the basis for subsequent segmentation and surface generation.

[0050] 2. Step S2: construct a reference joint plane P'.

[0051] According to the inclination and dip information of the joint surface, the approximate direction of the joint surface is determined. Any point on the dangerous rock boundary is selected as the center to construct the original plane P, and two three-dimensional rotations are performed through the inclination and dip angle to construct the reference joint plane P' parallel to the approximate joint surface. Figure 3 Schematic diagram of the original plane P and the constructed reference joint plane P' in step S2. Figure 3 The dip is 85.3 degrees and the inclination is 4.5 degrees. Figure 3The five dangerous rock masses in the figure involve four original planes P and four reference joint planes P'.

[0052] 3. Step S3: Obtain the intersection lines between the reference joint plane P' and the grid model, and eliminate the false intersection lines.

[0053] By intersecting the reference joint plane with the dangerous rock mass grid model, a series of intersection lines are obtained. These intersection lines can be divided into two categories: true plane intersection lines (true intersection lines) and false plane intersection lines (false intersection lines). True plane intersection lines correspond to plane joint surfaces obtained by cutting plane joint surfaces, while false plane intersection lines are non-joint surface intersection lines generated by plane cutting or small interference intersection lines generated by errors.

[0054] Intersection screening: All intersections are sorted by bubble sorting, and the longest intersection is selected as the true intersection by length. In some special cases, such as when the rock mass is severely embedded and the joint surface is much smaller than the complete rock surface at the location of the joint surface, it is necessary to manually remove such false plane intersections inside the rock mass to ensure that the intersection represents the actual joint surface.

[0055] Figure 4 This is the process of obtaining the intersection line in step S3. Figure 5 is the intersection line obtained in step S3. Figure 6 for Figure 5 The true intersection lines remaining after removing the false intersection lines. Figure 5 It can be seen that in the four-layer reference joint plane P', there are multiple intersection lines in the two lowest layers, and there are short-segment pseudo intersection lines in the two lowest layers. By screening and eliminating the pseudo intersection lines, we get Figure 6 Results shown.

[0056] 4. Step S4: Generate simplified joint surface plane intersection lines and then form an initial joint polygonal surface S.

[0057] right Figure 6 The true plane intersection line in is equidistantly selected, and the following is obtained: Figure 7 The results are shown. Figure 7 Connect the points in with straight lines to generate a set of simplified joint plane intersection lines, such as Figure 8 shown. Figure 8 The area within the simplified joint plane intersection line shown is formed as follows Fig. 9 The initial joint polygonal surface shown is used for further calculation of irregular joint surfaces.

[0058] 5. Step S5: Obtaining the intersection screening spatial point set U'

[0059] Based on the initial joint polygon surface S obtained in the previous step, a spatial point set U' of the intersection screening of the real joint surface S' (unknown) and the grid model (known) is generated. Specifically, by screening the initial joint polygon surface S and the points closest to the midpoint of the initial joint polygon surface S in the adjacent range as the partial intersection control points of the real joint surface S' and the grid model, the intersection lines obtained in this way have authenticity and universality. Fig. 9 The four initial joint polygonal surfaces S shown each include a midpoint and a normal vector (a vector perpendicular to the initial joint polygonal surface S).

[0060] The specific steps are as follows.

[0061] S5.1. Get the midpoint O and normal vector normal of the initial joint polygon S, take O as the center and normal as the normal direction, generate a circle beyond the model range and get the points {os|o1,o2,o3,…} on the circle at equal distances. The number of segmentation points in the example is n=60.

[0062] S5.2, with os as the center and the distance d between adjacent centers as the maximum diameter, generate n groups of m concentric circles in each group. In this example, m=10, for a total of 600 circles. Fig.10 shown.

[0063] S5.3. Take points equidistantly in each of the m×n concentric circles to obtain the search point set u0 of the screening space, such as Fig.11 As shown, points are taken at equal distances on each concentric circle, and the distances between points on different concentric circles with the same center are different. The number of points on each concentric circle is the same, so the points on the inner concentric circles are dense, and the points on the outer concentric circles are sparse.

[0064] S5.4. Expansion of the search point set: To eliminate the inadequate range of the search set caused by the error of the inclination angle, these points on the circumference are shifted up and down to a certain extent along the normal vector normal to expand the search range and obtain more accurate results, such as Fig.12 As shown. In the example, the three point sets {U|u1,u2,u3} are obtained by shifting up and down once. The data structure of the point set U is converted to obtain the intersection screening space point set U'. That is, the three point sets U are changed into a grouping method of n search ranges, namely {U'|u'1,u'2,u'3…u' n}.

[0065] 6. Step S6, obtaining the intersection control points of the real joint surface S' and the grid model.

[0066] Take the midpoint O of the initial joint polygon S as the starting point and each point on the n point sets (60 point sets) as the end point to draw a line segment, such as Fig.13 As shown, Fig.13 The intersection points of the line segments in the grid and the dangerous rock mass are obtained, and the intersection point set is as follows Fig.14 As shown, each group is sorted by the distance between the intersection point and O, and the intersection point with the closest distance is selected as the optimal intersection point of this group. Fig.15 Select these n intersection points and use them as interpolation points to generate a three-dimensional non-planar curve l ,See Fig.16 To improve the accuracy of the curve, the curve can be divided into finer point sets, see Fig.17 , in this example, Fig.15 and Fig.16 The 60 points in are interpolated to Fig.17 198 points in Fig.17 These points are further searched to obtain the nearest points on the dangerous rock mass grid, see Fig.18 . Fig.17 These points in are generally not on the grid, but close to it. Fig.18 All 198 points in are located on the grid. For this embodiment, after 60 points are interpolated to form 198 points, the accuracy of the intersection control points obtained is most appropriate. Fig.18 For Fig.17 The 198 points in the grid are closest to the 198 points on the grid, and these points on the grid are the intersection control points of the real joint surface S' and the grid model.

[0067] 7. Step S7: Generate a reference surface Ss of the real joint surface S'.

[0068] A surface mesh is generated using the point set of the aforementioned intersection control points. In the present invention, the essence of the real joint surface S' and the reference surface Ss are both surface meshes. The present invention adopts a non-uniform rational B-spline (NURBS) surface algorithm to generate a smooth surface Ss as the reference surface of the real joint surface S' by interpolating these points. The surface algorithm can well fit a surface model that meets actual needs. During the surface generation process, the parameters (span and flexibility) can be adjusted according to the actual engineering situation to ensure that the surface can not only fit the characteristics of the joint surface to obtain a high-quality surface joint surface, but also can autonomously adjust the smoothness. The present invention can autonomously adjust the parameters according to the actual engineering situation to obtain a high-quality real joint surface mesh that conforms to the laws of nature.

[0069] The parameter Span determines the resolution of the control point grid of the generated surface, that is, the number of control points. A higher Span value increases the number of surface control points, making the surface more complex and refined, and able to capture more details. Conversely, a lower Span value will generate a simpler, smoother surface, suitable for more regular shapes. Its function: Control the level of detail and complexity of the surface; the larger the value, the finer the surface and the closer it is to the input points; the smaller the value, the smoother and simpler the surface.

[0070] The parameter Flexibility determines the softness or smoothness of the surface. It adjusts the surface's responsiveness to input points. Higher flexibility makes the surface more compliant with input points, and the surface shape is more closely aligned with changes in input points. Lower flexibility produces a smoother surface, which tends to smoothly transition and ignore drastic changes between input points. Its function: Controls the surface's responsiveness to input points. The higher the flexibility, the more accurately the surface passes through the input points; the lower the flexibility, the smoother the surface is and does not completely follow the changes in the points.

[0071] Fig.19 These are the results of the reference plane Ss of different real joint surfaces S' obtained by adjusting different parameters Span and Flexibility in the present invention.

[0072] 8. Step S8: Mesh cutting and screening of real joint surfaces S'.

[0073] Use the dangerous rock mass mesh, that is, the mesh model of step S1, to cut the reference plane Ss of the real joint surface S' to obtain {Ss'|s1,s2...}, and randomly select a non-boundary point on each mesh and judge whether the point is inside the closed dangerous rock mass mesh, and select the joint surface mesh inside as the real joint surface surface mesh, that is, the real joint surface S'.

[0074] Fig. 20 It is a schematic diagram of the reference plane Ss of the real joint surface S' in step S8 being cut by the grid model in step S1, wherein the black dots in the inner white area and the black dots in the outer grey area represent a non-boundary point randomly selected on the grid of the reference plane Ss. Fig.21 is a real joint surface S' obtained in step S8. Fig. 22 These are the results of the four real joint surfaces S' obtained in step S8. Fig.23 The four real joint surfaces S' obtained in step S8 are compared with the initial joint polygonal surface S in step S4. The light grey in the figure is the initial joint polygonal surface S, and the dark grey is the real joint surface S'. Comparing the two, it can be seen that the area of ​​the joint surface is significantly reduced.

[0075] 9. Step S9: construct an accurate model of the dangerous rock mass.

[0076] According to the generated precise curved joint surface mesh, i.e., the mesh of the real joint surface S', the rock mass is segmented to obtain a mesh model. That is, the mesh of the real joint surface S' obtained in step S8 (including four real joint surfaces S' in this embodiment) and the mesh model obtained in step S1 are used as the outer contour to together constitute the precise model of the dangerous rock mass. The cutting described in the present invention is to use four real joint surfaces S' to cut the precise model of the dangerous rock mass into five parts. The precise model of the dangerous rock mass contains the real joint surface S', which can serve as the basis for the structural stability analysis and simulation of the dangerous rock mass. Fig.24 It is a mesh schematic diagram of the real joint surface S' in step S9. Fig.25 It is a schematic diagram of the accurate model of the dangerous rock mass obtained in step S9.

[0077] In general, the modeling method of the present invention has a high degree of automation, reduces human intervention, and significantly improves the accuracy and efficiency of dangerous rock mass model construction.

[0078] The present invention realizes accurate segmentation and model construction of dangerous rock mass through automated data processing, geometric screening, point set generation and surface construction. This method not only improves efficiency, but also greatly reduces errors in manual operation, and is suitable for large-scale dangerous rock mass segmentation and stability analysis.

[0079] The segmentation method and modeling method of the present invention can parameterize the search range of the joint surface and the complete grid intersection line and the irregularity of the joint surface, that is, the automated modeling method that is more in line with the actual situation can be parameterized. In step S5, the upper and lower ranges of the search can be parameterized to adapt to various irregular situations and obtain a true joint surface intersection line. In step S7, joint surfaces of different smoothness can be achieved by adjusting the number of control points and formula parameters.

[0080] The joint surface obtained by the method of the present invention will be more in line with the actual situation, and a more realistic simulation result will be obtained during numerical simulation. In the prior art, the block obtained by plane cutting will obtain conservative results after simulation due to the enlargement of the joint surface, that is, a relatively safe result, which is not conducive to the discovery of danger. The method of the present invention can avoid this situation, and the parameterized adjustment can obtain a dangerous rock model that conforms to the actual situation.

[0081] Example 2

[0082] This example is a real case of mushroom rock. Its modeling method and steps are consistent with those in Example 1. The real photo of the mushroom rock can be seen in Fig.26 and Fig. 27 ,in Fig.26 This is a photo of the Mushroom Rock from angle 1 in Example 2. Fig. 27 This is a photo of the mushroom rock from angle 2 in Example 2.

[0083] Fig.28 The figure is a comparison of the front view and top view of the planar joint surface (the initial joint polygonal surface S in step S4) and the curved joint surface (the real joint surface S' obtained in step S8) of the mushroom rock. The upper left corner is a schematic diagram of the front view of the initial joint polygonal surface S of the mushroom rock, the lower left corner is a schematic diagram of the front view of the real joint surface S' of the mushroom rock, and the picture on the right is a comparison of the top view of one of the initial joint polygonal surfaces S and the real joint surface S'.

[0084] From the analysis results of the finite difference after the computer cyclically calculates the plane joint surface and the curved joint surface 10,000 times, it can be seen that the result obtained by the plane joint surface first has inaccurate shape itself, and secondly, the calculation result will also change due to the difference in the shape of the block. Therefore, the use of the plane joint surface cannot simulate the accurate result well, and the method of the present invention is more reasonable for the disposal of dangerous rock mass. For the subsequent purpose of analyzing the motion trajectory of dangerous rocks by discrete element analysis, the accurate joint surface shape provided in the present invention also helps to realize the calculation simulation of accurate motion trajectory and stress conditions, and determine whether it will endanger the personal and property safety of people.

[0085] The formula for generating the curved joint surface (irregular real joint surface S') in step S7 of the present invention includes the following formula.

[0086] NURBS surface main formula:

[0087] Formula (1)

[0088] In formula (1), the summation formula of the entire numerator is obtained by applying the B-spline basis function to all control points. P i,j The influence of is accumulated to generate points on the surface S(u,v) , the denominator is a normalization calculation to make the sum of the weights equal to 1; S The surface is defined in parameter space by the parameters u and v Determined point; parameter u and v It is a two-dimensional parameter domain that defines the surface and is used to control the shape of the surface; N is the B-spline basis function, which is used to define the local characteristics of the surface; p and q is the order of the B-spline basis function, respectively defined u and v Surface smoothness in direction; P i,j are the coordinates of the control points, indicating the position of each control point in the control grid; is a double summation, indicating that u andv The contributions of the control points in each direction are accumulated; i is u The control point index in the direction, j is v Control point index in the direction; m+span The control points are defined at u and v The area of ​​influence in two directions; w i,j +flexibility is the weight of the control point, which defines the influence of each control point on the surface shape.

[0089] B-spline basis functions:

[0090] Formula (2)

[0091] In formula (2), N i,p-1 (u) It is a low-order basis function, which is used to recursively calculate the high-order basis function; is a parameter u Normalized value in the current node area, used to adjust the local influence of the B-spline; u i and u i+p These are the node values ​​of the node vectors, defining the segment space on the B-spline surface.

[0092] Weighted basis functions:

[0093] Formula (3)

[0094] Formula (3) represents the joint weighted basis function on a two-dimensional surface, which is used to control the local characteristics of the surface shape.

[0095] The present invention uses the non-uniform rational B-spline surface interpolation algorithm, namely the NURBS algorithm, to generate an accurate surface through the input known points. NURBS surfaces are extremely flexible and can adapt to complex geometric shapes while accurately describing smooth surfaces. By adjusting the span and flexibility in the NURBS parameters, we can control the smoothness and accuracy of the generated surface to better fit the real undulations of the joint surface.

[0096] Specifically, step S7 includes the following steps.

[0097] Step S7-1: First construct the main formula of the NURBS surface as shown in formula (1).

[0098] Specifically, we first construct the NURBS basis function as shown in equation (2): for each direction (usually u and v directions), we construct the B-spline basis function as shown in equation (2); then we construct the weighted basis function as shown in equation (3); and finally we establish the main surface formula as shown in equation (1). The span parameter controls the influence range of the point in the interpolation process, and flexibility affects the smoothness of the surface by adjusting the weight.

[0099] Step S7-2: Select parameters and generate an interpolation surface.

[0100] Specifically, according to the above NURBS surface equation, that is, the main formula of the NURBS surface as shown in formula (1), a fitting surface containing all known points is generated so that the surface passes through or approaches the position at these points, so that the generated surface can accurately reflect the morphology and characteristics of the actual joint surface.

[0101] Step S7-3: Adjust parameters to optimize surface fitting.

[0102] Specifically, by adjusting the span and flexibility parameters, the local smoothness and overall trend of the generated surface are controlled to avoid uneven surfaces caused by interpolation errors or local discontinuities, thereby ensuring that the generated surface has higher accuracy and meets actual needs.

[0103] In general, the present invention relates to a method for automated cutting and precise model construction of dangerous rock masses based on point cloud data. The present invention proposes an automated method capable of accurately segmenting the joint surfaces of dangerous rock masses and generating precise models by combining the point cloud data acquired by an unmanned aerial vehicle and the inclination angle information of the joint surfaces. The real joint surfaces in the precise model of the dangerous rock masses are curved meshes. The present invention aims to improve the modeling efficiency and accuracy of three-dimensional models of dangerous rock masses and avoid the complexity and errors caused by manual segmentation; and the present invention can overcome the problem of inaccurate dangerous rock mass models caused by traditional planar structural surface cutting. The precise model can make the static analysis and dynamic research of dangerous rock masses more accurate, thereby providing protection for the risks to human life and property safety that dangerous rock masses may bring.

[0104] The above contents are further detailed descriptions of the present invention in combination with specific preferred embodiments, and it cannot be determined that the specific implementation of the present invention is limited to these descriptions. For ordinary technicians in the technical field to which the present invention belongs, several simple deductions and substitutions can be made without departing from the concept of the present invention, which should be regarded as belonging to the protection scope of the present invention.

Claims

1. A method for automatic cutting and accurate modeling of dangerous rock mass based on point cloud data, characterized in that: The method comprises the following steps, Step S1, converting point cloud data into a grid model: the precise point cloud data of the dangerous rock mass obtained by the drone is processed by point cloud processing to generate a three-dimensional grid model of the dangerous rock mass and the inclination angle information of the approximate joint surface; the grid model includes the outer contour information of the dangerous rock mass; Step S2, constructing a reference joint plane P': according to the inclination and dip angle information of the approximate joint surface in step S1, determining the approximate direction of the reference joint plane P'; selecting any point on the dangerous rock boundary as the center, constructing the original plane P, performing two three-dimensional rotations through the inclination and dip angle, thereby constructing a reference joint plane P' parallel to the approximate joint surface; Step S3, obtaining the true intersection line between the reference joint plane P' and the grid model: first obtain the intersection line between the reference joint plane P' and the grid model in step S1, and then eliminate the false intersection lines to obtain the true intersection line; Step S4, forming an initial joint polygonal surface S; Step S5, obtaining the intersection screening spatial point set U'; Step S6, obtaining the intersection control points of the real joint surface S' and the grid model; Step S7, generating a reference surface Ss of the real joint surface; Step S8, mesh cutting and screening of real joint surfaces S'; using the mesh model of step S1 to cut the reference surface Ss of the real joint surface obtained in step S7, the curved mesh structure within the mesh model is screened as the real joint surface S'; Step S9, constructing an accurate model of the dangerous rock mass: the one or more real joint surfaces S' obtained in step S8 and the grid model in step S1 together constitute an accurate model of the dangerous rock mass.

2. The method for automated cutting and accurate modeling of dangerous rock mass based on point cloud data according to claim 1 is characterized in that: The step S4 comprises the following steps: firstly taking equidistant points on the true intersection line in step S3, and then connecting the adjacent points of the equidistant points with straight lines to obtain a simplified joint surface plane intersection line, and the range within the simplified joint surface plane intersection line is the initial joint polygon surface S.

3. The method for automated cutting and accurate modeling of dangerous rock mass based on point cloud data according to claim 2 is characterized in that: The step S5 comprises the following steps: S5.

1. Obtain the midpoint O and normal vector normal of the initial joint polygonal surface S, take O as the center and normal as the normal direction, generate a circle beyond the range of the mesh model, and divide the circle into points {os|o1,o2,o3,…} at equal intervals. The number of division points obtained by equal interval division is n. S5.2, with os as the center and the distance d between adjacent centers as the maximum diameter, generate n groups of m concentric circles in each group; S5.3, take points equidistantly in each of the m×n concentric circles to obtain the search point set u0 of the screening space; S5.

4. Expansion of the search point set: offset the points on these concentric circles up and down along the normal vector normal to obtain the point set U, transform the data structure of the point set U, and obtain the intersection screening space point set U'.

4. The method for automated cutting and accurate modeling of dangerous rock mass based on point cloud data according to claim 3 is characterized in that: In step S5, the number of segmentation points, that is, the number of groups of concentric circles n, ranges from 30 to 120, each group of concentric circles includes m concentric circles, and the value range of m is 5 to 20.

5. The method for automated cutting and accurate modeling of dangerous rock mass based on point cloud data according to claim 3 is characterized in that: The step S6 comprises the following steps: S6.

1. Take the midpoint O of the initial joint polygonal surface S as the starting point and draw a line segment with each point on the n points in the intersection point selection spatial point set U' as the end point; S6.2, finding the intersection points of the line segment in step S6.1 and the dangerous rock mass grid model in step S1 to obtain an intersection point set; S6.3, selecting the optimal intersection point from the intersection point set of step S6.2 to form the optimal intersection point set; S6.4, connect the adjacent optimal intersection points in step S6.3 with line segments to form a three-dimensional non-planar curve l ; S6.

5. On three-dimensional non-planar curves l Upper interpolation forms more interpolation points than the optimal number of intersection points; S6.6, searching the grid model described in step S1 for the point closest to each interpolation point in step S6.5 in a one-to-one correspondence, namely, the intersection control point of the real joint surface S' and the grid model.

6. The method for automated cutting and accurate modeling of dangerous rock mass based on point cloud data according to claim 5 is characterized in that: In step S6.5, the number of interpolation points is 2.5 to 5 times the number of optimal intersection points.

7. The method for automated cutting and accurate modeling of dangerous rock mass based on point cloud data according to claim 5, characterized in that: In step S7, the main formula of the NURBS surface is first constructed; then parameters are selected to generate an interpolation surface; and then the parameters are adjusted to optimize the surface fitting to obtain the reference surface Ss of the real joint surface.

Images