Method for updating new drawing of underground tunneling face in real time based on set logic driving

Abstract

The invention discloses an underground tunneling face real-time new drawing updating method based on set logic driving. The method comprises the following steps: firstly, installing a laser radar at an unshielded position of a tunneling roadway, performing initialization scanning on a tunneling surface, preprocessing an obtained initial point cloud to obtain a reference point cloud, and performing Poisson reconstruction after normal constraint to obtain a reference model; then, threshold filtering and ray sampling filtering are carried out on the scanned point cloud in the tunneling process, the initial point cloud and the reference model, and a tunneling change area is identified; and then, through the set logic of inward intersection solving and outward union solving, the change area and the initial point cloud are fused, a changed tunneling face model is obtained, and the purpose of quickly updating a new graph is achieved. Hardware equipment used in the method is low in cost, the recognition efficiency of the feature change of the heading face is high, real-time automatic mapping of the heading face can be achieved, and basic requirements of underground heading work are met.

Description

1. Technical Field

[0001] The invention belongs to the field of mine surveying, and in particular relates to a real-time updating three-dimensional mapping method for an underground excavation working face based on set logic drive. 2. Background Technology

[0002] The underground environment of coal mines is complex and changeable, with narrow space, limited visual range, insufficient light and other characteristics, which brings great challenges to the real-time monitoring and management of underground excavation working faces. Traditional underground monitoring and modeling technologies rely on manual operation, which is not only inefficient, but also difficult to meet the needs of high precision and real-time updates. Therefore, if 3D modeling of underground excavation working faces can be achieved, it will greatly improve the intelligence and digitalization level of mines, and also improve the efficiency and safety of underground production work.

[0003] In recent years, with the rapid development of computer vision, sensor technology and data processing technology, the real-time updating of 3D mapping technology for underground tunneling working faces has become a research hotspot. In existing studies, certain results have been achieved by installing high-precision visual sensors underground, using machine vision technology to monitor the working conditions of the tunneling face, or using millimeter-wave radar to achieve continuous scanning of the tunneling working face. However, the above methods still have many limitations in actual use, including high equipment cost, complex daily maintenance, easy to be affected by dust, and poor quality of 3D modeling.

[0004] The rapid development of laser radar technology can accurately measure the geometric structure of the underground environment and generate high-precision point cloud data. Based on this, the present invention designs a fast three-dimensional mapping and updating method for underground excavation working faces based on set logic drive, which can construct a three-dimensional model of the excavation face in real time, quickly and accurately, thereby guiding production work, with high economic benefits and social value. III. Summary of the invention

[0005] 1. Technical issues to be resolved

[0006] The technical problem to be solved by the present invention is that, in the current underground excavation process, due to the narrow and long restricted space of the tunnel, high dust, limited visual range and other factors, it is impossible to obtain the three-dimensional dynamic map of the excavation working face in a timely and accurate manner. In view of this problem, a method for quickly updating and building a map of the underground excavation working face based on set logic is designed. First, a laser radar device is fixed in an unobstructed place to perform an initial scan of the excavation face to obtain a reference point cloud. Then, the reference point cloud of the excavation face is preprocessed to remove noise, and a Poisson reconstruction algorithm after normal constraint is used to form a reference three-dimensional model of the excavation face. During the cutting process of the tunneling machine, a fixed laser radar is used to continuously scan the excavation face, and the scanned point cloud is compared with the reference point cloud and its reference model through threshold filtering and ray sampling filtering, and the set logic of seeking intersection inward and seeking union outward is used to identify the changed area of ​​the excavation face. Finally, the point cloud of the changed area is fused with the reference point cloud to generate a changed excavation face model to achieve the goal of quickly updating the map.

[0007] (II) Technical solution

[0008] In view of the fact that the underground excavation working face is limited by light, space and high dust concentration, it is difficult to obtain the three-dimensional map of the excavation face in real time. The present invention proposes a method for quickly updating the map. Before, during and after excavation, the strategies of seeking intersection inward and seeking union outward are respectively adopted to model the scanned point cloud data in real time. Through the three-dimensional models constructed at different excavation stages, the excavation work is effectively assisted, the problems of limited light, limited space and high dust concentration are solved, and the efficiency of underground excavation operations is improved. The main contents are as follows:

[0009] 1) Obtaining the benchmark point cloud and benchmark model: Fix the position of the laser radar to ensure that the position does not change during scanning, so as to obtain point cloud data in the same coordinate system. Scan before excavation to obtain the initial point cloud, perform point cloud denoising and voxel filtering preprocessing on the initial point cloud as the benchmark point cloud, and calculate the normal of the benchmark point cloud. Due to the complex structure of the excavation surface, the normal of the point cloud is not uniform, which affects the construction of the excavation surface model. Therefore, normal constraints are performed so that the normal vectors of the point cloud all point to the inside of the point cloud, and Poisson reconstruction is used to construct the benchmark model.

[0010] 2) Identification and screening of tunneling change areas: Identify and screen tunneling face change areas according to the set logic of seeking union inward and intersection outward. Scan the tunneling face point cloud during the tunneling process, compare the tunneling point cloud with the reference point cloud, traverse and calculate the Euclidean distance between each point in the tunneling point cloud and the reference point cloud, set a threshold T, and determine it as an unchanged point if it is less than the threshold, and as an outlier if it is greater than the threshold. Other point clouds are determined as potential change point clouds. Further screen the potential change point cloud, find the center C of the reference point cloud, and draw a ray from the center C of the point cloud to the potential change point P. Calculate the intersection I of the ray and the reference model, calculate the length of CP and CI and compare them. If CP is greater than CI, it is an external tunneling change point, and if it is less than or equal to CI, it is an internal point. All external points are retained and all internal points are deleted.

[0011] 3) Point cloud fusion and excavation face update and mapping: The filtered point cloud is fused to the reference point cloud, and the point cloud in a certain neighborhood of the intersection of the ray between the center point C of the reference point cloud and the external excavation point P and the reference model is deleted as the post-excavation point cloud, which is also the reference point cloud for the next update and mapping. The post-excavation point cloud is operated in the same way as in the first step to obtain a 3D model as the updated reference model.

[0012] (III) Beneficial effects

[0013] 1. The present invention can realize real-time three-dimensional visualization of the excavation face, accelerate the mining process, reduce manpower input, and thus significantly improve production efficiency.

[0014] 2. The present invention can reduce the need for personnel to directly participate in high-risk mining operations by adopting unmanned mining technology, thereby reducing the occurrence of accidents and injuries and improving safety. IV. Description of the drawings

[0015] Figure 1 Flowchart of the method for rapid updating and mapping of underground excavation working faces

[0016] Figure 2 Schematic diagram of the benchmark point cloud and benchmark model of the underground excavation surface

[0017] Figure 3 Schematic diagram of data scanning and mapping during underground excavation

[0018] Figure 4 Ray sampling filter diagram

[0019] Figure 5 Schematic diagram of point cloud fusion and updated mapping V. Specific implementation methods

[0020] by Figure 2-5 Taking the above as an example, the implementation process of the present invention and its specific implementation methods are described in detail.

[0021] 1. Acquisition of initial point cloud and 3D model of excavation face

[0022] 1) Acquisition of initial point cloud and 3D model of excavation face

[0023] Step 1-1: If Figure 2 As shown in the figure, the laser radar is placed in the center of the roof of the tunnel to ensure that the position remains unchanged during scanning. At time t, the initial point cloud P is obtained by scanning. init , perform body mass centroid downsampling on the initial point cloud according to formula (1) to obtain the reference point cloud P base ;

[0024]

[0025] Where m is the number of point cloud points in the voxel;

[0026] Step 1-2: Calculate the reference point cloud P base The normal n and the point cloud center C, for the point cloud P base The normal vector of each point is calculated according to formula (2) to calculate the angle between its normal vector n and the vector CP from the center of the point cloud to the point of the point cloud. If the angle is an obtuse angle, the direction of the normal vector n remains unchanged. If it is an acute angle, the normal vector is flipped. Then, the corrected normal vector N is obtained, and shielded Poisson reconstruction is performed based on N to construct the reference point cloud P. base The corresponding three-dimensional model M init ;

[0027]

[0028] Among them, CP is the vector from the center of the point cloud to the point cloud point, and n is the normal vector of the point cloud point;

[0029] 2) Identification and screening of tunneling change areas

[0030] Step 2-1: If Figure 3 As shown, scanning obtains the excavation process point cloud P process According to formula (3), the tunneling point cloud is denoised to eliminate the influence of dust noise and calculate the tunneling point cloud P process The shortest Euclidean distance d of each point to the reference point cloud i According to the distance thresholds α and β, internal noise points and morphologically unchanged points are deleted to screen out the potential excavation area point cloud P m ;

[0031] P m = {P process |α<d i <β}#(3)

[0032] Step 2-2: If Figure 4 As shown, starting from the center point C of the reference point cloud to the potential excavation area point cloud Pm Each point draws a ray R n , ray and benchmark model M init The intersection point is I i , calculate the distance CP from C to the potential tunneling point cloud mi and C to I i Distance CI i , compare CP according to formula (4) mi With CI i The size of CP mi Greater than CI i Then determine the potential excavation point P mi For external excavation point cloud, if CP mi Less than CI i Then determine P mi The internal noise point is deleted, and then the external excavation point P is obtained. out ;

[0033] P out = {P m |CP mi >CI i}#(4)

[0034] 3) Point cloud fusion and excavation face update and mapping

[0035] Step 3-1: Fuse the filtered excavation point cloud with the reference point cloud, and delete the ray R corresponding to the excavation point cloud according to formula (5). n Compared with the baseline model M init The intersection I n All points in the neighborhood η and P out Fusion is performed to obtain the updated reference point cloud P update , then, for P update Construct a 3D model according to steps 1-1 and 1-2 in Section 1. Figure 5 As shown, the updated baseline model M is obtained updsate , and serves as the benchmark model for subsequent map updates;

[0036] P update =P out ∪(P base -{P' base |d(I n )<η})#(5)

[0037] Among them, P′ base is the point set of the ray intersection neighborhood η, d(I n ) is the distance from the ray intersection point to the surrounding neighborhood points.

[0038] The specific embodiments described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A real-time updating and mapping method for underground excavation faces based on set logic drive, characterized in that: It includes the following: 1) Acquisition of initial point cloud and 3D model of excavation face Step 1-1: As shown in Figure 2, place the laser radar in the center of the top plate of the tunnel to ensure that the position remains unchanged during scanning. At time t, scan to obtain the initial point cloud P init , perform body mass centroid downsampling on the initial point cloud according to formula (1) to obtain the reference point cloud P base ; Where m is the number of point cloud points in the voxel; Step 1-2: Calculate the reference point cloud P base The normal n and the point cloud center C, for the point cloud P base The normal vector of each point is calculated according to formula (2) to calculate the angle between its normal vector n and the vector CP from the center of the point cloud to the point of the point cloud. If the angle is an obtuse angle, the direction of the normal vector n remains unchanged. If it is an acute angle, the normal vector is flipped. Then, the corrected normal vector N is obtained, and shielded Poisson reconstruction is performed based on N to construct the reference point cloud P. base The corresponding three-dimensional model M init ; Among them, CP is the vector from the center of the point cloud to the point cloud point, and n is the normal vector of the point cloud point; 2) Identification and screening of tunneling change areas Step 2-1: As shown in Figure 3, scan and obtain the point cloud P of the excavation process process According to formula (3), the tunneling point cloud is denoised to eliminate the influence of dust noise and calculate the tunneling point cloud P process The shortest Euclidean distance d of each point to the reference point cloud i According to the distance thresholds α and β, internal noise points and morphologically unchanged points are deleted to screen out the potential excavation area point cloud P m ; P m ={P process |a<d i <b} #(3) Step 2-2: As shown in Figure 4, start from the center point C of the reference point cloud to the potential excavation area point cloud P m Each point draws a ray R n , ray and benchmark model M init The intersection point is I i , calculate the distance CP from C to the potential tunneling point cloud mi and C to I i Distance CI i , compare CP according to formula (4) mi With CI i The size of CP mi Greater than CI i Then determine the potential excavation point P mi For external excavation point cloud, if CP mi Less than CI i Then determine P mi The internal noise point is deleted, and then the external excavation point P is obtained. out ; P out ={P m |CP mi >C.I. i }#(4) 3) Point cloud fusion and excavation face update and mapping Step 3-1: Fuse the filtered excavation point cloud with the reference point cloud, and delete the ray R corresponding to the excavation point cloud according to formula (5). n Compared with the baseline model M init The intersection I n All points in the neighborhood η and P out Fusion is performed to obtain the updated reference point cloud P update , then, for P update The 3D model is constructed according to the method of steps 1-1 and 1-2 in Section 1, as shown in Figure 5, to obtain the updated benchmark model M update , and serves as the benchmark model for subsequent map updates; P update =P out ∪(P base -{P′ base |d(I n )<η})#(5) Among them, P′ base is the point set of the ray intersection neighborhood η, d(I n ) is the distance from the ray intersection point to the surrounding neighborhood points.

Images