Method and device for extracting single object of power equipment

Abstract

The invention discloses a power equipment monomer object extraction method and device, and the method comprises the steps: carrying out the recognition and deletion of a connection cross rod of three-dimensional point cloud data of a to-be-extracted transformer substation, carrying out the deletion of points irrelevant to monomer equipment through primary screening, carrying out the secondary screening based on a local point cloud form factor index, and carrying out the extraction of a single object of the to-be-extracted transformer substation. And deleting noisy points included in the single equipment, and finally clustering the scene point cloud data of the single equipment obtained after two times of screening to complete extraction of the single equipment. According to the invention, automation of classification of the to-be-extracted single devices in the transformer substation is realized, and the simulation efficiency of different power devices in the transformer substation is improved. Meanwhile, the non-single equipment and the noisy points of the single equipment are screened out, errors caused by point clouds to which the non-single equipment participating in clustering belong can be reduced, and the accuracy of single equipment extraction is greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of three-dimensional point clouds, and in particular to a method and a device for extracting a single object of electric power equipment. Background Art

[0002] Digital twin technology is a key technology for building a digital twin sharing platform for 500kV terminal substations. It generally refers to an information technology that uses a variety of technologies to achieve real-time bidirectional synchronous mapping of physical real space and digital virtual space and virtual-real interaction. However, a substation often has hundreds of power equipment and their components, and there are large differences between different power equipment components. Therefore, in the current simulation of substations, manual correction methods are relied on to perform geometric correction on the three-dimensional model of power equipment in the actual installation and operation stage to make it more in line with the actual scene. However, the manual correction method is too cumbersome and inefficient.

[0003] Therefore, there is an urgent need for a strategy for extracting single power equipment objects to solve the technical problem of low simulation efficiency of different power equipment in substations. Summary of the invention

[0004] The embodiment of the present invention provides a method and device for extracting a single object of an electric power device, so as to solve the technical problem of low simulation efficiency of different electric power devices in a substation.

[0005] In order to solve the above problem, an embodiment of the present invention provides a method for extracting a single object of an electric device, comprising:

[0006] Acquire three-dimensional point cloud data of a substation to be extracted; wherein the substation to be extracted includes several devices;

[0007] Identify the connecting crossbars on the three-dimensional point cloud data, and determine a number of connecting crossbar seed points in the three-dimensional point cloud data; wherein the connecting crossbars are used to connect every two devices;

[0008] Based on the preset local point cloud morphological factor index, the target seed point is confirmed in the connecting crossbar seed point, and the target seed point is deleted to obtain the single device scene point cloud data of the substation to be extracted;

[0009] The single device scene point cloud data is clustered and merged to obtain a number of single devices corresponding to the substation to be extracted.

[0010] As an improvement of the above solution, the identification of the connecting crossbars on the three-dimensional point cloud data and the determination of a plurality of connecting crossbar seed points in the three-dimensional point cloud data include:

[0011] Performing a connecting rod recognition operation on each three-dimensional point cloud point in the three-dimensional point cloud data until the connecting rod recognition operation has been performed on all three-dimensional point cloud points to determine a number of connecting cross bar seed points;

[0012] The connection rod identification operation is as follows:

[0013] Selecting a 3D point cloud point on which a connecting rod recognition operation has not been performed in the 3D point cloud data;

[0014] Downsampling the three-dimensional point cloud points according to a preset spatial resolution to obtain downsampled point cloud points;

[0015] Based on a preset spatial query voxel resolution, perform a spatial voxel query on the downsampled point cloud points and a query on the three-dimensional column where the voxels are located to obtain the number of point clouds in the voxels and the number of point clouds in the three-dimensional column; wherein a three-dimensional column includes a plurality of voxels;

[0016] When the number of point clouds in the voxel is equal to the number of point clouds in the three-dimensional column, a connecting crossbar seed point is determined, and the connecting bar identification operation is re-executed;

[0017] Otherwise, perform the connecting rod identification operation again.

[0018] As an improvement of the above solution, the step of confirming the target seed point in the connecting crossbar seed point based on the preset local point cloud morphological factor includes:

[0019] Based on the preset point cloud resolution, a ball query is performed on each connecting crossbar seed point to construct a local point cloud area of ​​each connecting crossbar seed point;

[0020] Obtain the maximum height difference and crossbar diameter corresponding to each local point cloud area to determine whether the local point cloud morphological factor index is met; wherein the point cloud morphological factor index meets the following conditions;

[0021]

[0022] Where, dH local is the maximum height difference corresponding to a local point cloud area, max(H local ) is the maximum height corresponding to a local point cloud area, min(H locaL ) is the minimum height corresponding to a local point cloud area, D Bar is the crossbar diameter corresponding to a local point cloud area, α limit is the maximum inclination angle allowed for the connecting crossbar;

[0023] If it is satisfied, the connecting crossbar seed point that meets the local point cloud morphological factor index is confirmed as the target seed point;

[0024] If not satisfied, no operation is performed.

[0025] As an improvement of the above solution, clustering and merging the scene point cloud data of the single device to obtain a number of single devices corresponding to the substation to be extracted includes:

[0026] In the single device scene point cloud data, based on a preset clustering radius and a minimum number of three-dimensional points, the single device scene point cloud data is clustered by DBSCAN to obtain a plurality of cluster object point cloud sets;

[0027] Sequentially number each cluster object point cloud set;

[0028] Based on the serial number of each cluster object point cloud set, as the label of all three-dimensional point cloud points in the corresponding cluster object point cloud set, a number of three-dimensional point cloud points with labels are obtained;

[0029] All labeled three-dimensional point cloud points are merged and optimized to obtain several single devices corresponding to the substation to be extracted.

[0030] As an improvement of the above solution, the above-mentioned merging and optimizing operation is performed on all the three-dimensional point cloud points with labels to obtain several single devices corresponding to the substation to be extracted, including:

[0031] Based on the empirical distribution function of the number of 3D point cloud points, all tagged 3D point cloud points are divided according to a preset statistical distribution law to determine a large sample object set and a small sample object set;

[0032] In each small sample object set, each 3D point cloud point is queried to determine the adjacent 3D point cloud points corresponding to each 3D point cloud point; wherein the target 3D point cloud point is specifically: a point cloud point that is adjacent to a 3D point cloud point and belongs to a different set from that of the adjacent 3D point cloud point;

[0033] In each small sample object set, a neighboring merge operation is performed on all three-dimensional point cloud points, and after the neighboring merge operation is performed on all three-dimensional point cloud points in all small sample object sets, a number of single devices of the substation to be picked up are obtained; wherein each single device corresponds to a large sample object set;

[0034] The adjacent merging operation is specifically as follows:

[0035] Determine the adjacent three-dimensional point cloud points corresponding to each three-dimensional point cloud point;

[0036] If the adjacent 3D point cloud point belongs to the large sample object set, then all the 3D point cloud points in the small sample object set to which the 3D point cloud point corresponding to the current adjacent 3D point cloud point belongs are merged into the large sample object set to which the current adjacent point cloud point belongs;

[0037] Otherwise, remove the current 3D point cloud point.

[0038] Accordingly, an embodiment of the present invention further provides a device for extracting a single object of an electric power device, comprising: a data acquisition module, a data identification module, a data deletion module and a data merging module;

[0039] The data acquisition module is used to acquire three-dimensional point cloud data of a substation to be extracted; wherein the substation to be extracted includes several devices;

[0040] The data identification module is used to identify the connecting crossbars in the three-dimensional point cloud data and determine a number of connecting crossbar seed points in the three-dimensional point cloud data; wherein the connecting crossbars are used to connect every two devices;

[0041] The data deletion module is used to confirm the target seed point in the connecting crossbar seed point based on the preset local point cloud morphological factor index, and delete the target seed point to obtain the single device scene point cloud data of the substation to be extracted;

[0042] The data merging module is used to cluster and merge the single device scene point cloud data to obtain a number of single devices corresponding to the substation to be extracted.

[0043] As an improvement of the above solution, the data identification module includes:

[0044] Performing a connecting rod recognition operation on each three-dimensional point cloud point in the three-dimensional point cloud data until the connecting rod recognition operation has been performed on all three-dimensional point cloud points to determine a number of connecting cross bar seed points;

[0045] The connection rod identification operation is as follows:

[0046] Selecting a 3D point cloud point on which a connecting rod recognition operation has not been performed in the 3D point cloud data;

[0047] Downsampling the three-dimensional point cloud points according to a preset spatial resolution to obtain downsampled point cloud points;

[0048] Based on a preset spatial query voxel resolution, perform a spatial voxel query on the downsampled point cloud points and a query on the three-dimensional column where the voxels are located to obtain the number of point clouds in the voxels and the number of point clouds in the three-dimensional column; wherein a three-dimensional column includes a plurality of voxels;

[0049] When the number of point clouds in the voxel is equal to the number of point clouds in the three-dimensional column, a connecting crossbar seed point is determined, and the connecting bar identification operation is re-executed;

[0050] Otherwise, perform the connecting rod identification operation again.

[0051] As an improvement of the above solution, the step of confirming the target seed point in the connecting crossbar seed point based on the preset local point cloud morphological factor includes:

[0052] Based on the preset point cloud resolution, a ball query is performed on each connecting crossbar seed point to construct a local point cloud area of ​​each connecting crossbar seed point;

[0053] Obtain the maximum height difference and crossbar diameter corresponding to each local point cloud area to determine whether the local point cloud morphological factor index is met; wherein the point cloud morphological factor index meets the following conditions;

[0054]

[0055] Where, dH local is the maximum height difference corresponding to a local point cloud area, max(H local ) is the maximum height corresponding to a local point cloud area, min(H locaL ) is the minimum height corresponding to a local point cloud area, D Bar is the crossbar diameter corresponding to a local point cloud area, α limit is the maximum inclination angle allowed for the connecting crossbar;

[0056] If it is satisfied, the connecting crossbar seed point that meets the local point cloud morphological factor index is confirmed as the target seed point;

[0057] If not satisfied, no operation is performed.

[0058] As an improvement of the above solution, the data merging module includes: a clustering unit, a numbering unit, a labeling unit and a merging optimization unit;

[0059] The clustering unit is used to cluster the single device scene point cloud data through DBSCAN based on a preset clustering radius and a minimum three-dimensional point number in the single device scene point cloud data to obtain a plurality of cluster object point cloud sets;

[0060] The numbering unit is used to sequentially number each cluster object point cloud set;

[0061] The labeling unit is used to obtain a number of three-dimensional point cloud points with labels based on the serial number of each cluster object point cloud set as the label of all three-dimensional point cloud points in the corresponding cluster object point cloud set;

[0062] The merging and optimizing unit is used to perform a merging and optimizing operation on all three-dimensional point cloud points with labels to obtain a number of single devices corresponding to the substation to be extracted.

[0063] As an improvement of the above solution, the above-mentioned merging and optimizing operation is performed on all the three-dimensional point cloud points with labels to obtain several single devices corresponding to the substation to be extracted, including:

[0064] Based on the empirical distribution function of the number of 3D point cloud points, all tagged 3D point cloud points are divided according to a preset statistical distribution law to determine a large sample object set and a small sample object set;

[0065] In each small sample object set, each 3D point cloud point is queried to determine the adjacent 3D point cloud points corresponding to each 3D point cloud point; wherein the target 3D point cloud point is specifically: a point cloud point that is adjacent to a 3D point cloud point and belongs to a different set from that of the adjacent 3D point cloud point;

[0066] In each small sample object set, a neighboring merge operation is performed on all three-dimensional point cloud points, and after the neighboring merge operation is performed on all three-dimensional point cloud points in all small sample object sets, a number of single devices of the substation to be picked up are obtained; wherein each single device corresponds to a large sample object set;

[0067] The adjacent merging operation is specifically as follows:

[0068] Determine the adjacent three-dimensional point cloud points corresponding to each three-dimensional point cloud point;

[0069] If the adjacent 3D point cloud point belongs to the large sample object set, then all the 3D point cloud points in the small sample object set to which the 3D point cloud point corresponding to the current adjacent 3D point cloud point belongs are merged into the large sample object set to which the current adjacent point cloud point belongs;

[0070] Otherwise, remove the current 3D point cloud point.

[0071] As can be seen from the above, the present invention has the following beneficial effects:

[0072] The present invention provides a method for extracting single objects of electric power equipment, by identifying and deleting connecting cross bars of three-dimensional point cloud data of a substation to be extracted, deleting points irrelevant to single equipment through a primary screening, and at the same time, performing a secondary screening based on local point cloud morphological factor indicators to delete noise points included in single equipment, and finally clustering the single equipment scene point cloud data obtained after the two screenings, thereby completing the extraction of single equipment. The present invention realizes the automation of the classification of single equipment in the substation to be extracted, and improves the simulation efficiency of different electric power equipment in the substation. At the same time, the present invention can reduce the error caused by the point cloud belonging to the non-single equipment participating in the clustering by screening out the noise points of non-single equipment and single equipment, thereby greatly improving the accuracy of the extraction of single equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] Figure 1 It is a flowchart of a method for extracting a single object of an electric power device provided by an embodiment of the present invention;

[0074] Figure 2 It is a structural schematic diagram of a device for extracting a single object of an electric power device provided by an embodiment of the present invention;

[0075] Figure 3 This is a schematic diagram of the structure of a terminal device provided by an embodiment of the present invention;

[0076] Figure 4 is a schematic diagram of a query voxel and a stereo column provided by an embodiment of the present invention;

[0077] Figure 5 is a schematic diagram of the crossbar diameter provided by an embodiment of the present invention;

[0078] Figure 6 is a schematic diagram of the effect of merging provided by an embodiment of the present invention;

[0079] Figure 7 It is a schematic diagram comparing experimental results in two scenarios provided in one embodiment of the present invention.

[0080] Figure 8 It is a flowchart of a method for extracting a single object of electric power equipment provided in another embodiment of the present invention. DETAILED DESCRIPTION

[0081] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only 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 ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0082] Embodiment 1

[0083] See also Figure 1 , Figure 1 FIG. 1 is a flow chart of a method for extracting a single object of an electric power device provided by an embodiment of the present invention. Figure 1 As shown, this embodiment includes steps 101 to 104, and each step is specifically as follows:

[0084] Step 101: Acquire three-dimensional point cloud data of a substation to be extracted; wherein the substation to be extracted includes a plurality of equipment.

[0085] In this embodiment, in a specific embodiment, a three-dimensional scan is performed on the substation to be extracted to obtain three-dimensional data, and a cloth algorithm is used on the three-dimensional data, and then the ground point data in the three-dimensional data is identified and eliminated to obtain three-dimensional point cloud data, providing a data basis for the subsequent extraction of the single point cloud of the power equipment. The cloth algorithm is currently the most widely used algorithm for removing ground points from point clouds. Its basic principle is to identify and distinguish ground points in the point cloud scene by inverting the point cloud and simulating the process of cloth falling on the point cloud. In the scene of the substation to be extracted, since the ground terrain of the substation to be extracted is flat, the use of the cloth algorithm can efficiently and accurately eliminate the ground points. Therefore, this method implements the classic cloth algorithm to form an automatic removal module for ground points in the substation site cloud data, thereby completing the removal of ground points in the three-dimensional point cloud of the substation.

[0086] Step 102: Identify the connecting crossbars in the three-dimensional point cloud data, and determine a number of connecting crossbar seed points in the three-dimensional point cloud data; wherein the connecting crossbars are used to connect every two devices.

[0087] In this embodiment, the identifying of the connecting crossbars on the three-dimensional point cloud data and determining a number of connecting crossbar seed points in the three-dimensional point cloud data include:

[0088] Performing a connecting rod recognition operation on each three-dimensional point cloud point in the three-dimensional point cloud data until the connecting rod recognition operation has been performed on all three-dimensional point cloud points to determine a number of connecting cross bar seed points;

[0089] The connection rod identification operation is as follows:

[0090] Selecting a 3D point cloud point on which a connecting rod recognition operation has not been performed in the 3D point cloud data;

[0091] Downsampling the three-dimensional point cloud points according to a preset spatial resolution to obtain downsampled point cloud points;

[0092] Based on a preset spatial query voxel resolution, perform a spatial voxel query on the downsampled point cloud points and a query on the three-dimensional column where the voxels are located to obtain the number of point clouds in the voxels and the number of point clouds in the three-dimensional column; wherein a three-dimensional column includes a plurality of voxels;

[0093] When the number of point clouds in the voxel is equal to the number of point clouds in the three-dimensional column, a connecting crossbar seed point is determined, and the connecting bar identification operation is re-executed;

[0094] Otherwise, perform the connecting rod identification operation again.

[0095] It should be noted that the devices in the substation scene to be extracted are connected to each other through metal connecting bars. In order to accurately separate the devices, it is necessary to identify and eliminate the connecting bars between the devices from the point cloud.

[0096] In a specific embodiment, this embodiment proposes a module for automatically generating seed points for connecting crossbars of substation equipment, which mainly includes the following steps:

[0097] Firstly, a fast filtering operator based on point cloud voxelization and statistical distribution characteristics is designed to realize the automatic generation module of seed points of substation equipment connecting crossbars, obtain seed points that may belong to connecting crossbars from the substation scene point cloud, and use these seed points to calculate the morphological feature factors of the crossbar point cloud. It consists of the following sub-steps:

[0098] The first step is to perform voxel downsampling on the three-dimensional point cloud data in order to reduce the time consumption of data processing; the second step is to set the spatial resolution of downsampling to 0.1 meters in order to take into account the integrity of the amount of information expressed by the point cloud; the third step is to consider that the diameter of the connecting crossbar of the power equipment is about 0.25 meters, and we generate spatial query voxels with a resolution of 0.25 meters for the point cloud; the fourth step is to perform spatial voxel query of the point cloud and the three-dimensional column query where the voxel is located voxel by voxel, and count the number of point clouds N_voxel in the voxel (that is, the number of point clouds in the voxel described in the present invention) and the number of point clouds N_proj in the three-dimensional column (that is, the number of point clouds in the three-dimensional column described in the present invention). If N_voxel and N_proj are approximately equal, then all points in the spatial query voxel are added to the seed point set of the connecting crossbar of the power equipment; otherwise, it is considered that the target spatial query voxel contains the equipment object and does not contain the connecting crossbar, and all voxels within the query range of the three-dimensional column are removed from the set of voxels to be queried. By traversing each spatial query voxel, the final output is a set of seed points for connecting crossbars of power equipment.

[0099] For better explanation, see Figure 4 In the figure, the blue cube is the spatial query voxel, and the transparent cuboid is the three-dimensional column where the spatial query voxel is located. The three-dimensional column can contain one or more spatial query voxels.

[0100] Step 103: Based on the preset local point cloud morphological factor index, the target seed point is confirmed in the connecting crossbar seed point, and the target seed point is deleted to obtain the single device scene point cloud data of the substation to be extracted.

[0101] In this embodiment, the step of confirming the target seed point among the connecting crossbar seed points based on the preset local point cloud morphological factor includes:

[0102] Based on the preset point cloud resolution, a ball query is performed on each connecting crossbar seed point to construct a local point cloud area of ​​each connecting crossbar seed point;

[0103] Obtain the maximum height difference and crossbar diameter corresponding to each local point cloud area to determine whether the local point cloud morphological factor index is met; wherein the point cloud morphological factor index meets the following conditions;

[0104]

[0105] Where, dH local is the maximum height difference corresponding to a local point cloud area, max(H local ) is the maximum height corresponding to a local point cloud area, min(H locaL ) is the minimum height corresponding to a local point cloud area, D Bar is the crossbar diameter corresponding to a local point cloud area, α limit is the maximum inclination angle allowed for the connecting crossbar;

[0106] If it is satisfied, the connecting crossbar seed point that meets the local point cloud morphological factor index is confirmed as the target seed point;

[0107] If not satisfied, no operation is performed.

[0108] In a specific embodiment, a device connection rod elimination technology based on the local point cloud morphological factor index is proposed, which realizes the automatic elimination module of the substation equipment connection crossbar and completes the accurate elimination of the power equipment connection crossbar. The main direction of the power equipment connection crossbar is approximately 90° with the Z-axis direction of the three-dimensional space, so the maximum height difference in the height histogram of the local point cloud of the connection crossbar should be roughly similar to the diameter of the crossbar. Based on this structural feature, it is specifically composed of the following sub-steps:

[0109] In the first step, the seed points of the connecting rods are used as input, and the identification and elimination of the connecting rods of the equipment are performed point by point. In the first step, the seed points of the connecting crossbars are used as input, and the ball query is performed at the points with a point cloud resolution of 0.1m to construct the local point cloud of the connecting crossbars; in the second step, the height statistical histogram of the local point cloud is generated, and the maximum height difference dH of the local point cloud is obtained. local ; The third step is to define the local point cloud morphological factor index, which is used to determine whether the local point cloud is part of the connecting crossbar. This index compares the relationship between the local point cloud height and the crossbar diameter, and allows the crossbar to have a certain inclination (participate in Figure 5 ), to accurately identify the connecting crossbar, that is, when the maximum height difference dH of the local point cloud local With crossbar diameter D BarWhen the local point cloud morphological factor index is met, it is considered that the selected local point cloud conforms to the placement rule of the power equipment connection crossbar, and the selected local point cloud belongs to a part of the power equipment connection crossbar, and can be eliminated. By traversing the seed points of all connection crossbars and performing the above technical operations, the power equipment connection crossbars can be eliminated to obtain independent equipment point clouds.

[0110] Step 104: clustering and merging the single device scene point cloud data to obtain a number of single devices corresponding to the substation to be extracted.

[0111] In this embodiment, clustering and merging the single device scene point cloud data to obtain a number of single devices corresponding to the substation to be extracted includes:

[0112] In the single device scene point cloud data, based on a preset clustering radius and a minimum number of three-dimensional points, the single device scene point cloud data is clustered by DBSCAN to obtain a plurality of cluster object point cloud sets;

[0113] Sequentially number each cluster object point cloud set;

[0114] Based on the serial number of each cluster object point cloud set, as the label of all three-dimensional point cloud points in the corresponding cluster object point cloud set, a number of three-dimensional point cloud points with labels are obtained;

[0115] All labeled three-dimensional point cloud points are merged and optimized to obtain several single devices corresponding to the substation to be extracted.

[0116] In a specific embodiment, after eliminating the equipment connection crossbar, the substation equipment is relatively independent of each other in the single equipment scene point cloud data. Based on this, a DBSCAN-based power equipment reliability extraction method is proposed, which realizes the automatic recognition and extraction of a large number of single point cloud objects of power equipment in complex substation cloud scenes. It mainly includes a coarse-grained extraction module of substation power equipment single point cloud and a substation power equipment single point cloud merging optimization module. The coarse-grained extraction module of substation power equipment single point cloud mainly uses the DBSCAN method to complete the extraction of single equipment, but this clustering result is unreliable. On the one hand, the noise points in the point cloud may be clustered into noise classes. On the other hand, due to the complexity and uneven density of power equipment, its equipment components may be clustered into a single class. Therefore, the substation power equipment single point cloud merging optimization module aims to merge and optimize the coarse-grained clustering results according to the statistical characteristics of the clustering object point cloud size and the connectivity between different clustering objects, so as to complete the precise individualization of the equipment. It mainly includes the following steps:

[0117] Firstly, a point cloud coarse-grained clustering method based on DBSCAN technology was proposed, and the coarse-grained extraction module of the single point cloud of the substation power equipment was realized, and the coarse-grained single point cloud of the equipment in the substation scenario was obtained. This method uses the consistency of point cloud density as an indicator to perform growth clustering of the point cloud to form a set of several point cloud objects. In the clustering process, considering the actual size of the equipment object, the clustering radius of DBSCAN is set to 0.5 meters, and the minimum number of three-dimensional points of the clustering object is limited to 50, and the clustering objects with less than 50 points are removed as noise points. After the clustering is completed, the sequential encoding of each clustering object is obtained, and it is assigned to each point as a type label to obtain a three-dimensional point cloud point with a type label.

[0118] Secondly, a clustering merging optimization technology based on the statistical distribution of sample points and the connectivity of point clouds is proposed, which realizes the single point cloud merging optimization module of substation power equipment and completes the precise advance of the point cloud objects of power equipment. This technology consists of the following sub-steps: The first step is to generate the empirical distribution function of the number of three-dimensional points of the clustering object, determine the heterogeneity of the statistical distribution, and calculate the segmentation point as the threshold according to the 20 / 80 statistical distribution law, so as to divide the clustering object into a 20% large sample object set and an 80% small sample object set; the second step is to perform a ball query on each three-dimensional point in the object set for each small sample object set to obtain the three-dimensional point labels of all other objects connected to the object set; the third step is to analyze the adjacent sample labels. When there is a label of a large sample object in the adjacent sample labels, all the three-dimensional points in the small sample object are merged into the large sample object; otherwise, the small sample object is considered to be a discrete noise point and is removed from the sample set. By traversing each three-dimensional point in each small sample object, the small sample object is deleted or merged into the corresponding large sample object to obtain a single point cloud object of power equipment with higher precision. See the merging effect. Figure 6 .

[0119] In this embodiment, the merging and optimization operation of all labeled three-dimensional point cloud points is performed to obtain several single devices corresponding to the substation to be extracted, including:

[0120] Based on the empirical distribution function of the number of 3D point cloud points, all tagged 3D point cloud points are divided according to a preset statistical distribution law to determine a large sample object set and a small sample object set;

[0121] In each small sample object set, each 3D point cloud point is queried to determine the adjacent 3D point cloud points corresponding to each 3D point cloud point; wherein the target 3D point cloud point is specifically: a point cloud point that is adjacent to a 3D point cloud point and belongs to a different set from that of the adjacent 3D point cloud point;

[0122] In each small sample object set, a neighboring merge operation is performed on all three-dimensional point cloud points, and after the neighboring merge operation is performed on all three-dimensional point cloud points in all small sample object sets, a number of single devices of the substation to be picked up are obtained; wherein each single device corresponds to a large sample object set;

[0123] The adjacent merging operation is specifically as follows:

[0124] Determine the adjacent three-dimensional point cloud points corresponding to each three-dimensional point cloud point;

[0125] If the adjacent 3D point cloud point belongs to the large sample object set, then all the 3D point cloud points in the small sample object set to which the 3D point cloud point corresponding to the current adjacent 3D point cloud point belongs are merged into the large sample object set to which the current adjacent point cloud point belongs;

[0126] Otherwise, remove the current 3D point cloud point.

[0127] For a better explanation, see Figure 7 , to illustrate the method for extracting single objects of electric equipment provided by another embodiment of the present application, including: a module for automatically removing ground points of substation cloud data, a module for automatically generating seed points of connecting crossbars of substation equipment, a module for automatically removing connecting crossbars of substation equipment, a module for extracting coarse-grained single point cloud of electric equipment in substation, and a module for merging and optimizing single point cloud of electric equipment in substation. This method first processes the cloud data of substation site, i.e., removing ground points in the cloud data of substation site by using the cloth algorithm; secondly, eliminating the connecting crossbars of equipment in the cloud data of substation site, i.e., identifying the seed points of connecting crossbars based on the voxelization and statistical distribution characteristics of point clouds, and removing the connecting crossbars of equipment point clouds by designing and calculating the morphological factor index of local point clouds; thirdly, the electric equipment is monomerized, i.e., a coarse-grained single point cloud is formed by using the DBSCAN clustering algorithm, and on this basis, the equipment objects and component objects are classified according to the statistical characteristics of the single point cloud, and the component objects are assigned to the nearest equipment objects based on the spatial connectivity of the point cloud, so as to realize the reliable extraction and segmentation of single point cloud of electric equipment.

[0128] In a specific embodiment, in order to verify the applicability and generalization of this technology, the experiment selected substation 3D point cloud data obtained by two different types of equipment, and this method was verified in the 3D point clouds of these two substation scenes. Scene 1 is based on the 3D point cloud obtained by the ground-based laser scanner, and scene 2 is based on the 3D point cloud obtained by the drone laser scanner. Figure 8As shown in the figure, the experimental results show that in both scenarios, the complex single point clouds of power equipment in the substation can be identified and extracted, which illustrates the applicability of this technology in extracting single point clouds of different power equipment in substation scenarios, and can achieve good equipment individualization effect for substations of different designs, with high generalization.

[0129] See also Figure 2 , Figure 2 It is a structural schematic diagram of a power equipment monomer object extraction device provided by an embodiment of the present invention, comprising: a data acquisition module 201, a data identification module 202, a data deletion module 203 and a data merging module 204;

[0130] The data acquisition module is used to acquire three-dimensional point cloud data of a substation to be extracted; wherein the substation to be extracted includes several devices;

[0131] The data identification module is used to identify the connecting crossbars in the three-dimensional point cloud data and determine a number of connecting crossbar seed points in the three-dimensional point cloud data; wherein the connecting crossbars are used to connect every two devices;

[0132] The data deletion module is used to confirm the target seed point in the connecting crossbar seed point based on the preset local point cloud morphological factor index, and delete the target seed point to obtain the single device scene point cloud data of the substation to be extracted;

[0133] The data merging module is used to cluster and merge the single device scene point cloud data to obtain a number of single devices corresponding to the substation to be extracted.

[0134] As an improvement of the above solution, the data identification module includes:

[0135] Performing a connecting rod recognition operation on each three-dimensional point cloud point in the three-dimensional point cloud data until the connecting rod recognition operation has been performed on all three-dimensional point cloud points to determine a number of connecting cross bar seed points;

[0136] The connection rod identification operation is as follows:

[0137] Selecting a 3D point cloud point on which a connecting rod recognition operation has not been performed in the 3D point cloud data;

[0138] Downsampling the three-dimensional point cloud points according to a preset spatial resolution to obtain downsampled point cloud points;

[0139] Based on a preset spatial query voxel resolution, perform a spatial voxel query on the downsampled point cloud points and a query on the three-dimensional column where the voxels are located to obtain the number of point clouds in the voxels and the number of point clouds in the three-dimensional column; wherein a three-dimensional column includes a plurality of voxels;

[0140] When the number of point clouds in the voxel is equal to the number of point clouds in the three-dimensional column, a connecting crossbar seed point is determined, and the connecting bar identification operation is re-executed;

[0141] Otherwise, perform the connecting rod identification operation again.

[0142] As an improvement of the above solution, the step of confirming the target seed point in the connecting crossbar seed point based on the preset local point cloud morphological factor includes:

[0143] Based on the preset point cloud resolution, a ball query is performed on each connecting crossbar seed point to construct a local point cloud area of ​​each connecting crossbar seed point;

[0144] Obtain the maximum height difference and crossbar diameter corresponding to each local point cloud area to determine whether the local point cloud morphological factor index is met; wherein the point cloud morphological factor index meets the following conditions;

[0145]

[0146] Where, dH local is the maximum height difference corresponding to a local point cloud area, max(H local ) is the maximum height corresponding to a local point cloud area, min(H locaL ) is the minimum height corresponding to a local point cloud area, D Bar is the crossbar diameter corresponding to a local point cloud area, α limit is the maximum inclination angle allowed for the connecting crossbar;

[0147] If it is satisfied, the connecting crossbar seed point that meets the local point cloud morphological factor index is confirmed as the target seed point;

[0148] If not satisfied, no operation is performed.

[0149] As an improvement of the above solution, the data merging module includes: a clustering unit, a numbering unit, a labeling unit and a merging optimization unit;

[0150] The clustering unit is used to cluster the single device scene point cloud data through DBSCAN based on a preset clustering radius and a minimum three-dimensional point number in the single device scene point cloud data to obtain a plurality of cluster object point cloud sets;

[0151] The numbering unit is used to sequentially number each cluster object point cloud set;

[0152] The labeling unit is used to obtain a number of three-dimensional point cloud points with labels based on the serial number of each cluster object point cloud set as the label of all three-dimensional point cloud points in the corresponding cluster object point cloud set;

[0153] The merging and optimizing unit is used to perform a merging and optimizing operation on all three-dimensional point cloud points with labels to obtain a number of single devices corresponding to the substation to be extracted.

[0154] As an improvement of the above solution, the above-mentioned merging and optimizing operation is performed on all the three-dimensional point cloud points with labels to obtain several single devices corresponding to the substation to be extracted, including:

[0155] Based on the empirical distribution function of the number of 3D point cloud points, all tagged 3D point cloud points are divided according to a preset statistical distribution law to determine a large sample object set and a small sample object set;

[0156] In each small sample object set, each 3D point cloud point is queried to determine the adjacent 3D point cloud points corresponding to each 3D point cloud point; wherein the target 3D point cloud point is specifically: a point cloud point that is adjacent to a 3D point cloud point and belongs to a different set from that of the adjacent 3D point cloud point;

[0157] In each small sample object set, a neighboring merge operation is performed on all three-dimensional point cloud points, and after the neighboring merge operation is performed on all three-dimensional point cloud points in all small sample object sets, a number of single devices of the substation to be picked up are obtained; wherein each single device corresponds to a large sample object set;

[0158] The adjacent merging operation is specifically as follows:

[0159] Determine the adjacent three-dimensional point cloud points corresponding to each three-dimensional point cloud point;

[0160] If the adjacent 3D point cloud point belongs to the large sample object set, then all the 3D point cloud points in the small sample object set to which the 3D point cloud point corresponding to the current adjacent 3D point cloud point belongs are merged into the large sample object set to which the current adjacent point cloud point belongs;

[0161] Otherwise, remove the current 3D point cloud point.

[0162] This embodiment identifies and deletes the connecting crossbars of the three-dimensional point cloud data of the substation to be extracted, deletes the points unrelated to the single device through a primary screening, and at the same time, performs a secondary screening based on the local point cloud morphological factor index to delete the noise points included in the single device. Finally, the single device scene point cloud data obtained after the two screenings are clustered to complete the extraction of the single device. The present invention realizes the automation of the classification of single devices in the substation to be extracted, and improves the simulation efficiency of different power equipment in the substation. At the same time, the present invention can reduce the error caused by the point cloud belonging to the non-single device participating in the clustering by screening out the noise points of non-single devices and single devices, and greatly improves the accuracy of the extraction of single devices.

[0163] Embodiment 2

[0164] See also Figure 3 , Figure 3 It is a schematic diagram of the structure of a terminal device provided in one embodiment of the present invention.

[0165] A terminal device of this embodiment includes: a processor 301, a memory 302, and a computer program stored in the memory 302 and executable on the processor 301. When the processor 301 executes the computer program, the steps of the above-mentioned method for extracting a single object of a power device in the embodiment are implemented, for example: Figure 1 Alternatively, when the processor executes the computer program, the functions of each module in the above-mentioned device embodiments are realized, for example: Figure 2 The electric power equipment monomer object shown extracts all modules of the device.

[0166] In addition, an embodiment of the present invention further provides a computer-readable storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the method for extracting single objects of electric power equipment as described in any of the above embodiments.

[0167] Those skilled in the art will understand that the schematic diagram is merely an example of a terminal device and does not constitute a limitation on the terminal device. The terminal device may include more or fewer components than shown in the diagram, or a combination of certain components, or different components. For example, the terminal device may also include input and output devices, network access devices, buses, etc.

[0168] The processor 301 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc. The processor 301 is the control center of the terminal device, and uses various interfaces and lines to connect various parts of the entire terminal device.

[0169] The memory 302 can be used to store the computer program and / or module. The processor 301 implements various functions of the terminal device by running or executing the computer program and / or module stored in the memory and calling the data stored in the memory 302. The memory 302 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0170] Wherein, if the module / unit integrated in the terminal device is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium, etc.

[0171] It should be noted that the device embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the accompanying drawings of the device embodiments provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines. A person of ordinary skill in the art may understand and implement it without paying any creative effort.

[0172] The above is a preferred embodiment of the present invention. It should be pointed out that a person skilled in the art can make several improvements and modifications without departing from the principle of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for extracting a single object of electric power equipment, characterized in that: include: Acquire three-dimensional point cloud data of a substation to be extracted; wherein the substation to be extracted includes several devices; Identify the connecting crossbars on the three-dimensional point cloud data, and determine a number of connecting crossbar seed points in the three-dimensional point cloud data; wherein the connecting crossbars are used to connect every two devices; Based on the preset local point cloud morphological factor index, the target seed point is confirmed in the connecting crossbar seed point, and the target seed point is deleted to obtain the single device scene point cloud data of the substation to be extracted; The single device scene point cloud data is clustered and merged to obtain a number of single devices corresponding to the substation to be extracted.

2. The method for extracting a single object of electric power equipment according to claim 1, characterized in that: The step of identifying the connecting crossbars on the three-dimensional point cloud data and determining a plurality of connecting crossbar seed points in the three-dimensional point cloud data includes: Performing a connecting rod recognition operation on each three-dimensional point cloud point in the three-dimensional point cloud data until the connecting rod recognition operation has been performed on all three-dimensional point cloud points to determine a number of connecting cross bar seed points; The connection rod identification operation is as follows: Selecting a 3D point cloud point on which a connecting rod recognition operation has not been performed in the 3D point cloud data; Downsampling the three-dimensional point cloud points according to a preset spatial resolution to obtain downsampled point cloud points; Based on a preset spatial query voxel resolution, perform a spatial voxel query on the downsampled point cloud points and a query on the three-dimensional column where the voxels are located to obtain the number of point clouds in the voxels and the number of point clouds in the three-dimensional column; wherein a three-dimensional column includes a plurality of voxels; When the number of point clouds in the voxel is equal to the number of point clouds in the three-dimensional column, a connecting crossbar seed point is determined, and the connecting bar identification operation is re-executed; Otherwise, perform the connecting rod identification operation again.

3. The method for extracting a single object of electric power equipment according to claim 2, characterized in that: The step of confirming a target seed point among the connecting crossbar seed points based on a preset local point cloud morphological factor includes: Based on the preset point cloud resolution, a ball query is performed on each connecting crossbar seed point to construct a local point cloud area of ​​each connecting crossbar seed point; Obtain the maximum height difference and crossbar diameter corresponding to each local point cloud area to determine whether the local point cloud morphological factor index is met; wherein the point cloud morphological factor index meets the following conditions; Where, dH local is the maximum height difference corresponding to a local point cloud area, max(H local ) is the maximum height corresponding to a local point cloud area, min(H locaL ) is the minimum height corresponding to a local point cloud area, D Bar is the crossbar diameter corresponding to a local point cloud area, α limit is the maximum inclination angle allowed for the connecting crossbar; If it is satisfied, the connecting crossbar seed point that meets the local point cloud morphological factor index is confirmed as the target seed point; If not satisfied, no operation is performed.

4. The method for extracting a single object of electric power equipment according to claim 3, characterized in that: The clustering and merging of the single device scene point cloud data to obtain a number of single devices corresponding to the substation to be extracted includes: In the single device scene point cloud data, based on a preset clustering radius and a minimum number of three-dimensional points, the single device scene point cloud data is clustered by DBSCAN to obtain a plurality of cluster object point cloud sets; Sequentially number each cluster object point cloud set; Based on the serial number of each cluster object point cloud set, as the label of all three-dimensional point cloud points in the corresponding cluster object point cloud set, a number of three-dimensional point cloud points with labels are obtained; All labeled three-dimensional point cloud points are merged and optimized to obtain several single devices corresponding to the substation to be extracted.

5. The method for extracting a single object of electric power equipment according to claim 4, characterized in that: The above-mentioned merging and optimizing operation is performed on all the three-dimensional point cloud points with labels to obtain several single devices corresponding to the substation to be extracted, including: Based on the empirical distribution function of the number of 3D point cloud points, all tagged 3D point cloud points are divided according to a preset statistical distribution law to determine a large sample object set and a small sample object set; In each small sample object set, each 3D point cloud point is queried to determine the adjacent 3D point cloud points corresponding to each 3D point cloud point; wherein the target 3D point cloud point is specifically: a point cloud point that is adjacent to a 3D point cloud point and belongs to a different set from that of the adjacent 3D point cloud point; In each small sample object set, a neighboring merge operation is performed on all three-dimensional point cloud points, and after the neighboring merge operation is performed on all three-dimensional point cloud points in all small sample object sets, a number of single devices of the substation to be picked up are obtained; wherein each single device corresponds to a large sample object set; The adjacent merging operation is specifically as follows: Determine the adjacent three-dimensional point cloud points corresponding to each three-dimensional point cloud point; If the adjacent 3D point cloud point belongs to the large sample object set, then all the 3D point cloud points in the small sample object set to which the 3D point cloud point corresponding to the current adjacent 3D point cloud point belongs are merged into the large sample object set to which the current adjacent point cloud point belongs; Otherwise, remove the current 3D point cloud point.

6. A device for extracting single objects of electric power equipment, characterized in that: include: Data acquisition module, data identification module, data deletion module and data merging module; The data acquisition module is used to acquire three-dimensional point cloud data of a substation to be extracted; wherein the substation to be extracted includes several devices; The data identification module is used to identify the connecting crossbars in the three-dimensional point cloud data and determine a number of connecting crossbar seed points in the three-dimensional point cloud data; wherein the connecting crossbars are used to connect every two devices; The data deletion module is used to confirm the target seed point in the connecting crossbar seed point based on the preset local point cloud morphological factor index, and delete the target seed point to obtain the single device scene point cloud data of the substation to be extracted; The data merging module is used to cluster and merge the single device scene point cloud data to obtain a number of single devices corresponding to the substation to be extracted.

7. The device for extracting a single object of electric power equipment according to claim 6, characterized in that: The data identification module comprises: Performing a connecting rod recognition operation on each three-dimensional point cloud point in the three-dimensional point cloud data until the connecting rod recognition operation has been performed on all three-dimensional point cloud points to determine a number of connecting cross bar seed points; The connection rod identification operation is as follows: Selecting a 3D point cloud point on which a connecting rod recognition operation has not been performed in the 3D point cloud data; Downsampling the three-dimensional point cloud points according to a preset spatial resolution to obtain downsampled point cloud points; Based on a preset spatial query voxel resolution, perform a spatial voxel query on the downsampled point cloud points and a query on the three-dimensional column where the voxels are located to obtain the number of point clouds in the voxels and the number of point clouds in the three-dimensional column; wherein a three-dimensional column includes a plurality of voxels; When the number of point clouds in the voxel is equal to the number of point clouds in the three-dimensional column, a connecting crossbar seed point is determined, and the connecting bar identification operation is re-executed; Otherwise, perform the connecting rod identification operation again.

8. The device for extracting a single object of electric power equipment according to claim 7, characterized in that: The step of confirming a target seed point among the connecting crossbar seed points based on a preset local point cloud morphological factor includes: Based on the preset point cloud resolution, a ball query is performed on each connecting crossbar seed point to construct a local point cloud area of ​​each connecting crossbar seed point; Obtain the maximum height difference and crossbar diameter corresponding to each local point cloud area to determine whether the local point cloud morphological factor index is met; wherein the point cloud morphological factor index meets the following conditions; Where, dH local is the maximum height difference corresponding to a local point cloud area, max(H local ) is the maximum height corresponding to a local point cloud area, min(H locaL ) is the minimum height corresponding to a local point cloud area, D Bar is the crossbar diameter corresponding to a local point cloud area, α limit is the maximum inclination angle allowed for the connecting crossbar; If it is satisfied, the connecting crossbar seed point that meets the local point cloud morphological factor index is confirmed as the target seed point; If not satisfied, no operation is performed.

9. The device for extracting a single object of electric power equipment according to claim 8, characterized in that: The data merging module includes: a clustering unit, a numbering unit, a labeling unit and a merging optimization unit; The clustering unit is used to cluster the single device scene point cloud data through DBSCAN based on a preset clustering radius and a minimum three-dimensional point number in the single device scene point cloud data to obtain a plurality of cluster object point cloud sets; The numbering unit is used to sequentially number each cluster object point cloud set; The labeling unit is used to obtain a number of three-dimensional point cloud points with labels based on the serial number of each cluster object point cloud set as the label of all three-dimensional point cloud points in the corresponding cluster object point cloud set; The merging and optimizing unit is used to perform a merging and optimizing operation on all three-dimensional point cloud points with labels to obtain a number of single devices corresponding to the substation to be extracted.

10. The device for extracting a single object of electric power equipment according to claim 9, characterized in that: The above-mentioned merging and optimizing operation is performed on all the three-dimensional point cloud points with labels to obtain several single devices corresponding to the substation to be extracted, including: Based on the empirical distribution function of the number of 3D point cloud points, all tagged 3D point cloud points are divided according to a preset statistical distribution law to determine a large sample object set and a small sample object set; In each small sample object set, each 3D point cloud point is queried to determine the adjacent 3D point cloud points corresponding to each 3D point cloud point; wherein the target 3D point cloud point is specifically: a point cloud point that is adjacent to a 3D point cloud point and belongs to a different set from that of the adjacent 3D point cloud point; In each small sample object set, a neighboring merge operation is performed on all three-dimensional point cloud points, and after the neighboring merge operation is performed on all three-dimensional point cloud points in all small sample object sets, a number of single devices of the substation to be picked up are obtained; wherein each single device corresponds to a large sample object set; The adjacent merging operation is specifically as follows: Determine the adjacent three-dimensional point cloud points corresponding to each three-dimensional point cloud point; If the adjacent 3D point cloud point belongs to the large sample object set, then all the 3D point cloud points in the small sample object set to which the 3D point cloud point corresponding to the current adjacent 3D point cloud point belongs are merged into the large sample object set to which the current adjacent point cloud point belongs; Otherwise, remove the current 3D point cloud point.

Images