Ultra-high voltage maintenance safety control method, device, equipment and storage medium

Through point cloud data processing and Rtk positioning technology, safety hazards and low accuracy in overhead ultra-high voltage transmission lines are solved, and high-precision judgment of safety areas is achieved.

CN117994549BActive Publication Date: 2024-11-15STATE GRID HUBEI EXTRA HIGH VOLTAGE CO +1

Patent Information

Application Number
CN202410029241.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-01-09
Publication Date
2024-11-15
Estimated Expiration
2044-01-09

AI Technical Summary

Technical Problem

There are safety hazards during the maintenance of existing overhead ultra-high voltage transmission lines, and the traditional safety calculation method is not accurate, resulting in misjudgment problems.

Method used

By obtaining the point cloud data set of ultra-high voltage transmission channels, resampling is performed using the preset spatial distance dilution algorithm, the ICP algorithm performs point cloud registration, the clustering algorithm improves classification, sets the rectangular box as a safe area, and uses Rtk positioning Method to determine whether the operator is in a safe area.

Benefits of technology

It improves the accuracy of safety area judgment, realizes high-precision coordinate positioning at the centimeter level, and solves safety hazards and misjudgment problems.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117994549B_ABST
    Figure CN117994549B_ABST
Patent Text Reader

Abstract

The present invention relates to the field of ultra-high voltage maintenance technology, and discloses an ultra-high voltage maintenance safety management and control method, device, equipment and storage medium. The method includes resampling a point cloud data set based on a thinning algorithm; registering the thinned subset according to an ICP algorithm; improving a preset classification algorithm based on a clustering algorithm, and classifying the registered data set based on the improved classification algorithm; setting an operation center area based on the classified data, forming a rectangular box with the area as the center point and serving as a safe area; determining the point information of the operator based on the Rtk positioning method, so as to determine whether it is in the safe area. The present invention uses a thinning algorithm to quickly locate in massive point cloud data, and uses the ICP algorithm for registration without segmenting and extracting features from the data set. The classification algorithm is improved by using a clustering algorithm to improve the classification accuracy, and a subsequent safe area is framed for standby. High-precision coordinates are determined based on the Rtk positioning method to improve the accuracy of judgment.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of ultra-high voltage maintenance technology, and in particular to an ultra-high voltage maintenance safety management and control method, device, equipment and storage medium. Background Art

[0002] Overhead ultra-high voltage transmission lines are an important way of power transmission in the power industry. The safe and stable operation of transmission lines directly affects the high reliability and stable power supply of the power system. Therefore, regular and effective inspection and maintenance of transmission lines has become an important task for the power industry. At present, the inspection of overhead ultra-high voltage transmission lines mainly relies on manual and mechanical operations. Transmission line inspection must ensure that the various distances between transmission conductors and line corridors meet the requirements of safety distance regulations, especially the safety of maintenance personnel in manual operations.

[0003] Traditional safety distance measurement methods involve complex calculations such as ground filtering, DEM construction, object classification, and wire fitting. They are not accurate enough, which affects the subsequent safety zone analysis and judgment, and it is difficult to calculate the safety distance in real time. Safety distance measurement requires image recognition, analysis, and calculation technology to obtain analysis results. The analysis results will be affected by the hardware parameters such as the pixels, sensitivity, and focal length of the camera equipment and the accuracy of the software algorithm. There is a possibility of data deviation, which affects the accuracy of positioning and cannot achieve high-precision (centimeter-level) analysis results. Summary of the invention

[0004] The main purpose of the present invention is to provide a method, device, equipment and storage medium for ultra-high voltage maintenance safety management and control, aiming to solve the technical problems such as safety hazards existing in the maintenance process of existing overhead ultra-high voltage transmission lines, and the low accuracy of existing safety measurement methods and the possibility of misjudgment.

[0005] To achieve the above object, the present invention provides an ultra-high voltage maintenance safety management and control method, comprising:

[0006] Acquire a point cloud data set of an ultra-high voltage transmission channel, and resample the point cloud data set based on a preset spatial distance thinning algorithm to obtain a subset of the point cloud data set;

[0007] Performing point cloud registration on a subset of the point cloud dataset according to an ICP algorithm to obtain a target point cloud dataset;

[0008] Improving a preset classification algorithm based on a clustering algorithm, and classifying the target point cloud data set based on the improved classification algorithm to obtain classified tower body point cloud data;

[0009] Based on the classified tower body point cloud data, an operation center area is set, a rectangular box is formed with the operation center area as the center point, and the rectangular box is used as a tower point cloud safety area;

[0010] The position information of the operator is determined based on the Rtk positioning method, and whether the operator is within the safe area of ​​the tower point cloud is determined according to the position information.

[0011] In some embodiments, the step of acquiring a point cloud dataset of an ultra-high voltage power transmission channel and resampling the point cloud dataset based on a preset spatial distance thinning algorithm to obtain a subset of the point cloud dataset includes:

[0012] Obtain a point cloud data set of oblique photography of an ultra-high voltage transmission channel by a drone; wherein the point density of the point cloud data set is in 50m-2las format;

[0013] Constructing an octree structure based on the data of the point cloud data set according to a preset spatial distance thinning algorithm;

[0014] The data foot points in the octree structure are resampled based on a nearest neighbor search algorithm to obtain a subset of the point cloud data set.

[0015] In some embodiments, resampling the data foot points in the octree structure based on a nearest neighbor search algorithm to obtain a subset of the point cloud data set includes:

[0016] Determine the specified interval distance based on the nearest neighbor search algorithm;

[0017] Searching for the nearest neighbor of the data point in the octree structure in sequence according to the specified interval distance;

[0018] When the point cloud foot points of the specified interval distance are found, the foot points other than the point cloud foot points in the spherical region are removed to obtain a subset of the point cloud data set after the point cloud is thinned.

[0019] In some embodiments, the method of improving a preset classification algorithm based on a clustering algorithm and classifying the target point cloud data set based on the improved classification algorithm to obtain classified tower body point cloud data includes:

[0020] Based on the point cloud automatic classification software, the target point cloud data set is subjected to point cloud noise removal and classification to obtain a classified point cloud data set;

[0021] Dividing the classified point cloud data set and the unclassified point cloud data set into training samples and verification samples based on a preset ratio;

[0022] The PointCNN algorithm is improved based on the K-Means clustering algorithm to obtain the improved PointCNN algorithm;

[0023] The improved PointCNN algorithm is trained and verified according to the training samples and the verification samples to obtain a target point cloud classification algorithm;

[0024] The target point cloud data set is classified according to the target point cloud classification algorithm to obtain classified tower body point cloud data.

[0025] In some embodiments, the PointCNN algorithm is improved based on the K-Means clustering algorithm to obtain an improved PointCNN algorithm, including:

[0026] Performing cluster analysis on the target point cloud data set according to the K-Means clustering algorithm to obtain a clustering result for each classification;

[0027] Selecting the spatial mean of the clustering results of each classification;

[0028] The spatial mean is used as a representative point set of each layer X convolution of the PointCNN algorithm to obtain an improved PointCNN algorithm.

[0029] In some embodiments, the step of setting an operation center area based on the classified tower body point cloud data, forming a rectangular box with the operation center area as the center point, and using the rectangular box as a tower point cloud safety area includes:

[0030] According to the PCA principal component analysis method, three main directions and centroids of the classified tower body point cloud data are obtained;

[0031] Constructing a covariance matrix, and determining the eigenvalues ​​and eigenvectors of the covariance matrix; wherein the eigenvectors are the main directions;

[0032] Establishing a rectangular box of the point cloud transformed to the origin according to the main direction and the centroid;

[0033] Merge the point cloud data of the tower body located inside the rectangular box into a point set, and calculate the minimum bounding box of the point set;

[0034] The boundary range of the operation safety area is set according to the boundary of the minimum bounding box, and the tower point cloud safety area is constructed according to the boundary range.

[0035] In some embodiments, the determining the position information of the operator based on the Rtk positioning method, and judging whether the operator is inside the safe area of ​​the tower point cloud according to the position information, includes:

[0036] When the operator's helmet is turned on, the operator's point information is determined based on the Rtk positioning method;

[0037] Determine the RTK positioning point coordinates of the operator according to the point information;

[0038] Extracting normal vectors of each plane of the bounding box corresponding to the safe area of ​​the tower point cloud;

[0039] Determine the dot product of the rtk positioning point coordinates and the normal vector;

[0040] If the product of the numbers is less than or equal to zero, it is determined that the operator is inside the safe area of ​​the tower point cloud;

[0041] If the product of the quantities is greater than zero, it is determined that the operator is outside the safe area of ​​the tower point cloud.

[0042] In addition, to achieve the above-mentioned purpose, the present invention also proposes an ultra-high voltage maintenance safety control device, comprising:

[0043] A point cloud thinning module, used for acquiring a point cloud data set of an ultra-high voltage transmission channel, and resampling the point cloud data set based on a preset spatial distance thinning algorithm to obtain a subset of the point cloud data set;

[0044] A point cloud registration module, used for performing point cloud registration on a subset of the point cloud data set according to an ICP algorithm to obtain a target point cloud data set;

[0045] A point cloud classification module is used to improve a preset classification algorithm based on a clustering algorithm, and classify the target point cloud data set based on the improved classification algorithm to obtain classified tower body point cloud data;

[0046] A safety area setting module is used to set an operation center area based on the classified tower body point cloud data, form a rectangular box with the operation center area as the center point, and use the rectangular box as the tower point cloud safety area;

[0047] The positioning judgment module is used to determine the position information of the operator based on the Rtk positioning method, and judge whether the operator is inside the safe area of ​​the tower point cloud according to the position information.

[0048] In addition, to achieve the above-mentioned objectives, the present invention also proposes an ultra-high voltage maintenance safety control device, which includes: a memory, a processor, and an ultra-high voltage maintenance safety control program stored in the memory and executable on the processor, and the ultra-high voltage maintenance safety control program is configured to implement the ultra-high voltage maintenance safety control method as described above.

[0049] In addition, to achieve the above-mentioned purpose, the present invention also proposes a storage medium, which stores an ultra-high voltage maintenance safety management and control program, and the ultra-high voltage maintenance safety management and control program is used to enable the processor to implement the ultra-high voltage maintenance safety management and control method described above when executing it.

[0050] The present invention obtains a point cloud data set of an ultra-high voltage transmission channel, resamples the point cloud data set based on a preset spatial distance thinning algorithm to obtain a subset of the point cloud data set; performs point cloud registration on the subset of the point cloud data set according to an ICP algorithm to obtain a target point cloud data set; improves the preset classification algorithm based on a clustering algorithm, and classifies the target point cloud data set based on the improved classification algorithm to obtain classified tower body point cloud data; sets an operation center area based on the classified tower body point cloud data, forms a rectangular box with the operation center area as the center point, and uses the rectangular box as a tower point cloud safety area; determines the point position information of the operator based on the Rtk positioning method, and determines whether the operator is inside the tower point cloud safety area based on the point position information. In the present invention, a certain point or certain points can be quickly located in a large amount of point cloud data through a preset spatial distance thinning algorithm, which is convenient for subsequent processing. Point cloud registration is performed through the ICP algorithm, without segmenting and extracting features from the point cloud data set. Under the condition of good initial values, good algorithm convergence can be obtained. Using clustering algorithms to improve classification algorithms can avoid the problem of missing local point cloud features due to random sampling, making the selected points more objective and representative, thereby further improving the classification accuracy and framing the subsequent safe areas as backup. Based on the Rtk positioning method, centimeter-level high-precision coordinates are obtained to improve the accuracy of safe area judgment, solving the technical problems of safety hazards in the maintenance process of existing overhead ultra-high voltage transmission lines, and the low accuracy of existing safety measurement methods and the possibility of misjudgment. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 A schematic diagram of the structure of an ultra-high voltage maintenance safety control device in a hardware operating environment according to an embodiment of the present invention;

[0052] Figure 2 A schematic diagram of a process flow of an embodiment of an ultra-high voltage maintenance safety control method of the present invention;

[0053] Figure 3 Schematic diagram of different tower types related to the ultra-high voltage maintenance safety management and control method according to an embodiment of the present invention;

[0054] Figure 4 A schematic diagram of an octree storage structure of an ultra-high voltage maintenance safety management and control method according to an embodiment of the present invention;

[0055] Figure 5A schematic diagram of a spherical nearest neighbor search in an ultra-high voltage maintenance safety management and control method according to an embodiment of the present invention;

[0056] Figure 6 A comparison diagram of convolution point selection in an ultra-high voltage maintenance safety management and control method according to an embodiment of the present invention;

[0057] Figure 7 Schematic diagram of different bounding box algorithms for the ultra-high voltage maintenance safety management and control method according to an embodiment of the present invention;

[0058] Figure 8 A schematic diagram of a space plane and vectors of an ultra-high voltage maintenance safety management and control method according to an embodiment of the present invention;

[0059] Fig. 9 This is a structural block diagram of an embodiment of an ultra-high voltage maintenance safety control device of the present invention.

[0060] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0061] 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.

[0062] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative position relationship, movement status, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0063] In addition, the descriptions of "first", "second", etc. in the present invention are only used for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the technical solutions between the various embodiments can be combined with each other, but it must be based on the ability of ordinary technicians in the field to implement. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such combination of technical solutions does not exist and is not within the scope of protection required by the present invention. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0064] Reference Figure 1 , Figure 1This is a schematic diagram of the structure of the ultra-high voltage maintenance safety management and control equipment in the hardware operating environment involved in the embodiment of the present invention.

[0065] like Figure 1 As shown, the ultra-high voltage maintenance safety control equipment may include: a processor 1001, such as a central processing unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. Among them, the communication bus 1002 is used to realize the connection and communication between these components. The user interface 1003 may include a display screen (Display), an input unit such as a keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a wireless fidelity (Wireless-Fidelity, Wi-Fi) interface). The memory 1005 may be a high-speed random access memory (Random Access Memory, RAM memory) or a stable non-volatile memory (Non-Volatile Memory, NVM), such as a disk memory. The memory 1005 may also be a storage device independent of the aforementioned processor 1001.

[0066] Those skilled in the art will understand that Figure 1 The structure shown in does not constitute a limitation on the ultra-high voltage maintenance safety control equipment, and may include more or fewer components than shown in the figure, or a combination of certain components, or a different arrangement of components.

[0067] like Figure 1 As shown, the memory 1005 as a storage medium may include an operating system, a network communication module, a user interface module, and an ultra-high voltage maintenance safety management and control program.

[0068] exist Figure 1 In the ultra-high voltage maintenance safety control device shown, the network interface 1004 is mainly used for data communication with the network server; the user interface 1003 is mainly used for data interaction with the user; the processor 1001 and the memory 1005 in the ultra-high voltage maintenance safety control device of the present invention can be set in the ultra-high voltage maintenance safety control device, and the ultra-high voltage maintenance safety control device calls the ultra-high voltage maintenance safety control program stored in the memory 1005 through the processor 1001, and executes the ultra-high voltage maintenance safety control method provided by the embodiment of the present invention.

[0069] The present invention provides an ultra-high voltage maintenance safety management and control method, device, equipment and storage medium.

[0070] The embodiment of the present invention provides a method for safety management and control of ultra-high voltage maintenance, referring to Figure 2 , Figure 2 The present invention is a flowchart of an embodiment of an ultra-high voltage maintenance safety management and control method of the present invention.

[0071] like Figure 2 As shown, the ultra-high voltage maintenance safety management and control method includes:

[0072] Step S100: acquiring a point cloud data set of an ultra-high voltage transmission channel, and resampling the point cloud data set based on a preset spatial distance thinning algorithm to obtain a subset of the point cloud data set;

[0073] Step S200: performing point cloud registration on a subset of the point cloud data set according to an ICP algorithm to obtain a target point cloud data set;

[0074] Step S300: improving a preset classification algorithm based on a clustering algorithm, and classifying the target point cloud data set based on the improved classification algorithm to obtain classified tower body point cloud data;

[0075] Step S400: setting an operation center area based on the classified tower body point cloud data, forming a rectangular box with the operation center area as the center point, and using the rectangular box as a tower point cloud safety area;

[0076] Step S500: determining the location information of the operator based on the Rtk positioning method, and judging whether the operator is within the safe area of ​​the tower point cloud according to the location information.

[0077] It should be noted that the executing entity in this embodiment may be an ultra-high voltage maintenance safety control device, which may be a computer device with data processing functions, or other devices that can achieve the same or similar functions. This embodiment does not limit this. In this embodiment, a computer device is used as an example for explanation.

[0078] Specifically, obtain a point cloud dataset of an ultra-high voltage transmission channel: obtain a point cloud dataset of oblique photography of an ultra-high voltage transmission channel by a drone. Exemplarily, the point cloud dataset is an original point cloud dataset, which can be an original point cloud dataset in a 50m-2las format obtained by oblique photography of the transmission channel by a drone. The original point cloud dataset covers various types of poles and wires of different voltage levels in plateau and plain areas, including more than 1,000 poles and towers including dry-shaped towers, drum-shaped towers, wine glass-shaped towers, etc. (e.g. Figure 3 The conductor includes plain and mountain sections, totaling 500 km.

[0079] It can be understood that the core steps of the ultra-high voltage maintenance safety control method proposed in this embodiment include: Step 1, point cloud thinning: resample the original point cloud data to improve the point cloud processing efficiency without losing key features; Step 2, point cloud registration: align multiple segments of point cloud data according to feature points to ensure that there is no data offset; Step 3, point cloud classification: classify the tower body point cloud by type, and frame the subsequent safety area for backup; Step 4, safety area setting: based on the classified tower body point cloud data, set the operation center area, take this as the center point, generate a rectangular box with a length, width and height of x, y, and z respectively (the length, width and height can be customized according to user needs), determine which tower point clouds are inside the rectangular box, extract and merge them into the point set Ni, and finally calculate the minimum bounding box of the point set Ni. The boundary of this minimum bounding box is the boundary range of the operation safety area; Step 5, Rtk positioning: use GNSS real-time differential positioning to obtain centimeter-level precision point information to determine whether the operator is inside the safety area. The following will explain each step with specific examples.

[0080] In one example, the point cloud dataset is resampled based on a preset spatial distance thinning algorithm to obtain a subset of the point cloud dataset: an octree structure is constructed based on data of the point cloud dataset according to a preset spatial distance thinning algorithm; and data feet in the octree structure are resampled based on a nearest neighbor search algorithm to obtain a subset of the point cloud dataset.

[0081] Specifically, it can be understood that the preset spatial distance thinning algorithm can also be referred to as spatial thinning. The principle of this algorithm is to quickly construct an octree structure for massive point cloud data, and combine it with the nearest search method (nearest neighbor algorithm / K nearest neighbor algorithm / KNN) to determine whether to keep or remove data foot points, thereby achieving the purpose of point cloud data thinning. It is a simple and effective point cloud thinning algorithm. The spatial thinning algorithm can sample high-density, high-redundancy point clouds at a certain interval distance, so that the thinned data foot points maintain an equal interval distance and are evenly distributed, effectively removing redundant point clouds, while also retaining the original feature points of the point cloud as much as possible. The octree structure used in spatial distance thinning is to recursively divide the three-dimensional space in a regular manner. Each node of the octree represents a volume element of a cube, and each node has 8 child nodes (such as Figure 4 As shown in Figure 1.1). The initial cube bounding box of the point cloud is continuously divided into 8 equal cube units, and this splitting process is repeated recursively for each cube. The stopping condition is that there are no more points (such as 3 points) in the cube unit or when a maximum preset condition is reached. By constructing an octree structure for point cloud data, it is beneficial to data management and processing. This structure can quickly determine the points in a specific cube unit and cube triple. For example, one or some points can be quickly located in a massive amount of point cloud data, which is convenient for subsequent processing.

[0082] In one example, the data foot points in the octree structure are resampled based on a nearest neighbor search algorithm to obtain a subset of the point cloud data set, including: determining a specified interval distance based on a nearest neighbor search algorithm; searching for the nearest neighbor points of the data points in the octree structure in sequence according to the specified interval distance; when the point cloud foot points with the specified interval distance are found, the foot points other than the point cloud foot points in the spherical area are removed to obtain a subset of the point cloud data set after the point cloud is thinned.

[0083] Specifically, after the octree is constructed, the optimal tree depth (layer) is calculated according to the specified minimum distance to speed up the neighboring search efficiency. According to the specified interval distance (radius), the nearest neighbor points of the data points are searched in turn to find the point cloud foot points at a given distance, and the remaining foot points in the spherical area are removed, so as to achieve the purpose of point cloud data thinning. Spatial thinning is to uniformly sample the point cloud according to a certain minimum spatial interval (sampling interval) to simplify the original discrete redundant point cloud into a uniformly distributed point cloud, while achieving the purpose of data thinning. It should be noted that spatial thinning does not generate new point cloud data, and the point cloud obtained after thinning is a subset of the original point cloud. In addition, spatial thinning will only selectively eliminate points within the spherical search area, and outliers that exceed the sampling interval will be retained (such as Figure 5 as shown).

[0084] It can be understood that given a multidimensional space D, a vector in D is labeled as a sample point or data point. The finite set of sample points in D is called a sample set. Given a sample set E and a sample point d, the nearest neighbor of d is any sample point d′∈E that satisfies None-nearer (E, d, d′). None-nearer is defined as:

[0085]

[0086] The distance metric is the Euclidean distance .

[0087] In one example, a subset of the point cloud data set is aligned according to the ICP algorithm to obtain a target point cloud data set. Specifically, the ICP algorithm is essentially an optimal match of a least squares algorithm. The rotation and translation matrices are calculated by finding the closest point pair in the two point clouds, and the iteration is performed until a certain threshold accuracy is reached. Each iteration process can be divided into two steps: searching for the closest point pair of the two point clouds and solving the transformation matrix. The specific process is as follows:

[0088] 1) For a point in the source point cloud P , both search for the point closest to the point in the target point cloud Q , forming a corresponding point set.

[0089] 2) Solve the rotation matrix R and translation vector T between the source point cloud P and its nearest point set according to the transformation matrix calculation method.

[0090] 3) Use the rotation and translation matrix calculated in 2) to rotate the source point cloud and calculate the relative distance error between the transformed point cloud and the target point cloud:

[0091]

[0092] 4) Set the iteration stop condition, if , then stop the calculation, the current rotation matrix R and translation vector T are the final registration parameter matrix; otherwise return 1) to continue iterative calculation.

[0093] It is understandable that the ICP algorithm is simple and intuitive to calculate and has good registration accuracy, and is the most widely used method in the fine registration algorithm. However, the convergence of the algorithm to the global optimum depends largely on the distance of the initial position and the quality of the corresponding relationship during the iteration process. Therefore, this embodiment can also provide other coarse registration algorithms to find accurate corresponding point pairs to ensure that the algorithm does not fall into a local minimum.

[0094] In this embodiment, the advantages of using the ICP algorithm are as follows: there is no need to segment and extract features from the point cloud set; when the initial value is good, good algorithm convergence can be obtained. However, there are also some disadvantages: when searching for corresponding points, the amount of calculation is large and the calculation speed is slow; at the same time, the algorithm has certain requirements for the initial position of the registration point cloud, and an unreasonable initial position will cause the algorithm to fall into a local optimum; when the ICP algorithm is looking for corresponding points, the model will take the point with the closest Euclidean distance between any two point clouds as the corresponding point, and this assumption will produce a certain number of erroneous corresponding points.

[0095] In one example, a preset classification algorithm is improved based on a clustering algorithm, and the target point cloud data set is classified based on the improved classification algorithm to obtain classified tower body point cloud data: point cloud noise is removed and classified based on point cloud automatic classification software to obtain a classified point cloud data set; the classified point cloud data set and the unclassified point cloud data set are divided into training samples and verification samples based on a preset ratio; the PointCNN algorithm is improved based on the K-Means clustering algorithm to obtain an improved PointCNN algorithm; the improved PointCNN algorithm is trained and verified based on the training samples and the verification samples to obtain a target point cloud classification algorithm; the target point cloud data set is classified according to the target point cloud classification algorithm to obtain classified tower body point cloud data.

[0096] It should be noted that point cloud classification is usually achieved through supervised learning methods, including traditional supervised machine learning and deep learning. It is to mark each point or each group after segmentation with a certain class and give it a certain semantics. Therefore, point cloud classification is also called point cloud semantic segmentation or point labeling. In this embodiment, the principle of automatic point cloud classification based on deep learning: the data used is a point cloud dataset with a point density of 50m-2las format obtained by oblique photography of the transmission channel by a drone, covering various types of poles and wires of different voltage levels in plateau areas and plains. Among them, the poles include dry-shaped towers, drum-shaped towers, wine glass-shaped towers, etc. (such as Figure 3 There are more than 1,000 such points, including the ones shown in Figure 1. The wires include plain sections and mountain sections totaling 500 km. First, the noise of the point cloud is removed and classified by combining the point cloud automatic classification software TerraSolid and manual fine classification to obtain a classified point cloud data set. Then, the classified data set and the unclassified data set are divided into training samples and verification samples in a ratio of 7:3. Among them, 70% of the classified data is used to establish the model and adjust the parameters, and 30% of the unclassified data is used to verify the effect of the established model.

[0097] Specifically, feature point extraction:

[0098] Pole tower: Based on the point cloud data of the transmission line and the center position of the pole tower, the Kd-tree distance segmentation method is used to separate the pole tower point cloud from the transmission line point cloud; then the Euclidean clustering algorithm is used to remove the noise around the pole tower point cloud, and the moving surface fitting algorithm and the spatial grid area downward growth algorithm are used to remove the ground points at the bottom of the pole tower and the surrounding clutter. According to the spatial structural characteristics of the pole tower, it is divided into two parts: the tower head and the tower body. First, the key segmentation position of the tower head and the tower body is identified by analyzing the length-width ratio-height histogram of the tower; then the point cloud of the power line and insulator string around the tower head is removed based on the spatial grid area upward growth algorithm to obtain the tower head point cloud; finally, the RANSAC (Random Sample Consensus) linear fitting algorithm is used to remove the vegetation point cloud around the tower body to obtain the tower body point cloud; the tower head and tower body are integrated to obtain the pole tower point cloud.

[0099] Ground wire: The original point cloud data of the airborne LiDAR of the transmission line channel includes power line point cloud and non-power line point cloud. In this embodiment, the power line point cloud classification algorithm based on the local range elevation distribution pattern is first used to roughly extract the original point cloud to obtain a rough power line point cloud data set (including most power line point clouds, some power tower point clouds, vegetation points, and low-voltage line points, etc.). Then, based on the shape characteristics of the power line, the RANSAC (Random Sample Consensus) algorithm is used to further extract the power line point cloud, and the power line segmentation and power line modeling are performed simultaneously.

[0100] Insulators: Insulators are mainly located between the crossarm hanging point and the conductor hanging point. They mainly include suspension insulators and tension insulators, which are divided into single string and double string. Insulators obtained by airborne laser point cloud may be partially missing due to occlusion. This embodiment can use an insulator identification method based on geometric features.

[0101] Ground channel: The ground and vegetation categories can be obtained through the point cloud filtering algorithm. The channel point cloud is filtered based on the mathematical morphology filtering algorithm. The main operations of mathematical morphology filtering are corrosion, expansion, opening operation, and closing operation. The "open" operator is used to obtain the lower layer of data in the LiDAR point cloud data, that is, to obtain the ground points, which can remove isolated targets smaller than the structural elements.

[0102] Buildings: The overall process of extracting buildings from lidar point clouds is as follows: first, the laser point cloud of the extracted ground points is divided into grids, and the number of point clouds, maximum value, minimum value and other attributes in each grid are counted; then, candidate building grids are found based on the height difference between the building and the ground; secondly, since there are a large number of non-building point clouds, such as vegetation, in the candidate building grids, the grids are divided again, and the grid spacing is much smaller than the grid spacing of the first division, so as to realize the extraction of building point clouds; finally, the connected domain analysis is performed on the extracted building point clouds to realize accurate building point cloud extraction.

[0103] In one example, the PointCNN algorithm is improved based on the K-Means clustering algorithm to obtain an improved PointCNN algorithm: cluster analysis is performed on the target point cloud data set according to the K-Means clustering algorithm to obtain a clustering result for each classification; the spatial mean of the clustering result of each classification is selected; and the spatial mean is used as a representative point set of each layer of X convolution of the PointCNN algorithm to obtain an improved PointCNN algorithm.

[0104] It should be noted that the improved PointCNN algorithm: The algorithm idea of ​​PointCNN is to perform matrix transformation based on a core module X-Conv to give the point cloud a potential order. First, the domain point is subtracted from the center point to form a relative position relationship, and a local feature F' is learned using the relative MLP (multi-layer perceptron). The local feature is added to the input global feature to obtain a new feature, and then the relative position relationship is learned using MLP to obtain a K×K matrix. This matrix is ​​the X transformation, which can ensure the disorder of the point cloud, and then it is convolved with the convolution kernel K. The feature Fp finally obtained represents the feature of point p and completes the feature mapping of point p.

[0105] This embodiment takes the key ground feature identification of the power transmission channel as an example, and represents the point cloud data as a point set F containing spatial features. 1.

[0106]

[0107] A set of points The feature set corresponding to each point set (where C1 represents the initial feature channel), the input F is transformed into 1 Convert to Among them, {p 2 ,i} is from {p 1 ,i}. After the X convolution transformation, F 2 The number of feature points is reduced while the corresponding feature channels are increased.

[0108] Specifically, the original point cloud forms higher-dimensional point cloud features through the X convolution layer. The higher the level, the richer the features of these point cloud feature layers. Through information extraction, fewer points with richer features are obtained. This also means the importance of these representative points. How to select reasonable points is a major improvement space for the PointCNN algorithm. Compared with the random sampling point selection method of the conventional PointCNN algorithm, cluster analysis is used in this embodiment to select points with similar features, so that the spatial mean of the clustering results obtained for each classification is selected as the representative point set of each layer of X convolution.

[0109] It is understandable that the purpose of the K-Means clustering algorithm is to divide the point cloud set into non-overlapping point cloud clusters according to specific standards, so that the similarity of point cloud features in the same cluster is as large as possible, and the similarity of point cloud features in different clusters is as small as possible, and the point cloud after clustering analysis is more representative. In the process of PointCNN multi-layer convolution, it is necessary to determine the number of points participating in the convolution of each layer, and the number of clustering categories also needs to be determined in advance in the K-Means clustering process. This embodiment uses the clustering algorithm to improve the PointCNN algorithm to avoid the problem of missing local point cloud features due to random sampling, making the selected points more objective and representative. In this embodiment, Figure 6 The convolution point selection comparison shown in Figure 2 is shown in Figure 2. Figure 6 (a) and (b) shown are the random selection methods of the original PointCNN algorithm; Figure 6 (c) and (d) shown are selection methods based on cluster analysis.

[0110] The core purpose of improving the PointCNN point cloud deep learning algorithm based on the K-Means clustering algorithm is to obtain the point cloud through clustering. 1 , i point cloud subsets are selected from p 2 ,i is used for convolution, the specific steps of this process are as follows:

[0111] Calculate the unordered point cloud subset p 1 ,i relative to p i The relative coordinate tensor L of the image is obtained; the K-Means clustering algorithm is used to perform k clustering operations on the coordinate tensor L, where k is the number of points involved in the convolution layer and is set according to the point deep learning parameters; the position means of the r categories are calculated; the position means of the r categories are connected to obtain the tensor P, and the tensor P is used in the convolution layer, where the convolution layer involves multiple convolution processes. The clustering algorithm is used multiple times to select recursive convolution points to complete the deep learning feature extraction process.

[0112] Specifically, the evaluation indicators of the PointCNN improved algorithm are: based on the real classified data, the accuracy of the model prediction results is evaluated. The main indicators involved in the evaluation are as follows:

[0113] 1. Precision and recall

[0114] The accuracy rate refers to the ratio of the number of correctly classified point clouds to all point cloud data sets involved in data processing based on the LiDAR point cloud data classification algorithm. Generally speaking, the larger the ratio, the better the accuracy of the classifier. The calculation formula of the accuracy rate A is:

[0115]

[0116] Among them, the expressions of TN, TP, FP, and FN are as follows: TP means correctly predicted as a positive example; TN means correctly predicted as a negative example; FP means incorrectly predicted as a positive example; and FN means incorrectly predicted as a negative example. TP+FP+TN+FN=total number of samples.

[0117] The recall rate is the ratio of the number of correctly classified positive samples to the number of all true positive samples. Generally, the higher the recall rate, the stronger the classification ability of the model. The calculation formula of the recall rate R is:

[0118]

[0119] Among them, the expressions of TP and FN are as follows: TP means correctly predicted as a positive example; FN means incorrectly predicted as a negative example.

[0120] 2. Mean Square Error Loss Function

[0121] The loss function is a calculation function used to measure the difference between the predicted value and the true value of the model. It is a non-negative real-valued function. The smaller the loss function, the better the robustness of the model. This embodiment uses the mean square error loss function to calculate the distance between the predicted point cloud and the true point cloud based on the Euclidean distance. The loss function MSE calculation formula is as follows:

[0122]

[0123] in, Represents the spatial information of the real point, Represents the spatial coordinate information of the prediction point.

[0124] In one example, an operation center area is set based on the classified tower body point cloud data, a rectangular box is formed with the operation center area as the center point, and the rectangular box is used as the tower point cloud safety area: the three main directions and the center of mass of the classified tower body point cloud data are obtained according to the PCA principal component analysis method; a covariance matrix is ​​constructed, and the eigenvalues ​​and eigenvectors of the covariance matrix are determined; wherein the eigenvector is the main direction; a rectangular box of the point cloud transformed to the origin is established according to the main direction and the center of mass; the tower body point cloud data located inside the rectangular box is merged into a point set, and the minimum bounding box of the point set is calculated; the boundary range of the operation safety area is set according to the boundary of the minimum bounding box, and the tower point cloud safety area is constructed according to the boundary range.

[0125] Specifically, the safe area setting: the bounding box, also called the minimum circumscribed rectangle, is an algorithm for solving the optimal bounding space of a discrete point set. The basic idea is to use a slightly larger and simpler geometric body (called a bounding box) to approximately replace complex geometric objects. Figure 7 As shown, common bounding box algorithms include AABB bounding box, bounding sphere, directional bounding box OBB and fixed direction convex hull FDH.

[0126] The calculation process of the minimum bounding box is roughly as follows:

[0127] 1. Use PCA principal component analysis to obtain the three main directions of the point cloud, obtain the centroid, calculate the covariance, obtain the covariance matrix, and find the eigenvalues ​​and feature vectors of the covariance matrix. The eigenvector is the main direction.

[0128] 2. Using the main direction and center of mass obtained in step 1, transform the input point cloud to the origin, and restore the main direction to the coordinate system direction, and establish the bounding box of the point cloud transformed to the origin.

[0129] 3. Set the main direction and bounding box for the input point cloud, which is achieved by the inverse transformation of the input point cloud to the origin point cloud.

[0130] It should be noted that PCA (Principal Component Analysis): Principal component analysis can be used to calculate the coordinate space of a data set composed of multiple variables. The multiple variables of the data set can be divided into multiple unrelated components. The x, y, z coordinate values ​​in the vertex position array are such a data set (point cloud). The most different directions in the data set represent the basic principal components of the principal components.

[0131] Set each point to , the point cloud is regarded as P 1 , P 2 , P 3 , ...P N The data set consists of N points. First, the position mean m is calculated using the following formula:

[0132]

[0133] Then construct a covariance matrix C:

[0134]

[0135] The covariance matrix is ​​a symmetric matrix consisting of the following six elements:

[0136]

[0137] Specifically, the covariance matrix represents the relationship between the x, y, and z coordinate values. If the three coordinate values ​​are independent of each other, the values ​​of their elements in the covariance matrix are 0. In order to evenly distribute all points along the coordinate axis, the covariance matrix needs to be converted into a diagonal matrix. You can use the content in linear algebra:

[0138] (1) Solving for the eigenvalue λ 1 ,λ 2 ,λ 3 ;

[0139] (2) Substitute the solved eigenvalue back into (A-λE) = 0 to solve for the corresponding eigenvector ε 1 , ε 2 , ε 3 ;

[0140] (3) After orthogonal normalization of each eigenvector, the three orthogonal normalized vectors form an orthogonal matrix:

[0141]

[0142] Same QCQ T The result should be a diagonal matrix C′:

[0143]

[0144] by , , As the direction corresponding to the natural axis of the object, the maximum and minimum positions along the three directions of X, Y, and Z from the vertex are calculated below. Based on these maximum and minimum values, the six planes of the bounding box can be easily constructed.

[0145] To determine the maximum and minimum ranges, we can calculate the position coordinates P of each vertex. i It is done by the inner product with the unit vector. The 6 planes of the bounding box are:

[0146]

[0147] The size of the bounding box is the difference between the maximum and minimum values ​​of the corresponding inner products in the three directions of X, Y, and Z. The center O of the bounding box is the intersection of the three planes in the three pairs of reverse planes. Let a, b, and c be the average of the maximum and minimum values ​​on X, Y, and Z respectively:

[0148]

[0149] Then the coordinates of the center point are O=aX+bY+cZ.

[0150] In one example, the position information of the operator is determined based on the Rtk positioning method, and whether the operator is inside the tower point cloud safety area is judged according to the position information: when the operator's helmet is turned on, the position information of the operator is determined based on the Rtk positioning method; the rtk positioning point coordinates of the operator are determined according to the position information; the normal vectors of each plane of the bounding box corresponding to the tower point cloud safety area are extracted; the dot product of the rtk positioning point coordinates and the normal vector is determined; if the dot products are less than or equal to zero, it is judged that the operator is inside the tower point cloud safety area; if the dot products are greater than zero, it is judged that the operator is outside the tower point cloud safety area.

[0151] It should be noted that Rtk is actually a measurement technology called "carrier phase differential technology", which is a differential method for real-time processing of carrier phase observations of two measurement stations. The carrier phase collected by the base station is sent to the user receiver to calculate the difference in coordinates, thereby obtaining high-precision coordinates.

[0152] Specifically, the satellite data found by Rtk will contain the satellite's orbital information (position) and time information. The time taken by the known satellite signal to propagate to Rtk is multiplied by the propagation speed (the default speed of light) to get the distance between the satellite and Rtk. The formula is established through the distance and the known coordinates (X, Y, Z) of each satellite to calculate the current coordinates (X, Y, Z) of Rtk. The more satellites found and the more known quantities, the higher the positioning accuracy of Rtk.

[0153] Specifically, differential source, base station: Since satellite signals are affected by factors such as the ionosphere, troposphere, and multipath effects during propagation, the positioning accuracy of a single Rtk can only reach about 5 meters, which is the conventional single-point solution. To improve the positioning accuracy of Rtk, it is necessary to introduce a differential source and use relative positioning to eliminate common unknowns. That is, one Rtk acts as a base station, sends the received satellite signals and other information to the mobile station in real time, and performs differential operations with the same satellite signals and other information received by the mobile station to achieve centimeter-level positioning accuracy, which is the so-called fixed solution. If the number of satellites searched in the environment where Rtk is located is small or the quality of the satellite search is poor (such as under trees, buildings, etc.), it will affect the fixed effect and positioning accuracy of Rtk to a certain extent. Using the principle of relative positioning to achieve centimeter-level positioning accuracy actually assumes that the errors received by the base station and the mobile station are the same. The farther the distance between the two, the worse the accuracy.

[0154] Specifically, radio mode: the base station transmits differential information to the mobile station in the form of electromagnetic waves through the radio. The base station and the mobile station must ensure that the radio channel, protocol, and frequency are consistent before communication can be achieved. Network mode / CORS base station: the base station transmits differential information to the mobile station through the network mode. The mobile station needs to enter the base station's IP, port, access point, account, password and other information, and the mobile station needs to be able to access the Internet.

[0155] It is understandable that the above-mentioned radio mode and network mode have their own advantages and disadvantages: the network mode has a wide operating distance. Now there are CORS base stations covering the whole country, and basically one account can be used throughout the country. Since the radio mode is transmitted through electromagnetic waves, if it encounters high buildings and trees during the transmission process, the signal will be attenuated, affecting the operating distance. The general operating distance is 2 to 15 kilometers. In addition, the network mode is affected by the operator's network signal. In mountainous areas where the network is underdeveloped, the use of the network mode may be affected. In this case, the radio mode is generally used.

[0156] It should be noted that this embodiment is described by taking an actual operation process as an example:

[0157] In this embodiment, the Rtk hardware can be integrated into the intelligent integrated helmet. Obtain the scope of the survey area and the control points of the survey area: including the coordinates of the control points, elevation, the coordinate system used and the central meridian. If the parameters of the survey area already exist, the parameters of the survey area can be used directly. Determine the base station to be used according to the size and characteristics of the working area: radio 1+1 or single station + network account. Use the helmet to start the machine, and the live working platform can obtain the Rtk point information and reach a fixed solution.

[0158] Specifically, working conditions: After the test personnel have debugged the equipment on the ground, an operator will carry the equipment to the high-voltage tower for testing (try to ensure that the Rtk antenna is always in a vertical upward state when climbing). In actual applications, the operator's movement path can be to move vertically upward along the tower and then move horizontally. Test results: The intelligent integrated safety helmet is well covered when worn on the head. The Rtk fixed solution ratio is 31.8% from the start of the operation to entering the equipotential zone, and the Rtk fixed solution ratio is 25.4% after entering the equipotential zone. According to the test results, the fixed solution ratio before and after entering the equipotential zone is not much different. Therefore, in an ultra-high voltage and strong electric field environment, the above method of this embodiment has a greater probability of obtaining accurate Rtk point information and reaching a fixed solution, and thereby judging whether the operator is inside the safe area of ​​the tower point cloud, thereby further improving the accuracy of the safe area judgment.

[0159] Specifically, the relative position relationship between the Rtk point and the bounding box is determined as follows:

[0160] Two vectors and The cosine of the angle is:

[0161]

[0162] Among them, θ is the angle between the two vectors; For vector and The number product of Represents the magnitude of a vector.

[0163] refer to Figure 8 , α is a plane in space, and the normal vector of the plane is specified to point upward, is the plane normal vector, A is any point on the α plane, P is any point in space, and the vector from point A to point P and the normal vector The scalar product can be obtained:

[0164]

[0165] For any convex polyhedron in space, if the normal vector of each surface of the polyhedron is stipulated to face outward, then the necessary and sufficient condition for any point P in space to be inside the polyhedron is that the product of the vector from any point on each surface to point P and the vector of the surface normal is less than or equal to 0 (if all are less than 0, it means the point is inside the polyhedron, otherwise, it means the point is on the polyhedron).

[0166] Specifically, take all the planes of the tower point cloud safety zone bounding box, and calculate the scalar product of the plane normal vectors of the bounding box with the coordinates of the Rtk positioning point. If they are all less than or equal to 0, it can be judged that the positioning point is inside the bounding box, that is, it is in the safe area, which means that the operator is judged to be inside the tower point cloud safety area. Otherwise, it is outside the safe area, which means that the operator is judged to be outside the tower point cloud safety area.

[0167] In this embodiment, a point cloud data set of an ultra-high voltage transmission channel is obtained, and the point cloud data set is resampled based on a preset spatial distance thinning algorithm to obtain a subset of the point cloud data set; the subset of the point cloud data set is point cloud registered based on the ICP algorithm to obtain a target point cloud data set; the preset classification algorithm is improved based on a clustering algorithm, and the target point cloud data set is classified based on the improved classification algorithm to obtain classified tower body point cloud data; the operation center area is set based on the classified tower body point cloud data, a rectangular box is formed with the operation center area as the center point, and the rectangular box is used as the tower point cloud safety area; the point position information of the operator is determined based on the Rtk positioning method, and whether the operator is inside the tower point cloud safety area is determined based on the point position information. In this embodiment, a certain point or certain points can be quickly located in a large amount of point cloud data through a preset spatial distance thinning algorithm, which is convenient for subsequent processing. Point cloud registration is performed through the ICP algorithm, and there is no need to segment and extract features from the point cloud data set. Under the condition of good initial values, good algorithm convergence can be obtained. Using clustering algorithms to improve classification algorithms can avoid the problem of missing local point cloud features due to random sampling, making the selected points more objective and representative, thereby further improving the classification accuracy and framing the subsequent safe areas as backup. Based on the Rtk positioning method, centimeter-level high-precision coordinates are obtained to improve the accuracy of safe area judgment, solving the technical problems of safety hazards in the maintenance process of existing overhead ultra-high voltage transmission lines, and the low accuracy of existing safety measurement methods and the possibility of misjudgment.

[0168] In addition, an embodiment of the present invention also proposes a storage medium, on which an ultra-high voltage maintenance safety management and control program is stored. When the ultra-high voltage maintenance safety management and control program is executed by a processor, the steps of the ultra-high voltage maintenance safety management and control method described above are implemented.

[0169] Reference Fig. 9 , Fig. 9 This is a structural block diagram of an embodiment of an ultra-high voltage maintenance safety control device of the present invention.

[0170] like Fig. 9 As shown, the ultra-high voltage maintenance safety control device includes:

[0171] A point cloud thinning module 10 is used to obtain a point cloud data set of the ultra-high voltage transmission channel, and resample the point cloud data set based on a preset spatial distance thinning algorithm to obtain a subset of the point cloud data set;

[0172] A point cloud registration module 20, configured to perform point cloud registration on a subset of the point cloud data set according to an ICP algorithm to obtain a target point cloud data set;

[0173] The point cloud classification module 30 is used to improve the preset classification algorithm based on the clustering algorithm, and classify the target point cloud data set based on the improved classification algorithm to obtain the classified tower body point cloud data;

[0174] A safety area setting module 40 is used to set an operation center area based on the classified tower body point cloud data, form a rectangular box with the operation center area as the center point, and use the rectangular box as the tower point cloud safety area;

[0175] The positioning judgment module 50 is used to determine the position information of the operator based on the Rtk positioning method, and judge whether the operator is inside the safe area of ​​the tower point cloud according to the position information.

[0176] It should be noted that the technical details that are not described in detail in the embodiment of the ultra-high voltage maintenance safety control device can be referred to the ultra-high voltage maintenance safety control method as described above provided in any embodiment of the present invention, and will not be repeated here.

[0177] It should be understood that the above is only an example and does not constitute any limitation on the technical solution of the present invention. In specific applications, technicians in this field can make settings as needed, and the present invention does not limit this.

[0178] It should be noted that the workflow described above is merely illustrative and does not limit the scope of protection of the present invention. In practical applications, technicians in this field can select part or all of them according to actual needs to achieve the purpose of the present embodiment, and no limitation is made here.

[0179] In addition, it should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or system. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or system including the element.

[0180] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0181] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as a read-only memory (ROM) / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0182] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for safety control of ultra-high voltage maintenance, characterized in that: The ultra-high voltage maintenance safety management and control method comprises: Acquire a point cloud data set of an ultra-high voltage transmission channel, and resample the point cloud data set based on a preset spatial distance thinning algorithm to obtain a subset of the point cloud data set; Performing point cloud registration on a subset of the point cloud dataset according to an ICP algorithm to obtain a target point cloud dataset; Improving a preset classification algorithm based on a clustering algorithm, and classifying the target point cloud data set based on the improved classification algorithm to obtain classified tower body point cloud data; Based on the classified tower body point cloud data, an operation center area is set, a rectangular box is formed with the operation center area as the center point, and the rectangular box is used as a tower point cloud safety area; Determine the location information of the operator based on the Rtk positioning method, and judge whether the operator is inside the safe area of ​​the tower point cloud according to the location information; Among them, the preset classification algorithm is improved based on the clustering algorithm, including: performing cluster analysis on the target point cloud data set according to the K-Means clustering algorithm to obtain the clustering result of each classification; selecting the spatial mean of the clustering result of each classification; using the spatial mean as the representative point set of each layer of X convolution of the PointCNN algorithm to obtain the improved PointCNN algorithm; The method comprises the following steps: setting an operation center area based on the classified pole tower body point cloud data, forming a rectangular box with the operation center area as the center point, and using the rectangular box as the pole tower point cloud safety area, including: obtaining three main directions and the centroid of the classified pole tower body point cloud data according to the PCA principal component analysis method; constructing a covariance matrix, and determining the eigenvalues ​​and eigenvectors of the covariance matrix; wherein the eigenvector is the main direction; establishing a rectangular box of the point cloud transformed to the origin according to the main direction and the centroid; merging the pole tower body point cloud data located inside the rectangular box into a point set, and calculating the minimum bounding box of the point set; setting a boundary range of the operation safety area according to the boundary of the minimum bounding box, and constructing the pole tower point cloud safety area according to the boundary range.

2. The ultra-high voltage maintenance safety management and control method according to claim 1, characterized in that: The step of acquiring a point cloud data set of the ultra-high voltage transmission channel and resampling the point cloud data set based on a preset spatial distance thinning algorithm to obtain a subset of the point cloud data set includes: Obtain a point cloud data set of oblique photography of an ultra-high voltage transmission channel by a drone; wherein the point density of the point cloud data set is in 50m-2las format; Constructing an octree structure based on the data of the point cloud data set according to a preset spatial distance thinning algorithm; The data foot points in the octree structure are resampled based on a nearest neighbor search algorithm to obtain a subset of the point cloud data set.

3. The ultra-high voltage maintenance safety management and control method according to claim 2, characterized in that: The resampling of the data foot points in the octree structure based on the nearest neighbor search algorithm to obtain a subset of the point cloud data set includes: Determine the specified interval distance based on the nearest neighbor search algorithm; Searching for the nearest neighbor of the data point in the octree structure in sequence according to the specified interval distance; When the point cloud foot point, i.e., the nearest neighbor point, with the specified interval distance is found, the foot points other than the point cloud foot point in the spherical area are removed to obtain a subset of the point cloud data set after point cloud thinning; wherein the spherical area is a sphere with the data point as the center and the specified interval distance as the radius.

4. The ultra-high voltage maintenance safety management and control method according to claim 1, characterized in that: The method of improving the preset classification algorithm based on the clustering algorithm and classifying the target point cloud data set based on the improved classification algorithm to obtain the classified tower body point cloud data includes: Based on the point cloud automatic classification software, the target point cloud data set is subjected to point cloud noise removal and classification to obtain a classified point cloud data set; Dividing the classified point cloud data set and the unclassified point cloud data set into training samples and verification samples based on a preset ratio; The PointCNN algorithm is improved based on the K-Means clustering algorithm to obtain the improved PointCNN algorithm; The improved PointCNN algorithm is trained and verified according to the training samples and the verification samples to obtain a target point cloud classification algorithm; The target point cloud data set is classified according to the target point cloud classification algorithm to obtain classified tower body point cloud data.

5. The ultra-high voltage maintenance safety management and control method according to claim 1, characterized in that: The determining of the position information of the operator based on the Rtk positioning method, and judging whether the operator is within the safety area of ​​the tower point cloud according to the position information, includes: When the operator's helmet is turned on, the operator's point information is determined based on the Rtk positioning method; Determine the RTK positioning point coordinates of the operator according to the point information; Extracting normal vectors of each plane of the bounding box corresponding to the safe area of ​​the tower point cloud; Determine the dot product of the rtk positioning point coordinates and the normal vector; If the product of the numbers is less than or equal to zero, it is determined that the operator is inside the safe area of ​​the tower point cloud; If the product of the quantities is greater than zero, it is determined that the operator is outside the safe area of ​​the tower point cloud.

6. An ultra-high voltage maintenance safety control device, characterized in that: The ultra-high voltage maintenance safety control device comprises: A point cloud thinning module, used for acquiring a point cloud data set of an ultra-high voltage transmission channel, and resampling the point cloud data set based on a preset spatial distance thinning algorithm to obtain a subset of the point cloud data set; A point cloud registration module, used for performing point cloud registration on a subset of the point cloud data set according to an ICP algorithm to obtain a target point cloud data set; A point cloud classification module is used to improve a preset classification algorithm based on a clustering algorithm, and classify the target point cloud data set based on the improved classification algorithm to obtain classified tower body point cloud data; A safety area setting module is used to set an operation center area based on the classified tower body point cloud data, form a rectangular box with the operation center area as the center point, and use the rectangular box as the tower point cloud safety area; A positioning judgment module is used to determine the position information of the operator based on the Rtk positioning method, and judge whether the operator is inside the safe area of ​​the tower point cloud according to the position information; Among them, the preset classification algorithm is improved based on the clustering algorithm, including: performing cluster analysis on the target point cloud data set according to the K-Means clustering algorithm to obtain the clustering result of each classification; selecting the spatial mean of the clustering result of each classification; using the spatial mean as the representative point set of each layer of X convolution of the PointCNN algorithm to obtain the improved PointCNN algorithm; The method comprises the following steps: setting an operation center area based on the classified pole tower body point cloud data, forming a rectangular box with the operation center area as the center point, and using the rectangular box as the pole tower point cloud safety area, including: obtaining three main directions and the centroid of the classified pole tower body point cloud data according to the PCA principal component analysis method; constructing a covariance matrix, and determining the eigenvalues ​​and eigenvectors of the covariance matrix; wherein the eigenvector is the main direction; establishing a rectangular box of the point cloud transformed to the origin according to the main direction and the centroid; merging the pole tower body point cloud data located inside the rectangular box into a point set, and calculating the minimum bounding box of the point set; setting a boundary range of the operation safety area according to the boundary of the minimum bounding box, and constructing the pole tower point cloud safety area according to the boundary range.

7. An ultra-high voltage maintenance safety control device, characterized in that: The ultra-high voltage maintenance safety control device includes: a memory, a processor, and an ultra-high voltage maintenance safety control program stored in the memory and executable on the processor, wherein the ultra-high voltage maintenance safety control program is configured to implement the ultra-high voltage maintenance safety control method as described in any one of claims 1 to 5.

8. A storage medium, characterized in that: The storage medium stores an ultra-high voltage maintenance safety management and control program, and the ultra-high voltage maintenance safety management and control program is used to enable the processor to implement the ultra-high voltage maintenance safety management and control method as described in any one of claims 1 to 5 when executing.

Citation Information

Patent Citations

  • Safe distance calculation method for electrified operation of transformer substation

    CN107704879A

  • RTK positioning-based power transmission maintenance safety early warning method and system, and terminal

    CN113990034A

  • Method for realizing on-line monitoring and intelligent inspection of power transmission line based on three-dimensional simulation technology

    CN114637026A

  • Feature vector-based feature extraction method for laser point cloud data of power transmission line

    CN115356705A

  • Extrinsic parameter calibration method and apparatus, computing device and computer storage medium

    WO2021031157A1

Cited By

  • An overhead line engineering open operation area fusion positioning management and control method and system

    CN122983982A