A single-line laser radar-based distribution line identification and positioning system

Abstract

The application provides a kind of network distribution wire identification positioning system based on single line laser radar, single line laser radar (21) is fixedly installed at the axis end of six-degree-of-freedom mechanical arm wrist joint two (34), and the scanning plane normal vector is parallel to the axis of wrist joint two (32). The scanning dimension and position of the laser radar are dynamically adjusted by the mechanical arm, and the identification and positioning of the network distribution wire component are completed by using component point cloud segmentation and operation point identification and positioning based on geometric constraint-frequency statistics-semantic environment-operation standard. The operation point of the insulator and the wire can be accurately identified and precisely positioned. The system has the advantages of high efficiency and economy, and can adapt to the three-dimensional scene of pole and wire intersection, providing a full-automatic solution for live working scene of network distribution, and promoting the intelligent transformation of power operation and maintenance.

Description

Technical Field

[0001] This invention belongs to the technical fields of live-line working on power distribution networks and 3D visual perception, and specifically relates to a power distribution network conductor identification and positioning system based on a single-line lidar. Background Technology

[0002] With the deepening of power distribution network construction, both power supply companies and electricity users have placed higher demands on the reliability of power supply. High-altitude live-line work requires high physical fitness from workers and carries hidden dangers of electric shock and falls from heights. To address the safety issues associated with live-line work, the replacement of humans with intelligent robots for relatively complex tasks such as power connection is becoming an inevitable trend. Currently, live-line work sites are distributed across different cities and suburbs, with complex distribution and diverse environments, and are affected by ambient light intensity. This presents significant challenges to the identification and positioning of thin, cylindrical conductors with weak textures and to the realization of autonomous perception functions for live-line work robots.

[0003] Existing methods for identifying and locating power transmission lines during live-line work in power distribution networks mainly fall into two categories: visual image recognition and 3D point cloud recognition. In the field of visual image recognition, patent CN118015032A discloses a high-precision identification and location method, device, and storage medium for power transmission lines. It utilizes color images and depth maps, fitting the location of the power transmission line based on the slope of the normal curves of its left and right edges. However, the camera is sensitive to changes in outdoor light intensity, making target identification and location easily affected by lighting conditions, which significantly limits the applicability of this method. Regarding 3D point cloud recognition, patent CN117036826A discloses a method for identifying and locating power transmission lines during live-line work in power distribution networks. It uses multiple 3D LiDARs to collect more comprehensive data, providing strong feature description capabilities for power transmission lines. Combined with an SVM classifier and correlation analysis, the identification accuracy is quite high. However, it is costly, complex, and has stringent hardware requirements. Furthermore, it can only identify power transmission lines, limiting its application scenarios. Patent CN114627374A discloses a point cloud acquisition system and insulator identification and positioning method based on lidar and a gimbal. The lidar is vertically placed in the center of the gimbal to increase the scanning range. Patent CN119340847A discloses a lightweight live-line working robot system and operating method based on an insulated bucket truck. The radar installation position is fixed. Both of these solutions have blind spots in the scanning space and cannot flexibly change the scanning position. Patent CN118093706A proposes a live-line working robot, system, and operating method for power distribution networks. Based on the characteristics of different targets (leads, lead ends, lines), it matches multiple identification methods such as depth camera point cloud acquisition, binocular camera video image acquisition, and lidar scanning, which improves the identification accuracy of targets in complex environments to a certain extent. However, the equipment cost is high and the hardware dependence is high. Patent CN116883606A discloses a method and system for constructing a three-dimensional model. It decomposes clustered point cloud clusters into different components by geometric model fitting, analyzes and determines the connection topology between components, uses a model library to reconstruct the local three-dimensional structure of small blocks, and then combines the small blocks to form a complete three-dimensional model. Although it mentions the modeling of insulators and conductors, its method relies on a general model library and geometric fitting. Furthermore, the use of the model library requires that the point cloud has obvious geometric appearance features for fitting. It does not mention the identification and localization of irregular or low-density point clouds and lacks a refined identification and localization method specific to power scenarios. Summary of the Invention

[0004] In view of this, and in view of the defects and deficiencies of the existing technology, the purpose of this invention is to provide a distribution network conductor identification and positioning system based on single-line lidar, focusing on the refined identification and positioning of power scenarios, embedding industry standards into the algorithm, and realizing a closed loop from reconstruction to operation guidance, so as to effectively solve the problem of high-precision and automated identification and positioning of conductors and distribution network components in complex distribution environments and under changing conditions. This effectively improves hardware utilization and operation efficiency, and reduces equipment switching and operation steps.

[0005] This invention aims to improve the automated operation and maintenance level of power facilities through high-precision 3D perception technology. A single-line LiDAR is installed at the wrist joint 2 of a six-DOF collaborative robotic arm. By dynamically adjusting the scanning dimension and position of the LiDAR through the robotic arm, the field of view limitations of traditional fixed scanning are broken, significantly expanding the scanning range. Simultaneously, the robotic arm can collaboratively perform tool-grabbing tasks while performing scanning and modeling tasks, greatly improving work efficiency and accuracy. The system employs depth-intensity dual-modal filtering, 3D reconstruction, and point cloud processing methods, combined with KD tree spatial indexing, improved voxel grid downsampling, DBSCAN and PCA collaborative high-voltage line component point cloud segmentation technology, and work point identification and positioning based on geometric constraints, frequency statistics, semantic environment, and work standards. This accurately identifies and precisely locates work points for insulators and conductors. The system combines high efficiency and economy, adapting to three-dimensional scenarios with intersecting poles and conductors, providing a fully automated solution for live-line work in distribution networks, and promoting the intelligent transformation of power operation and maintenance.

[0006] The specific technical solution adopted by this invention to solve its technical problem is as follows:

[0007] A distribution network conductor identification and positioning system based on single-line lidar: The single-line lidar (21) is fixedly installed at the end of the axis of the wrist joint 2 (34) of the six-degree-of-freedom robotic arm, and the scanning plane normal vector is parallel to the axis of the wrist joint 2 (34); the scanning dimension and position of the lidar are dynamically adjusted by the robotic arm, and the identification and positioning of the distribution network conductor components are completed by component point cloud segmentation and operation point identification and positioning based on geometric constraints-frequency statistics-semantic environment-operation standards.

[0008] Furthermore, the rigid bracket (22) is connected to the single-line laser radar (21). The upper cover (24) and lower cover (23) of the symmetrical structure of the rigid bracket (22) wrap the wrist joint (34), fixing the wrist joint (34) to the rotation position of the wrist joint (33) and the wrist joint (35) in a coaxial manner to ensure that the position and posture of the single-line laser radar (21) are fixed. The first four joints of the robotic arm are used to adjust the position and posture of the single-line laser radar, including the base (31) for changing the position of the single-line laser radar (21), the three joints of the shoulder (32) and the elbow (36), and the wrist joint (33) for adjusting the angle of the single-line laser radar (21).

[0009] Furthermore, the robotic arm in the retracted state performs an initialization motion, and after the motion is completed, it grabs the end-effector tool. During the process, the attitude of the single-line laser radar is adjusted in coordination, and point cloud scanning and modeling are performed simultaneously. When the initialization is completed, the point cloud data acquisition is completed. The method of controlling the attitude of the single-line laser radar is to adjust the wrist joint (34) to keep the single-line laser radar (21) at an upward scanning angle during the process of the robotic arm end position moving to the initialization end position. When the robotic arm moves to the tool library to grab the tool, the algorithm for identifying and locating high-voltage lines and insulators is executed simultaneously to complete the pose estimation of the target point.

[0010] Furthermore, the process of identifying and locating power distribution network conductor components by dynamically adjusting the scanning dimension and position of the LiDAR using a robotic arm, and employing component point cloud segmentation and operation point identification and positioning based on geometric constraints, frequency statistics, semantic environment, and operation standards includes:

[0011] By controlling the attitude of the LiDAR and enabling LiDAR scanning during the movement of the robotic arm to the tool gripping point, the robot joint angles and the point cloud data of the power distribution environment collected by the LiDAR are acquired simultaneously.

[0012] Point cloud data is subjected to point cloud tailing filtering and noise reduction based on depth and intensity dual-modality;

[0013] Perform 3D reconstruction of point clouds, including pose solving and point cloud coordinate transformation;

[0014] Processing 3D point cloud data to identify target components;

[0015] The 3D point cloud instance is fitted and segmented, and the pose of the task point is determined based on the semantic environment and task standards.

[0016] Furthermore, the point cloud trailing filter combines the changes in depth and reflection intensity between each point and its neighboring points. When the depth change of neighboring point clouds exceeds a preset threshold, it is identified as a trailing point cloud and removed. The remaining point clouds are then optimized by removing abnormal values ​​of reflection intensity frequency of neighboring points.

[0017] Furthermore, the pose solution is based on the structural parameters of the robotic arm, the height parameters of the emitted laser provided by the radar, and the structural parameters of the radar mounting bracket to obtain the DH parameters of the vision system robotic arm. After 4-link kinematic modeling, the point cloud coordinate transformation matrix is ​​obtained, and the point cloud data is transformed to the robotic arm coordinate system to complete the 3D point cloud stitching.

[0018] Furthermore, the processing of the 3D point cloud data to identify the target component specifically includes the following steps:

[0019] Perform pass-through filtering on the 3D point cloud data to filter out invalid point cloud data;

[0020] Based on the KD tree search model, we improve the voxel grid method for point cloud data downsampling to reduce the number of point clouds while retaining important details.

[0021] Perform component point cloud segmentation using DBSCAN-PCA collaboration;

[0022] Work point identification and location are based on geometric constraints, frequency statistics, semantic environment, and work standards.

[0023] Furthermore, the improved voxel grid method uses cuboid voxel units. For each point after voxel downsampling, the KNN algorithm in the KD tree model is used to find the centroid point that is closest to the original point cloud, which is then used as a new point in the improved downsampling point cloud.

[0024] Furthermore, the DBSCAN-PCA collaborative component point cloud segmentation is as follows: The DBSCAN algorithm parameters r and k are set, and each point is analyzed to determine if it is a core point. If a point's neighborhood contains at least k points, it is marked as a core point. A core point is selected as the initial point of a cluster, and the KD-tree model is used to accelerate the neighborhood search, gradually expanding the cluster until it cannot be expanded further. All points directly or indirectly connected to the core point will eventually be grouped into the same cluster. Each cluster is assigned a unique label, and points not grouped into any cluster are marked as noise points. PCA principal component analysis is used on each point cloud cluster to extract the first principal component. Judgment is made based on the following geometric constraints: if the absolute value of the angle between the principal component and the Y-axis is less than 60 degrees and the absolute value of the angle between the principal component and the Z-axis is greater than 60 degrees, the cluster is determined to be a high-voltage line component point cloud; if the absolute value of the angle between the principal component and the Z-axis is less than 60 degrees and the absolute value of the angle between the principal component and the Y-axis is greater than 60 degrees, the cluster is determined to be a pole point cloud.

[0025] Furthermore, the operation point identification and positioning based on geometric constraints, frequency statistics, semantic environment, and operation standards specifically involves: using a custom cylindrical model to regularize the irregular insulator porcelain disc point cloud shape; using the RANSAC straight line fitting algorithm to identify and fit the high-voltage line; setting the base point of the custom cylindrical model as a point on the straight line fitted by RANSAC, rotating around the base point and gradually and uniformly increasing the angle between the cylinder axis and the direction of the first principal component, with a maximum angle of 30 degrees; finding the location that satisfies the insulator point cloud region by statistically analyzing the distribution of intensity values ​​and setting a threshold to filter out intensity points with lower frequencies; and using the location of the insulator point cloud region and the straight line direction fitted by the RANSAC algorithm, while meeting the technical requirement that the length of the insulation stripping position from the insulator centerline is between 50cm and 150cm.

[0026] The formula for determining the expected location of the work point for a high-voltage power line is as follows:

[0027]

[0028] Among them, the position of the insulator Centerline direction The distance L from the work point to the insulator, and the expected location of the work point on the high-voltage line. .

[0029] Compared with the prior art, the beneficial effects of the present invention and its preferred embodiments include at least the following:

[0030] 1. The dynamically adjustable collaborative robot and lidar system provided by this invention not only breaks through the field of view limitations of fixed scanning, but also can flexibly adjust the scanning angle and range as needed to ensure the acquisition of comprehensive and accurate point cloud data in complex environments; the process of the robotic arm entering the initial working posture and grabbing the end-effector tool from the tool library completes point cloud scanning and modeling by coordinating the adjustment of the lidar posture, improving hardware utilization, reducing equipment switching and operation steps, and effectively improving work efficiency.

[0031] 2. The present invention provides a dual-mode filtering solution for reflection intensity and depth to solve the problem of trailing noise in dynamic scanning; it uses cuboid voxels instead of traditional cube voxels to overcome the problem of poor dynamic adaptability of cube voxels, and combines the KNN algorithm to extract the nearest neighbor points in the original point cloud as representative points after downsampling, preserving local geometric details; DBSCAN-PCA is combined with the angle constraint of the coordinate axis to segment the point cloud of high-voltage line components; by constructing a custom cylindrical model and RANSAC conductor fitting, the insulator is identified and located based on geometric constraints and frequency statistics, and the conductor direction fitted by RANSAC is combined with industry standards (such as the distance of the insulation stripping position from the insulator is 50-150cm) to calculate the spatial position of the conductor operation point, embedding industry standards into the algorithm to realize a closed loop from reconstruction to operation guidance.

[0032] 3. The system provided by this invention has high adaptability and can work efficiently in complex three-dimensional environments where poles and conductors are intertwined. It is particularly suitable for live-line work scenarios in power distribution networks and provides a fully automated solution for intelligent operation and maintenance of power facilities. Attached Figure Description

[0033] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments:

[0034] Figure 1 A diagram of a power distribution network conductor identification and positioning system based on a single-line lidar provided in an embodiment of the present invention;

[0035] Figure 2 This is a diagram showing the installation location of a single-line lidar in the visual perception module provided in an embodiment of the present invention.

[0036] Figure 3 A schematic diagram of an embodiment of a distribution network conductor identification and positioning system based on a single-line lidar provided in this invention;

[0037] Figure 4 This is a schematic diagram of the control function of the control system provided in an embodiment of the present invention;

[0038] Figure 5 This is a flowchart of the work point identification and positioning method provided in an embodiment of the present invention;

[0039] Figure 6 A flowchart illustrating the point cloud processing algorithm for identifying insulators and conductors provided in this embodiment of the invention.

[0040] In the picture:

[0041] 1. Human-computer interaction terminal; 2. Visual perception module; 3. Robotic arm motion module; 4. Independent power supply module; 5. Dual-mode communication module; 6. Control module; 7. Insulator; 8. Crossarm; 9. Conductor; 10. Pole.

[0042] 21. Single-line lidar; 22. Rigid bracket; 23. Lower cover of bracket; 24. Upper cover of bracket; 31. Base; 32. Shoulder; 33. Wrist joint 1; 34. Wrist joint 2; 35. Wrist joint 3; 36. Elbow. Detailed Implementation

[0043] To make the features and advantages of this patent more apparent and understandable, specific embodiments are provided below for detailed explanation:

[0044] It should be noted that the following detailed descriptions are illustrative and intended to provide further explanation of this application. Unless otherwise specified, all technical and scientific terms used in this specification have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0045] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this application. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0046] refer to Figures 1-6 The overall structure of the power distribution network conductor identification and positioning system based on single-line lidar provided in this embodiment of the invention is as follows: Figure 1 As shown, the power distribution line identification and positioning system based on single-line lidar includes a human-machine interaction terminal 1 and a live-line working robot; the live-line working robot includes a vision perception module 2, a robotic arm motion module 3, an independent power supply module 4, a dual-mode communication module 5, and a control module 6; the control module 6 is composed of an industrial control computer, which serves as the core control unit and controls the vision perception module 2 and the robotic arm motion module 3 through the dual-mode communication module 5.

[0047] like Figure 2 As shown, the visual perception module 2 uses a single-line LiDAR 21 for high-precision 3D reconstruction of the power distribution network environment and target identification and positioning; the dual-mode communication module 5 consists of an industrial router, adopting a wired and wireless collaborative layout, responsible for feeding back the status information of the control module 6, visual perception module 2, robotic arm motion module 3 and independent power supply module 4 to the human-machine interaction terminal module 1 for display; the power supply module 4 consists of a 48V battery, which powers each module; the single-line LiDAR 21 is fixed to the end of the axis of the collaborative robotic arm wrist joint 34 by a rigid bracket structure 22-24 (including the rigid bracket 22 connecting the single-line LiDAR, the lower cover 23 covering the robotic arm wrist joint 2 bracket, and the upper cover 24 covering the robotic arm wrist joint 2 bracket) and its scanning plane normal vector is parallel to the axis of the wrist joint 2 bracket 34; the robotic arm motion module 3 consists of a collaborative robotic arm, used to change the scanning dimension of the visual module 2 and move the working tool to the target point. The two functions can be completed collaboratively to improve modeling and operation efficiency.

[0048] More specifically, as a preferred embodiment, the installation position of the single-line lidar in the visual perception module provided by the present invention is as follows: Figure 2As shown, a portion of the rigid support structure, the rigid support 22 of the single-line lidar is connected to the single-line lidar 21. The other portion, the symmetrical support upper cover 24 and lower cover 23, enclose the robotic arm's wrist joint 34, fixing its rotational position coaxially with wrist joint 33 and wrist joint 35, ensuring the fixed position and orientation of the single-line lidar 21. Furthermore, the collaborative robotic arm 3 is a six-joint robot. The six-joint robot includes the first four joints: base 31, shoulder 32, elbow 36, and wrist joint 33, used to adjust the pose of the single-line lidar 21. These include three joints (base 31, shoulder 32, elbow 36) primarily used to change the position of the single-line lidar 21, and the fourth joint (wrist joint 33) primarily used to adjust the angle of the single-line lidar 21. The six-joint robot also includes three joints (base 31, shoulder 32, elbow 36) primarily used to change the position of the robot's end effector, and three joints (wrist joint 33, wrist joint 34, wrist joint 35) primarily used to adjust the orientation of the robot's end effector.

[0049] The working process of the power distribution network conductor identification and positioning system based on single-line lidar provided by this invention is as follows: Figure 3 As shown, the live-line working robot is lifted by an insulated bucket truck to the area under the conductor for which live-line work is required. Then, the robot is operated by a human-computer interaction terminal to control the robotic arm motion module 3 and the vision perception module 2 to identify the conductor 9, insulator 7, crossarm 8, and pole 10, determine the position and posture of the work point on the conductor 9, and control the robotic arm to move to the target position to complete the work task.

[0050] Based on the above system device structure design, the control function of the control system provided in this embodiment of the invention is as follows: Figure 4 As shown, the control system communicates with the robotic arm and the single-line LiDAR through a dual-mode communication module. It controls the movement of the robotic arm to perform scanning modeling and end flange target point operations, controls the scanning state of the single-line LiDAR, and simultaneously acquires the angles of each joint of the robotic arm and the point cloud data collected by the single-line LiDAR. After processing by the point cloud coordinate transformation algorithm, the three-dimensional point cloud information is obtained.

[0051] As a preferred embodiment, the control module controls the robotic arm motion module to enter the initial working posture and grab the end-effector tool from the tool library through dual-mode communication. During the movement to the target position, the system pays attention to controlling the posture of the single-line lidar and performs point cloud scanning and modeling simultaneously. When the point cloud data is collected at the target position, the recognition and positioning algorithm is used to process the point cloud data to obtain the working point pose. The robotic arm can then return to the initial working posture and move to the target working point with the working tool.

[0052] More specifically, controlling the attitude of the single-line lidar involves constantly adjusting the wrist joint of the robotic arm as it moves to the target point, ensuring the lidar maintains an upward scanning angle. Once it enters the area above or below the target point, the end effector's attitude is adjusted for precise alignment and tool grasping.

[0053] Based on this, such as Figure 5 As shown, the work point identification and positioning method of the distribution network conductor identification and positioning system based on single-line lidar provided in this embodiment includes the following steps:

[0054] Step 1: By controlling the attitude of the LiDAR and enabling LiDAR scanning during the movement of the robotic arm to the tool gripping point, the robot joint angles and the point cloud data of the power distribution environment collected by the LiDAR are acquired simultaneously.

[0055] Step 2: Perform point cloud tailing filtering on the point cloud data based on depth and intensity characteristics to remove noise;

[0056] Step 3: Perform 3D reconstruction of the point cloud, which includes pose solving and point cloud coordinate transformation;

[0057] Step 4: Use point cloud processing algorithms to identify insulators and conductors from the 3D point cloud data;

[0058] Step 5: Fit and segment the 3D point cloud instance, and determine the pose of the task point based on the semantic environment and task standards.

[0059] As a preferred embodiment, Step 2 employs an innovative dual-modal trailing filter method. By combining the changes in depth and reflection intensity between each point and its neighboring points, when the depth change of adjacent point clouds exceeds a preset threshold, it is identified as a trailing point cloud and removed. The remaining point clouds are then optimized by removing outliers in the reflection intensity frequency of adjacent points. In this solution, the dual-modal data collected by a single sensor combined with this method reduces hardware complexity and cost. Furthermore, the filtering based on neighboring point relationships improves the sensitivity for filtering out trailing noise from small cross-section wires. The sensor is mounted on a robotic arm, avoiding the spatial coverage problem of fixed-site radar.

[0060] As a preferred embodiment, in Step 3, the pose is solved based on the robotic arm structural parameters, the radar-provided laser emission height parameters, and the radar mounting bracket structural parameters to obtain the DH parameters of the vision system robotic arm. After the above four-link kinematic modeling, the point cloud coordinate transformation matrix is ​​obtained, and the point cloud data is transformed to the robotic arm coordinate system to complete the 3D point cloud stitching. This scheme can significantly reduce the number of point cloud registration iterations through robotic arm kinematic constraints.

[0061] As a preferred embodiment, in Step 4, the point cloud processing algorithm identifies insulators and conductors, including the following steps: Figure 6 As shown:

[0062] Step 41: Perform pass-through filtering on the 3D point cloud data to filter out invalid point cloud data;

[0063] Step 42: Establish a KD tree search model to shorten the algorithm's running time and improve efficiency;

[0064] Step 43: Improve the voxel grid method for downsampling point cloud data to reduce the number of point clouds while retaining important details;

[0065] Step 44: Point cloud segmentation technology for high-voltage line components using DBSCAN and PCA in collaboration;

[0066] Step 45: Identification and localization of work points based on geometric constraints, frequency statistics, semantic environment, and work standards.

[0067] In a preferred embodiment, the improved voxel grid method uses cuboid voxel units instead of cube voxel units. For each point after voxel downsampling, the KNN algorithm in the KD tree model is used to find the centroid point that is closest to the original point cloud, which is then used as a new point in the improved downsampled point cloud. This scheme significantly improves the accuracy and fidelity of the downsampled point cloud while reducing the errors introduced by traditional methods.

[0068] In a preferred embodiment, the specific process of the DBSCAN and PCA collaborative point cloud segmentation technology for high-voltage line components in Step 44 is as follows: First, the parameters r and k of the DBSCAN algorithm are set, and each point is analyzed to determine whether it is a core point. If a point's neighborhood contains at least k points, it is marked as a core point. Then, a core point is selected as the initial point of a cluster, and the KD-tree model is used to accelerate the neighborhood search, gradually expanding the cluster until it cannot be expanded further. All points directly or indirectly connected to the core point will eventually be grouped into the same cluster. Then, each cluster is assigned a unique label, and points not assigned to any cluster are marked as noise points. Finally, PCA principal component analysis is used on each point cloud cluster to extract its first principal component. Based on the geometric characteristics of high-voltage lines and combined with the directional information of principal components, specific geometric constraints are set: (1) If the absolute value of the angle between the principal component and the Y-axis is less than 60 degrees and the absolute value of the angle between the principal component and the Z-axis is greater than 60 degrees, then the cluster is determined to be a high-voltage line component point cloud; (2) If the absolute value of the angle between the principal component and the Z-axis is less than 60 degrees and the absolute value of the angle between the principal component and the Y-axis is greater than 60 degrees, then the cluster is determined to be a pole point cloud.

[0069] In a preferred embodiment, in Step 45, the specific process for identifying and locating the work point based on geometric constraints, frequency statistics, semantic environment, and work standards is as follows: First, a custom cylindrical model is used to regularize the irregular insulator porcelain disc point cloud shape; then, the RANSAC straight line fitting algorithm is used to identify and fit the high-voltage line; then, the base point of the custom cylindrical model is a point on the straight line fitted by RANSAC, and the model is rotated around this base point with the angle between the cylinder axis and the direction of the first principal component gradually and uniformly increasing, with a maximum angle of 30 degrees. By statistically analyzing the distribution of intensity values ​​and setting a threshold, low-frequency intensity points are filtered out, retaining more representative point clouds in the data, eliminating interference points from connected objects, and finding the location that satisfies the insulator point cloud region. Based on the location of the insulator point cloud region and the straight line direction fitted by the RANSAC algorithm, and meeting the technical requirement that the length of the insulation stripping position from the insulator centerline should be between 50cm and 150cm, the formula for solving the expected work point location of the high-voltage line is as follows:

[0070]

[0071] Among them, the position of the insulator Centerline direction The distance L from the work point to the insulator, and the expected location of the work point on the high-voltage line. .

[0072] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0073] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

[0074] This patent is not limited to the above-described preferred embodiment. Anyone can derive other forms of a distribution network wire identification and positioning system based on single-line lidar under the guidance of this patent. All equivalent changes and modifications made within the scope of this patent application shall fall within the scope of this patent.

Claims

1. A power distribution network conductor identification and positioning system based on single-line lidar, characterized in that: The single-line lidar (21) is fixedly installed at the end of the axis of the wrist joint 2 (34) of the six-degree-of-freedom robotic arm, and the scanning plane normal vector is parallel to the axis of the wrist joint 2 (34). The scanning dimension and position of the lidar are dynamically adjusted by the robotic arm, and the identification and positioning of the power distribution wire components are completed by using component point cloud segmentation and operation point identification and positioning based on geometric constraints-frequency statistics-semantic environment-operation standards. The rigid bracket (22) is connected to the single-line laser radar (21). The upper cover (24) and lower cover (23) of the symmetrical structure of the rigid bracket (22) wrap the wrist joint (34). The wrist joint (34) is fixed to the rotation position of the wrist joint (33) and the wrist joint (35) in a coaxial manner to ensure that the position and posture of the single-line laser radar (21) are fixed. The first four joints of the robotic arm are used to adjust the position and posture of the single-line laser radar, including the base (31) for changing the position of the single-line laser radar (21), the three joints of the shoulder (32) and the elbow (36), and the wrist joint (33) for adjusting the angle of the single-line laser radar (21).

2. The distribution network conductor identification and positioning system based on single-line lidar according to claim 1, characterized in that: The robotic arm in the retracted state performs an initialization motion. After the motion is completed, it grabs the end-effector tool. During the process, the attitude of the single-line laser radar is adjusted in coordination, and point cloud scanning and modeling are performed simultaneously. When the initialization is completed, the point cloud data acquisition is completed. The method to control the attitude of the single-line laser radar is as follows: during the process of the robotic arm end position moving to the initialization end position, the wrist joint 2 (34) is adjusted to keep the single-line laser radar (21) at an upward scanning angle. When the robotic arm moves to the tool library to grab the tool, the algorithm for identifying and locating high-voltage lines and insulators is executed simultaneously to complete the pose estimation of the target point.

3. The distribution network conductor identification and positioning system based on single-line lidar according to claim 1, characterized in that: The process of identifying and locating power distribution network conductor components by dynamically adjusting the scanning dimension and position of the LiDAR using a robotic arm, and employing component point cloud segmentation and operation point identification and positioning based on geometric constraints, frequency statistics, semantic environment, and operation standards includes: By controlling the attitude of the LiDAR and enabling LiDAR scanning during the movement of the robotic arm to the tool gripping point, the robot joint angles and the point cloud data of the power distribution environment collected by the LiDAR are acquired simultaneously. Point cloud data is subjected to point cloud tailing filtering and noise reduction based on depth and intensity dual-modality. Perform 3D reconstruction of point clouds, including pose solving and point cloud coordinate transformation; Processing 3D point cloud data to identify target components; The 3D point cloud instance is fitted and segmented, and the pose of the task point is determined based on the semantic environment and task standards.

4. A power distribution network conductor identification and positioning system based on a single-line lidar according to claim 3, characterized in that: The point cloud trailing filter combines the changes in depth and reflection intensity between each point and its neighboring points. When the depth change of neighboring point clouds exceeds a preset threshold, it is identified as a trailing point cloud and removed. The remaining point clouds are then optimized by removing abnormal values ​​of reflection intensity frequency of neighboring points.

5. A power distribution network conductor identification and positioning system based on a single-line lidar according to claim 3, characterized in that: The pose calculation is based on the structural parameters of the robotic arm, the height parameters of the emitted laser provided by the radar, and the structural parameters of the radar mounting bracket to obtain the DH parameters of the vision system robotic arm. After 4-link kinematic modeling, the point cloud coordinate transformation matrix is ​​obtained, and the point cloud data is transformed to the robotic arm coordinate system to complete the 3D point cloud stitching.

6. A power distribution network conductor identification and positioning system based on a single-line lidar according to claim 3, characterized in that: The process of processing 3D point cloud data to identify target components specifically includes the following steps: Perform pass-through filtering on the 3D point cloud data to filter out invalid point cloud data; Based on the KD tree search model, we improve the voxel grid method for point cloud data downsampling to reduce the number of point clouds while retaining important details. Perform DBSCAN-PCA collaborative component point cloud segmentation; Work point identification and location are based on geometric constraints, frequency statistics, semantic environment, and work standards.

7. A power distribution network conductor identification and positioning system based on a single-line lidar according to claim 6, characterized in that: The improved voxel grid method uses cuboid voxel units. For each point after voxel downsampling, the KNN algorithm in the K-Dtree model is used to find the centroid point that is closest to the original point cloud, which is then used as the new point in the improved downsampling point cloud.

8. A power distribution network conductor identification and positioning system based on a single-line lidar according to claim 6, characterized in that: The DBSCAN-PCA collaborative component point cloud segmentation is as follows: set the parameters r and k of the DBSCAN algorithm, analyze each point, and determine whether it is a core point; if a point contains at least k points in its neighborhood, it is marked as a core point; select a core point as the initial point of the cluster, use the KD tree model to accelerate the neighborhood search, and gradually expand the cluster until it can no longer be expanded. All points that are directly or indirectly connected to the core point will eventually be grouped into the same cluster; each cluster is assigned a unique label, and points that are not grouped into any cluster are marked as noise points. For each point cloud cluster, PCA principal component analysis was performed to extract the first principal component; The determination is based on the following geometric constraints: if the absolute value of the angle between the principal component and the Y-axis is less than 60 degrees and the absolute value of the angle between the principal component and the Z-axis is greater than 60 degrees, then the cluster is determined to be a point cloud of a high-voltage line component. If the absolute value of the angle between the principal component and the Z-axis is less than 60 degrees and the absolute value of the angle between the principal component and the Y-axis is greater than 60 degrees, then the cluster is determined to be a pole point cloud.

9. A power distribution network conductor identification and positioning system based on a single-line lidar according to claim 6, characterized in that: The specific steps for identifying and locating work points based on geometric constraints, frequency statistics, semantic environment, and work standards are as follows: A custom cylindrical model is used to regularize the irregular point cloud of the insulator porcelain disc; the RANSAC linear fitting algorithm is used to identify and fit the high-voltage line; the base point of the custom cylindrical model is set as a point on the line fitted by RANSAC; the model is rotated around the base point, and the angle between the cylinder axis and the direction of the first principal component gradually and uniformly increases, with a maximum angle of 30 degrees; by statistically analyzing the distribution of intensity values ​​and setting a threshold to filter out intensity points with lower frequencies, the location satisfying the insulator point cloud region is found; the location of the insulator point cloud region and the direction of the line fitted by the RANSAC algorithm are used, and the technical requirement that the length of the insulation stripping position from the insulator centerline is between 50cm and 150cm is met; The formula for determining the expected location of the work point for a high-voltage power line is as follows: Among them, the position of the insulator Centerline direction The distance L from the work point to the insulator, and the expected location of the work point on the high-voltage line. .

Images