Distribution line equipment parameter calculation method using three-dimensional point cloud

Abstract

The invention discloses a distribution line equipment parameter calculation method using a three-dimensional point cloud, and the method comprises the following steps: employing advanced visible light shooting equipment or other three-dimensional sensing equipment to obtain the point cloud data of distribution line equipment and the surrounding environment of the distribution line equipment; the detailed geometrical shape and spatial distribution of the distribution line equipment are obtained through the point cloud data, and a high-quality data source is provided for subsequent equipment parameter extraction; after the point cloud data are obtained, the point cloud data are preprocessed so as to improve the quality of the point cloud data. Three-dimensional point cloud data are collected through visible light shooting equipment and a high-precision camera, and high-precision extraction of equipment parameters is realized in combination with principal component analysis and an optimization algorithm. The measurement efficiency is improved through automatic processing of point cloud data, parameters are quickly calculated by adopting an octree and recursive segmentation algorithm for complex equipment, and the method has relatively high adaptability and anti-noise capability and is suitable for various detection scenes.

Description

Technical Field

[0001] The present invention relates to the technical field of distribution line equipment parameter calculation, and specifically relates to a method for calculating distribution line equipment parameters using three-dimensional point clouds. Background Art

[0002] With the rapid development of the power system, the quantity and types of distribution network equipment are constantly increasing. Traditional state detection and maintenance of distribution line equipment often rely on manual inspections and traditional measurement methods. However, traditional measurement methods have problems such as low accuracy, low efficiency, and high labor costs. Especially for high-altitude equipment (such as poles, crossarms, insulators, etc.) and equipment in complex environments (such as high-voltage lines, densely tree-covered areas, etc.), the accuracy and reliability of traditional measurement methods are often greatly limited. Manual measurement not only consumes a large amount of time and labor, but is also easily affected by factors such as weather, environment, and operator experience, resulting in large errors in measurement results, which affects the operation and maintenance efficiency of distribution lines and the reliability of equipment. In addition, the real-time performance of traditional methods is poor, and it is impossible to achieve real-time monitoring and dynamic maintenance of the state of distribution network equipment, thereby affecting the overall safety and stability of the power system.

[0003] In recent years, with the rapid development of three-dimensional point cloud technology, they have achieved remarkable applications in various fields such as equipment detection and environmental modeling. However, there are still certain deficiencies in the existing distribution network equipment detection technology based on three-dimensional point cloud technology during application. First, although three-dimensional point cloud data can provide accurate spatial geometric information, how to accurately and quickly extract the key parameters of equipment (such as pole height, crossarm length, insulator diameter, etc.) from a large amount of point cloud data is still a major challenge. Existing methods mostly rely on manual intervention or use simple geometric algorithms, and it is difficult to handle situations of complex equipment shapes or interference between equipment, and the extraction accuracy and calculation efficiency are not ideal. Second, although the existing three-dimensional point cloud processing methods can extract basic geometric parameters, there are still certain limitations when dealing with the complex structures of equipment (such as insulator volume, pole tilt angle, etc.), especially when dealing with high-resolution and dense point cloud data, the calculation complexity is high and the processing speed is slow, which cannot meet the actual needs of large-scale distribution line equipment detection. Therefore, there is an urgent need for a new technical solution to solve the problems of insufficient accuracy, low efficiency, and complex calculation existing in the prior art through more efficient algorithms and automated methods, and to achieve accurate calculation and automated extraction of distribution line equipment parameters.

[0004] The above information disclosed in the background art section is only used to enhance the understanding of the background of the present disclosure, and thus it may include information that does not constitute the prior art known to those of ordinary skill in the art. Summary of the Invention

[0005] The object of the present invention is to provide a method for calculating the parameters of distribution line equipment using three-dimensional point clouds. High-resolution three-dimensional point cloud data is obtained through the combined acquisition of a visible light photographing device and a high-precision camera, and combined with the principal component analysis algorithm and various optimization algorithms, high-precision extraction of the parameters of distribution line equipment is realized, solving the problem of insufficient measurement accuracy of traditional methods in complex environments. Through the automatic processing of point cloud data, including preprocessing, spatial alignment, and parameter extraction, without manual intervention, the measurement efficiency is greatly improved. Especially for complex equipment such as insulators, combined with the octree structure and the recursive segmentation algorithm, the parameter calculation is quickly completed. The present invention has strong adaptability and robustness, can eliminate the influence of noise in complex environments, ensure high-quality data processing, and is applicable to various equipment detection scenarios to solve the problems in the above-mentioned background technology.

[0006] To achieve the above object, the present invention provides the following technical solution: A method for calculating the parameters of distribution line equipment using three-dimensional point clouds, comprising the following steps:

[0007] Step 1: Obtain the three-dimensional point cloud data of the distribution line equipment;

[0008] In this step, an advanced visible light photographing device or other three-dimensional sensing devices (such as high-precision cameras, LiDAR sensors, etc.) are used to obtain the point cloud data of the distribution line equipment and its surrounding environment. The detailed geometric shape and spatial distribution of the distribution line equipment are obtained through the point cloud data, providing a high-quality data source for subsequent equipment parameter extraction. The acquisition of three-dimensional point cloud data effectively solves the problems of accuracy and efficiency of traditional measurement methods.

[0009] Step 2: Preprocess the point cloud data and perform spatial alignment;

[0010] After obtaining the point cloud data, it needs to be preprocessed, mainly including steps such as denoising, filtering, and density optimization to improve the quality of the point cloud data. In addition, the principal component analysis (PCA) method is used to perform spatial alignment on the point cloud to ensure that the main axis direction of the point cloud data is consistent with the global coordinate system, thereby providing a standardized reference coordinate system for subsequent parameter calculation.

[0011] Step 3: Extract the tower height;

[0012] In this step, the RANSAC algorithm is used to fit the ground point cloud, determine the ground plane parameters, and determine the height of the tower according to this plane. The tower height is realized by measuring the vertical distance between the highest point of the tower point cloud and the ground plane, thereby obtaining accurate tower height data. The technology of this step can accurately measure the tower height automatically and efficiently.

[0013] Step 4: Extract the cross-arm length;

[0014] In this step, the length of the cross arm is extracted by the minimum circumscribed circle algorithm. First, a reference plane is fitted, and the cross arm point cloud is projected onto this plane. Then, the length of the cross arm is calculated based on the diameter of the minimum circumscribed circle. Through the optimization algorithm, the projection error can be minimized to provide accurate cross arm length data.

[0015] Step 5: Extract the structural parameters of the insulator;

[0016] In this step, the structural parameters of the insulator are calculated by multiple algorithms. First, the oriented bounding box (OBB) method is used to determine the main direction and position of the insulator point cloud and calculate its structural height. Then, the point cloud is projected onto the horizontal plane, and the disk diameter of the insulator is extracted using the minimum circumscribed circle. Finally, the convex hull and concave hull algorithms are used to calculate the volume of the insulator to obtain the geometric shape information of the insulator. This technology enables the precise measurement of complex-shaped devices such as insulators.

[0017] Preferably, the three-dimensional point cloud data acquisition in Step 1 is jointly performed by a visible light imaging device scan (LiDAR) and a high-precision camera to ensure more complete and high-resolution point cloud data acquisition under different environmental conditions. The combined use of the visible light imaging device scan and the camera can effectively improve the density and accuracy of the point cloud data.

[0018] Preferably, the point cloud data preprocessing in Step 2 further includes using the Gaussian filtering algorithm to smooth the point cloud, removing the pseudo points generated by factors such as device noise and external interference, and performing hierarchical processing on the point cloud data through the grid method to improve the data density uniformity and the stability of subsequent processing.

[0019] Preferably, the tower height extraction in Step 3 is improved in accuracy through multi-stage processing. The specific steps are as follows:

[0020] The RANSAC algorithm is used to roughly fit the ground point cloud to obtain the preliminary parameters of the ground plane;

[0021] The tower point cloud is further refined and fitted to eliminate the possible errors between the ground and the tower;

[0022] Through the maximum vertical distance calculation method, the vertical distance between the highest point of the tower and the fitted ground plane is extracted to obtain the accurate height of the tower.

[0023] Preferably, the cross arm length extraction in Step 4 is realized through the following detailed steps:

[0024] A reference plane is fitted based on the three-dimensional data of the cross arm point cloud as the projection reference plane of the cross arm point cloud;

[0025] Vertically project the cross-arm point cloud onto the reference plane, and use the minimum circumscribed circle algorithm to extract the diameter of the minimum circumscribed circle of the cross-arm point cloud as the length of the cross-arm;

[0026] In order to eliminate the influence of uneven point cloud density or noise, the weighted average method is used to synthesize multiple fitting results to optimize the calculation accuracy of the cross-arm length.

[0027] Preferably, the extraction of the insulator structure parameters in step 5 is realized in detail through the following steps:

[0028] Use the oriented bounding box (OBB) method to preliminarily enclose the insulator point cloud, extract the axial direction of the insulator and determine its position;

[0029] By rotating and projecting the insulator point cloud, select a suitable projection plane (such as the XZ plane), and use the minimum circumscribed circle algorithm to calculate the nominal disc diameter of the insulator;

[0030] Based on the high-density area of the point cloud, use the convex hull and concave hull algorithms to extract the convex hull volume and concave hull volume of the insulator respectively;

[0031] For high-resolution point clouds, adopt a voxel-based block processing method to improve the accuracy and robustness of the concave hull volume calculation.

[0032] Preferably, the point cloud data preprocessing in step 2 optimizes the density distribution of the point cloud data by combining a multi-scale filtering algorithm. The multi-scale filtering algorithm filters at different scales, dynamically adjusts the density of the point cloud, adapts to the needs of different devices and environments, so as to provide more processing capabilities for high-density point cloud areas and filter out noise in low-density areas.

[0033] Preferably, the insulator volume extraction in step 5 is performed by recursively dividing the insulator point cloud to ensure the accuracy of the volume calculation. The recursive division method gradually divides the insulator point cloud into multiple voxel blocks, and uses an octree data structure for spatial division and adjacent point search, and finally accurately calculates the insulator volume, reducing errors and improving calculation efficiency.

[0034] In the above technical solution, the technical effects and advantages provided by the present invention:

[0035] The present invention uses a combined acquisition method that combines a visible light imaging device and a high-precision camera to obtain high-resolution three-dimensional point cloud data of distribution line equipment, ensuring that the data has a higher spatial density and geometric integrity. For distribution line equipment in complex environments, the spatial alignment of the point cloud data is achieved by using the principal component analysis algorithm, and combined with a variety of optimization algorithms, the accuracy of equipment parameter extraction is significantly improved. For example, the height of the pole tower is accurately calculated by the vertical distance from the fit of the ground plane to the highest point, the length of the cross arm is obtained by the circumscribed circle algorithm optimized multiple times, and the volume of the insulator is calculated by recursive segmentation of the high-resolution point cloud. These methods not only reduce the errors in manual measurement but also effectively solve the problem of insufficient measurement accuracy of traditional methods in complex environments.

[0036] The present invention greatly improves the efficiency of distribution line equipment measurement by realizing the automated processing of three-dimensional point cloud data. The preprocessing, spatial alignment, and key parameter extraction of the point cloud data are all completed with the support of algorithms without manual intervention. Especially for parameter extraction of equipment with complex shapes such as insulators, the present invention combines the octree structure and the recursive segmentation algorithm, which not only improves the calculation efficiency but also can quickly complete the calculation of volume and other parameters under high-resolution point clouds. This efficient measurement method significantly reduces the time cost required for equipment inspection and maintenance, and at the same time provides a practical solution for large-scale distribution line equipment detection.

[0037] The three-dimensional point cloud technology of the present invention combined with a variety of optimization algorithms makes the equipment parameter measurement have higher adaptability and robustness. Through the combined acquisition of a visible light imaging device and a high-precision camera, high-quality point cloud data can still be obtained even under complex environmental conditions. In addition, the RANSAC algorithm and region growing algorithm adopted by the present invention have strong anti-noise capabilities and can eliminate the influence of outliers and environmental noise on data processing. Combined with the multi-stage fitting method and optimization strategy, accurate geometric feature extraction can be achieved even if the point cloud data density is uneven or there are local missing parts. Therefore, the present invention is not only applicable to conventional equipment detection scenarios but also can maintain stable performance under special conditions, providing strong technical support for the measurement of distribution line equipment parameters. Description of the Drawings

[0038] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required to be used in the embodiments. Obviously, the drawings described below are only some embodiments recorded in the present invention, and those of ordinary skill in the art can also obtain other drawings based on these drawings.

[0039] Figure 1 It is a method flow chart of a method for calculating distribution line equipment parameters using three-dimensional point clouds according to the present invention.

[0040] Figure 2 Schematic diagram for extracting the minimum bounding box of the present invention.

[0041] Figure 3 Schematic diagram for extracting the convex hull in the schematic diagram for extracting the volume of a tree of the present invention.

[0042] Figure 4 Schematic diagram for extracting the concave hull in the schematic diagram for extracting the volume of a tree of the present invention.

[0043] Figure 5 Schematic diagram for extracting the height of a tree of the present invention.

[0044] Figure 6 Point cloud map of the projection of a tree of the present invention.

[0045] Figure 7 Schematic diagram for extracting the projected area of a tree of the present invention.

[0046] Figure 8 Schematic diagram for extracting the leaf area of the present invention.

[0047] Figure 9 Schematic diagram of the height and stem diameter of tree 1 of the present invention.

[0048] Figure 10 Schematic diagram of the height and stem diameter of tree 2 of the present invention.

[0049] Figure 11 Schematic diagram of the height and stem diameter of tree 3 of the present invention.

[0050] Figure 12 Flow chart for extracting the structural parameters of the tower of the present invention.

[0051] Figure 13 Schematic diagram for measuring the height of the tower of the present invention.

[0052] Figure 14 Schematic diagram for calculating the length of the cross arm of the present invention.

[0053] Figure 15 Circumscribed rectangular bounding box diagram of the insulator of the present invention.

[0054] Figure 16 Structural height diagram of the insulator of the present invention.

[0055] Figure 17 Nominal disc diameter diagram of the present invention.

[0056] Figure 18 Schematic diagram of the volume of the insulator of the present invention.

[0057] Figure 19 Schematic diagram for extracting the structural parameters of the insulator based on 3D point cloud of the present invention.

[0058] Figure 20 The pavement point cloud map after segmentation for the present invention.

[0059] Figure 21 Schematic diagram of the minimum distance retrieval algorithm for the present invention.

[0060] Figure 22 Schematic diagram of the region growing algorithm for the present invention. Detailed implementation manners

[0061] Example embodiments will now be described more fully with reference to the accompanying drawings. However, the example embodiments can be implemented in various forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the concept of the example embodiments to those skilled in the art.

[0062] The present invention provides a method for calculating parameters of distribution line equipment using three-dimensional point clouds as shown in Figure 1 the following, including the steps:

[0063] Step 1: Obtain three-dimensional point cloud data of distribution line equipment;

[0064] In this step, an advanced visible light photographing device or other three-dimensional sensing devices (such as high-precision cameras, LiDAR sensors, etc.) are used to obtain the point cloud data of the distribution line equipment and its surrounding environment. The detailed geometric shape and spatial distribution of the distribution line equipment are obtained through the point cloud data, providing a high-quality data source for subsequent extraction of equipment parameters. The acquisition of three-dimensional point cloud data effectively solves the problems of accuracy and efficiency of traditional measurement methods.

[0065] A visible light photographing device refers to an imaging device that can capture natural light (in the wavelength band of 400 nm to 700 nm), mainly including high-precision digital cameras, single-lens reflex cameras, industrial cameras, and drone-mounted cameras, etc. These devices obtain image data of the distribution line equipment and its surrounding environment through photographing, and have the characteristics of high resolution, rich details, and flexible operation. In the acquisition of three-dimensional point cloud data, the commonly used multi-view stereo photogrammetry (MVS) relies on visible light photographing devices to take multiple images of the target object from different angles, and generates high-precision three-dimensional point cloud data by calculating the overlapping areas of the images.

[0066] Step 2: Preprocess the point cloud data and perform spatial alignment;

[0067] After obtaining the point cloud data, it is necessary to preprocess it, mainly including steps such as denoising, filtering, and density optimization to improve the quality of the point cloud data. In addition, the principal component analysis (PCA) method is used to perform spatial alignment on the point cloud to ensure that the main axis direction of the point cloud data is consistent with the global coordinate system, thereby providing a standardized reference coordinate system for subsequent parameter calculation.

[0068] Step 3: Extract the pole height;

[0069] In this step, the RANSAC algorithm is used to fit the ground point cloud, determine the ground plane parameters, and determine the height of the pole according to this plane. The pole height is achieved by measuring the vertical distance between the highest point of the pole point cloud and the ground plane, thereby obtaining accurate pole height data. The technology of this step can accurately measure the pole height automatically and efficiently.

[0070] Step 4: Extract the cross-arm length;

[0071] In this step, the length of the cross-arm is extracted by the minimum circumscribed circle algorithm. First, a reference plane is fitted, and the cross-arm point cloud is projected onto this plane, and then the length of the cross-arm is calculated according to the diameter of the minimum circumscribed circle. By optimizing the algorithm, the projection error can be minimized to provide accurate cross-arm length data.

[0072] Step 5: Extract the insulator structure parameters;

[0073] In this step, the structure parameters of the insulator are calculated by various algorithms. First, the oriented bounding box (OBB) method is used to determine the main direction and position of the insulator point cloud and calculate its structural height; then, the point cloud is projected onto the horizontal plane, and the disk diameter of the insulator is extracted using the minimum circumscribed circle; finally, the convex hull and concave hull algorithms are used to calculate the volume of the insulator to obtain the geometric shape information of the insulator. This technology makes it possible to accurately measure complex-shaped devices such as insulators.

[0074] Embodiment 1: Joint point cloud data acquisition and processing based on a visible light photographing device and a high-precision camera;

[0075] In this embodiment, a joint scheme of a visible light photographing device (LiDAR) and a high-precision camera is used for the three-dimensional point cloud data acquisition of the distribution line equipment. The visible light photographing device scanning technology can operate stably under complex environmental conditions (such as rainy weather, night work, etc.). Especially when there are high-altitude devices (such as poles) and there are obstacles (such as trees, buildings, etc.) around the devices, it can provide relatively accurate three-dimensional data. The visible light photographing device can quickly scan a large area by emitting visible light beams and receiving reflected signals, generating point cloud data with spatial positions (XYZ coordinates). These point cloud data not only reflect the spatial geometric shape of the distribution equipment but also contain the structural information of the surrounding environment, providing a solid foundation for the subsequent extraction of equipment parameters.

[0076] However, when a visible light imaging device scans to provide geometric shape information, it usually lacks details (such as color, texture, etc.). In this case, combining high-precision camera data will play an important role. The high-precision camera can supplement the missing detailed information in the point cloud data, provide clear image texture and object edges, and help further improve the extraction accuracy of device parameters. Therefore, jointly collecting data using a visible light imaging device and a high-precision camera can provide comprehensive and high-quality three-dimensional point cloud data under different environments and device structures.

[0077] After the data acquisition is completed, it enters the preprocessing stage of the point cloud data. Since the collected point cloud data often contains noise and outliers, denoising processing is required. In this embodiment, the Gaussian filtering algorithm is used to smooth the point cloud, thereby removing the spurious points generated due to equipment failures, environmental interference, or motion errors. The Gaussian filtering algorithm can retain the main features of the data and remove most of the high-frequency noise, ensuring the stability and accuracy of subsequent processing.

[0078] Next, spatial alignment of the point cloud data is performed. Through the principal component analysis (PCA) method, the point cloud data is rotated and translated to align it with the global coordinate system. The PCA algorithm calculates the covariance matrix of the point cloud data to obtain the main direction and secondary main direction of the point cloud, thereby determining the axis of the point cloud and rotating it to be consistent with the standard coordinate axes. The spatially aligned point cloud data is easier to process, ensuring consistency in subsequent steps.

[0079] In the process of extracting the tower height, first, the RANSAC algorithm is used to fit the ground point cloud to determine the parameters of the ground plane. The RANSAC algorithm is a robust fitting method that can effectively identify the data set containing noise from a large amount of point cloud data and fit a relatively accurate ground plane. Then, the height of the tower is calculated, that is, the vertical distance from the highest point of the tower to the ground plane. In the calculation process, a multi-stage fitting method is used. First, rough fitting is performed, and then refined fitting is carried out to eliminate the errors caused by environmental complexity (such as ground undulation).

[0080] The extraction of the cross-arm length uses the minimum circumscribed circle algorithm. After fitting the three-dimensional point cloud of the cross-arm, first, a reference plane is determined as the measurement reference, and the point cloud of the cross-arm is vertically projected onto this plane. Subsequently, the minimum circumscribed circle algorithm is applied to calculate the diameter of the minimum circumscribed circle of the cross-arm point cloud, thereby obtaining the length of the cross-arm. To reduce the influence of uneven point cloud density or noise, the weighted average method is used for optimization among multiple fitting results, further improving the calculation accuracy.

[0081] For the extraction of the structural parameters of insulators, first use the oriented bounding box (OBB) algorithm to determine the orientation and position of the insulators. The OBB algorithm can calculate a smallest rectangular bounding box based on the point cloud data, which can tightly wrap the insulators and reduce the errors caused by the sparse or different densities of the point cloud data. Through this method, the axial position and height of the insulators can be extracted. Subsequently, project the point cloud data of the insulators onto a suitable plane, usually choose the XZ plane for projection, and then use the minimum circumscribed circle algorithm to calculate the nominal disc diameter. To further improve the calculation accuracy, the convex hull and concave hull algorithms are combined to extract the volume of the insulators. The convex hull volume can describe the external contour of the insulators, while the concave hull volume can more accurately reflect its actual geometric shape and provide a more accurate volume estimate.

[0082] Through this data acquisition method that combines visible light photographing equipment and high-precision cameras, and combines a variety of efficient algorithms, it is possible to efficiently and accurately extract the parameters of distribution line equipment in complex environments, significantly improving the accuracy and efficiency of distribution network equipment detection.

[0083] Embodiment 2: Parameter extraction of poles, cross arms and insulators under high-resolution point cloud data;

[0084] This embodiment details the extraction of distribution line equipment parameters under high-resolution point cloud data. In this embodiment, first collect high-precision point cloud data through the combined use of visible light photographing equipment and high-precision cameras. High-resolution point cloud data has a high spatial density and can more finely depict the geometric shape of distribution equipment, which provides the necessary data support for the accurate extraction of equipment parameters. However, since point cloud data is usually affected by factors such as noise, environmental factors and equipment accuracy, strict filtering and denoising processing are required in the data preprocessing stage.

[0085] In the data preprocessing process, a multi-scale filtering algorithm is adopted to dynamically adjust the point cloud in different density regions. In regions with higher density, a higher-precision filtering method is used to retain more detailed information; while in low-density regions, the sampling density of the point cloud is increased to ensure the uniformity of the data. In addition, to eliminate the noise and outliers in the point cloud, a region growing algorithm is combined for point cloud denoising. The region growing algorithm can effectively eliminate the noise points far from the main region and retain the main structure point cloud, improving the quality of the point cloud data.

[0086] After data preprocessing, the point cloud is spatially aligned through the PCA algorithm, and the point cloud data is transformed into the standard coordinate system. This process ensures the consistency of the point cloud data, making the subsequent parameter extraction more accurate. In terms of tower height extraction, first, the RANSAC algorithm is used to fit the ground point cloud to determine the position of the ground plane, and then the tower height is calculated by the maximum vertical distance method. To further improve the calculation accuracy, in complex terrains, a multi-stage fitting strategy is used: fine fitting is carried out on the basis of rough fitting to remove the influence caused by equipment errors or ground undulations.

[0087] The extraction of the cross-arm length is achieved through the minimum circumscribed circle algorithm. First, the reference plane of the cross-arm point cloud is fitted, and it is vertically projected onto this plane. On the projected plane, the minimum circumscribed circle algorithm is used to calculate the diameter of the cross-arm, thereby obtaining the length of the cross-arm. To further improve the calculation accuracy, the weighted average method is adopted to synthesize multiple fitting results and reduce the influence of the point cloud density difference and noise on the results.

[0088] In terms of insulator parameter extraction, the oriented bounding box (OBB) algorithm is adopted to determine the direction of the insulator, and its structural height is calculated based on the point cloud data. At the same time, the nominal disc diameter is calculated through the minimum circumscribed circle algorithm, and the volume of the insulator is calculated in combination with the convex hull and concave hull algorithms. To optimize the volume calculation accuracy, the recursive segmentation method is adopted, and spatial segmentation is carried out through the octree structure, making the calculation more accurate and with higher calculation efficiency.

[0089] This parameter extraction method based on high-resolution point cloud data can effectively improve the accuracy and efficiency of equipment parameter extraction, and comprehensively monitor and accurately evaluate the distribution line equipment in complex environments.

[0090] Embodiment 3: A volume calculation method based on region growing and octree optimization algorithm;

[0091] This embodiment mainly solves the problem of how to efficiently and accurately calculate the volume of the complex morphological structure (such as insulators) of distribution line equipment under high-resolution point cloud data. In traditional point cloud volume calculation methods, due to the complexity and irregularity of point cloud data, traditional methods often require a large amount of computing resources, the calculation process is time-consuming and vulnerable to noise interference, and the accuracy is poor. Therefore, in this embodiment, through the optimized calculation method, the calculation efficiency and accuracy are greatly improved.

[0092] First, in the point cloud data preprocessing stage, a region growing algorithm is adopted for denoising. The region growing algorithm can automatically select the growth region according to the density of the point cloud, and effectively remove false points and noise points, ensuring that subsequent calculations are based on high-quality point cloud data. In this algorithm, first, a high-density region is selected as the initial point, and then, based on the radius of the sphere, the radius range is gradually increased, and the points within the sphere are processed to ensure that only the point cloud data related to the main structure is retained, removing outliers and noise.

[0093] Next, the recursive segmentation method is used to divide the point cloud data into multiple small voxel blocks, and the octree data structure is applied for spatial segmentation. The octree is an efficient data structure that recursively divides the point cloud data into multiple small blocks, and each small block can be independently processed and calculated. The point cloud data within each voxel block is analyzed spatially through the adjacent point search method to accurately calculate its volume. This method can significantly reduce the error in calculations, improve the accuracy of volume calculation, and at the same time reduce the computational complexity.

[0094] Through this optimized recursive segmentation and octree structure, the volume calculation not only improves the calculation accuracy but also speeds up the calculation speed. Especially when dealing with high-resolution point cloud data, it has great advantages. This method is not only applicable to the volume calculation of complex-shaped devices such as insulators but can also be widely applied to the extraction of other types of device parameters, ensuring the efficiency and accuracy of the distribution line equipment detection work.

[0095] Through the above innovative point cloud data processing method, the parameter extraction of distribution line equipment can be completed quickly and accurately, greatly improving the operation and maintenance efficiency and management level of the distribution network.

[0096] The bounding box is a geometric space that can replace a discrete point set or an object with a relatively simple geometric body. The bounding box includes the AABB type, OBB type, bounding sphere type, etc. The AABB type bounding box is easily affected by the position of the object, and there will be a gap between the object and the box body, resulting in measurement errors. Due to its own geometric shape problems, the bounding sphere type is also difficult to apply in practice. The OBB type can avoid the volume error caused by the three-dimensional coordinates of the plant and can more closely wrap the entire plant. Therefore, the OBB type bounding box is used in this paper. The main steps are as follows: Use the PCA principal component analysis method to calculate the covariance matrix of the point cloud, determine the eigenvectors of the target point cloud in three directions. Among them, the eigenvector corresponding to the maximum eigenvalue is the main axis direction of the OBB type bounding box. After determining the three main directions of the point cloud, project all points onto this direction, and calculate the maximum and minimum coordinate values of the point cloud within each main axis respectively. Half of the sum of the two coordinate values is the coordinate center of the bounding box, and the sum of the two coordinate values is the side length of the bounding box. The bounding box extraction is as Figures 2-4 shown.

[0097] The tree volume is divided into the concave hull volume and the convex hull volume. The concave hull can reflect the actual volume of the canopy. In this paper, the Alphashapes method is extended from two dimensions to three dimensions to extract the concave hull volume. The convex hull refers to the smallest convex set that contains a finite point set P in any-dimensional space. It is composed of the vertices of the convex hull and appears as a convex polyhedron in three dimensions, which can reflect the contour shape of the canopy. In this paper, the Quick Hull algorithm is used to extract the convex hull. The construction of the concave hull and convex hull volumes is as Figures 3-4 shown.

[0098] Plant height is a basic parameter of plant growth. The lowest point of the branch is used as the starting point for measurement. The numerical value of the position of this point in the vertical direction is the smallest. The ending point of the measurement is selected at the place where the tree growth ends. The numerical value of the position of this point in the vertical direction is the largest. The difference between these two points is taken as the tree height. When measuring the point cloud data, since the main direction of the point cloud is not consistent with the world coordinate axis, first, the spatial position of the point cloud is adjusted by rotation and translation to convert the main direction of the point cloud to the world coordinate. In this paper, the method of principal component analysis is used to make their directions consistent. The conversion steps are as follows:

[0099] Calculate the centroid coordinates of the target point cloud;

[0100] Use the principal component analysis method to calculate the coordinate basis vectors of the target point cloud. The target point cloud is centered, and the eigenvalues and eigenvectors of the covariance matrix of the centered point cloud are solved. The eigenvectors are the basis vectors of the point cloud, that is, the main direction.

[0101] Calculate the rotation matrix and translation matrix of the point cloud centroid and the main direction relative to the world coordinate.

[0102] Through the rotation matrix and translation matrix, transfer the point cloud centroid to the origin of the world coordinate, and the position of the main direction of the point cloud is the position of the coordinate axis of the world coordinate.

[0103] The point cloud coordinate conversion formula is as shown in the formula:

[0104] P z = RP y + T

[0105] In the formula, P z represents the converted point cloud, R represents the rotation matrix, T represents the translation matrix, and P y represents the original point cloud.

[0106] After the coordinate conversion, the Z-axis direction is the main growth direction of the tree. Traverse the target point cloud to extract the maximum and minimum values of the point cloud coordinates in the Z-axis direction. The difference between the two is the plant height (Liang Xiuying et al., 2020). The schematic diagram of tree plant height extraction is as Figure 5 shown.

[0107] Calculate the projected area based on the segmented blade point cloud. First, for the plane model of AX + BY + CZ + D = 0, create a plane with coefficients A = B = C = 0 and D = 1, which is the X - Y plane. All the point clouds related to the Z - axis are projected onto the Z - plane. Traverse all the point clouds and repeat the above steps. In this plane model, the Z - coordinate values of all the point clouds are 0, that is, all the point clouds are projected onto the specified plane. Then use the Alpha shapes method to extract the outer contour of the point cloud, and after meshing, calculate the area of all triangular meshes. The sum of all the areas is the projected area of the point cloud. The extraction of the tree projected area is shown in the figure, and the process of extracting the outer contour of the point cloud is as follows:

[0108] If there are N points, choose any two points to connect. The total number of connected lines is N(N - 1) / 2 pairs. If there are I sets, the total possible number of connection pairs is N(I - 1)(N - 1).

[0109] During the rolling process, based on the radius R of the sphere, disconnect the connections with a distance exceeding 2R and roll along the tangent direction of the circle.

[0110] For each group of remaining connections, construct a circumcircle with a radius of R with two points as the diameter. Then there are at most 2 feasible solutions.

[0111] All the remaining circumcircles corresponding to the connections form the boundary information.

[0112] Based on the above - mentioned meshed model after triangulation, the blade area can be calculated. The reconstructed meshed model consists of numerous triangular meshes. The vertex coordinates of each triangular mesh are the same as those of the original point cloud. Through the vertex coordinates, the lengths of each side of each triangular mesh can be calculated. Then, with the help of Heron's formula, the area of each triangular mesh can be calculated. Finally, traverse all the triangular meshes of the blade and accumulate the areas using the summation formula to obtain the blade area. The schematic diagram of blade area extraction is as Figure 8 shown.

[0113]

[0114] In the formula: S represents the area of a single triangular mesh, p l represents half of the perimeter of the triangular mesh, a, b, c represent the lengths of each side of the triangular mesh, i represents the triangular mesh index, j represents the total number of triangular meshes, and S z represents the total area of the blade.

[0115] Compare the measured values with the simulated values obtained from 3D reconstruction to evaluate the accuracy of the 3D reconstruction model. Use R2, MAPE, and RMSE as the evaluation criteria for the modeling accuracy. The calculation formulas are as follows:

[0116]

[0117] In the formula, n represents the number of trees, and Y i represents the measured parameter value, represents the average value of the measured parameters, and X i represents the simulated parameter value, represents the average value of the simulated parameters.

[0118] The tree height starts from the lowest point of the tree growth by removing the ground point cloud and ends at the end point of the growth. The actual height of the plant height is measured three times using a tape measure, and the average value of the three times is selected as the actual plant height.

[0119] For the tree stem diameter, first determine the elevation of the measured stem diameter. In this paper, the measurement position is selected at 5 cm above the starting height of the tree trunk growth as the measurement position of the stem diameter. The actual stem diameter of the tree is measured three times using a vernier caliper, and the average value of the three times is selected as the actual stem diameter of the tree.

[0120] To evaluate the accuracy of the tree model, 40 trees of different heights are selected respectively. By measuring the true values of the phenotypic parameters on-site, the simulated values of the phenotypic parameters extracted by using the above method for modeling are compared with the measured values. Using R 2 , MAPE, and RMSE as evaluation indicators to evaluate the accuracy of the tree height. The comparison results are as Figures 9-11 shown.

[0121] Table 1 Accuracy evaluation results of measurement parameters of trees with different heights

[0122]

[0123] As can be seen from the table, the accuracies of the heights and stem diameters of trees with different heights are 95.215%, 95.565%, and 97.955% respectively. R 2 is above 0.9, RMSE is within the accuracy requirement range of 0.1 - 1, and MAPE reaches below 10%. It can be seen that the error between the simulated values obtained by this method and the actually measured values is small, with good consistency, and can relatively accurately measure the phenotypic parameters of trees.

[0124] The tower height is measured by the vertical distance from its highest point to the ground of the tower foundation. The ground point cloud can use the RANSAC algorithm to fit a plane PL = [A, B, C, D], as shown in the figure. Let the point cloud of the tower be H = [h1, h2, hn], (h i =(xh i , yh i , zh i ), i = 1, 2,... n height ), and the pole height can be calculated according to H, where the highest point is P0.

[0125]

[0126] Where: T height is the pole height, n height is the total number of point clouds of the pole tower, and A, B, C, D are the parameters of the fitted ground plane, xh i , yh i , zh i are the coordinates of the i-th point cloud for pole height calculation.

[0127] The cross-arm length can be measured by the diameter of the minimum circumscribed circle of its three-dimensional point cloud. First, fit a plane from the cross-arm point cloud as the measurement reference plane, then project the cross-arm point cloud vertically onto this plane, and fit a minimum circumscribed circle based on the planar point cloud. The diameter of the circumscribed circle is the cross-arm length, as Figure 14 shown.

[0128] Analyze and process the insulator data after completing point cloud preprocessing. Use the MATLAB program to extract structural parameters of the crop, including the structural height Hj, nominal disc diameter Dj, convex hull volume Vc, and concave hull volume Va of the insulator. The definitions and calculation formulas of these parameters are shown in Table 2.

[0129] Table 2 Description of insulator structural parameters

[0130]

[0131] The structural height of the insulator is the vertical height from the top steel cap to the bottom iron foot of the insulator. First, generate a rectangular bounding box for the insulator point cloud through an oriented bounding box (OBB). Consider the direction perpendicular to the cross-arm plane as the axial direction Z of the insulator, and consider the insulator point cloud located in the cross-arm plane as the origin. The rectangular bounding box is as Figure 15 shown.

[0132] Traverse all the point clouds of the insulator, find the maximum and minimum values of Z in the point cloud coordinate system, and the structural height is the difference between the maximum and minimum values of Z in the three-dimensional coordinate system.

[0133] The nominal disc diameter of the insulator is the diameter of the circular part of the insulator. Project the insulator point cloud vertically onto the horizontal plane, and fit a minimum circumscribed circle based on the planar point cloud. The diameter of the circumscribed circle is the nominal disc diameter.

[0134] The convex hull volume is the volume of the smallest convex polyhedron that can enclose the insulator point cloud, and the concave hull volume is the volume of the boundary of the most compact region that encloses the insulator point cloud. The spatial.ConvexHull and spatial.Delaunay in the Scipy library are used to calculate the convex hull and concave hull volumes of the insulator point cloud respectively. Sampling Delaunay triangulation to calculate the concave hull volume can capture the geometry of the point cloud more accurately, which is suitable for high-resolution point cloud data, but the computational complexity is relatively high.

[0135] In addition, a voxel-based method can be used to calculate the insulator volume. This method is easier to handle complex shapes and noise, and the resolution and calculation time can be controlled by adjusting the voxel size. The minimum bounding box of the insulator is divided into N voxels using an octree structure, and the number of voxels Nv occupied by the insulator point cloud is calculated using the point cloud proximity search method of the octree. Then the voxel volume calculation formula of the insulator is as follows:

[0136]

[0137] To verify the effectiveness of the proposed method, in the data collection stage of this project, artificial measurement data of the insulator structure parameters was collected, and the accuracy of the result parameters extracted from the 3D point cloud was evaluated by calculating the relative error between the structure parameters extracted from the 3D point cloud and the artificial measurement results.

[0138] Taking the insulator in the following figure as an example, the algorithm measurement value and the artificial measurement value of the insulator are calculated according to the above method, and their relative error is shown in Table 3.

[0139] Table 3 Comparison of the structural height of the insulator and the artificial measurement height

[0140]

[0141] As can be seen from the above table, the relative error between the insulator point cloud calculation and the artificial measurement height is within 0.13%, and the average relative error is 0.04%. The comparison between the structural height of the insulator extracted from the point cloud and the artificial measurement result is shown in the figure. The error of the structural height is about 6%. The results show that the measurement value of the algorithm in this paper is in good agreement with the artificial measurement value. The 3D reconstruction method of this project can reconstruct a relatively dense, fine and less noisy point cloud. The 3D point cloud segmentation and parameter calculation method has high accuracy and certain robustness, and can be effectively applied to the measurement of the structural parameters of distribution network equipment.

[0142] After segmenting the road surface point cloud by pointnet++, the edge point cloud data on both sides of the road is extracted, and a minimum distance retrieval algorithm is designed to determine the minimum width of the road surface by calculating the shortest distance between the two sides, as Figure 20 shown.

[0143] To calculate the minimum width between the two edges of a road, this project uses a region growing algorithm to retrieve the two points with the minimum distance among the point cloud data on both sides. First, select the starting point of the point cloud on one side as the initial point, and then define the initial point as the center of a sphere to retrieve the minimum distance from this point to the other side, as Figure 21 shown.

[0144] The core objective of calculating the minimum distance between the two edges is to calculate the distance from any point in space to a point cloud set. To achieve this goal, based on the idea of a point cloud denoising algorithm of a region growing (RG) algorithm introduced in the literature, this paper makes improvements and realizes the calculation of the distance from any point in space to a point cloud set. The RG algorithm first selects the point with the highest point cloud density as the initial point, then defines the initial point as the center of a sphere, and dynamically selects a threshold for the diameter of the sphere. Finally, the points exceeding the sphere diameter threshold are removed for noise reduction, as Figure 22 shown.

[0145] Based on the idea of the above algorithm, this paper uses the RG algorithm to calculate the distance from any point p0 in space to the point cloud data set P. The above figure shows the schematic diagram of the RG algorithm for distance calculation. For the to-be-determined point p0, given the sphere diameter growth rate v, the sphere diameter after the k-th iteration is k×v. In each iteration, it is determined whether there is a point inside the sphere. If so, the sphere diameter k×v is determined as the distance from p0 to P.

[0146] However, the calculation accuracy and time of the above algorithm are greatly affected by the sphere diameter growth rate v. When v is large, the calculation time is short, but the calculation accuracy is poor. On the contrary, when v is small, the calculation accuracy is high, but the calculation time is too long. To solve this problem, this paper combines the bisection method with the region growing algorithm, and uses the bisection method to shorten the number of iterations for calculating the sphere diameter, which is called the bisection region growing (BRG) algorithm. The BRG algorithm can ensure high calculation accuracy while greatly improving the calculation efficiency.

[0147] The present invention obtains high-resolution three-dimensional point cloud data of distribution line equipment through a combined acquisition method that combines a visible light imaging device (LiDAR) and a high-precision camera, ensuring that the data has a higher spatial density and geometric integrity. For distribution line equipment in complex environments (such as poles, crossarms, and insulators), the present invention uses the principal component analysis (PCA) algorithm to achieve spatial alignment of the point cloud data, and combines multiple optimization algorithms (such as the RANSAC algorithm for fitting the ground plane, the minimum circumscribed circle algorithm for extracting the geometric diameter, and the convex hull and concave hull algorithms for calculating the volume, etc.), significantly improving the accuracy of equipment parameter extraction. For example, the height of the pole is accurately calculated by the vertical distance between the fitted ground plane and the highest point, the length of the crossarm is obtained through the optimized circumscribed circle algorithm, and the volume of the insulator is calculated by recursively segmenting the high-resolution point cloud. These methods not only reduce the errors in manual measurement but also effectively solve the problem of insufficient measurement accuracy of traditional methods in complex environments.

[0148] The present invention greatly improves the efficiency of distribution line equipment measurement by realizing the automated processing of three-dimensional point cloud data. The preprocessing of point cloud data (such as Gaussian filtering, denoising, and density optimization), spatial alignment (such as the PCA algorithm), and extraction of key parameters (such as pole height, crossarm length, insulator disc diameter, and volume calculation, etc.) are all completed under the support of algorithms without manual intervention. In particular, for parameter extraction of complex-shaped equipment such as insulators, the present invention combines the octree structure and the recursive segmentation algorithm, which not only improves the calculation efficiency but also can quickly complete the calculation of volume and other parameters under high-resolution point clouds. This efficient measurement method significantly reduces the time cost required for equipment inspection and maintenance, and at the same time provides a practical solution for large-scale distribution line equipment detection.

[0149] The three-dimensional point cloud technology of the present invention combined with multiple optimization algorithms makes the equipment parameter measurement more adaptable and robust. Through the combined acquisition of the visible light imaging device and the high-precision camera, high-quality point cloud data can still be obtained even under complex environmental conditions (such as high-altitude operations, tree shading, or areas with complex terrain). In addition, the RANSAC algorithm and region growing algorithm adopted by the present invention have strong anti-noise capabilities, which can eliminate the influence of outliers and environmental noise on data processing. Combining multi-stage fitting methods and optimization strategies, accurate geometric feature extraction can be achieved even if the point cloud data has uneven density or local missing data. Therefore, the present invention is not only applicable to conventional equipment detection scenarios but also can maintain stable performance under special conditions (such as bad weather or complex-structured equipment), providing strong technical support for the measurement of distribution line equipment parameters.

[0150] Only some exemplary embodiments of the present invention have been described by way of illustration. Without doubt, for those of ordinary skill in the art, the described embodiments can be modified in various different ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and description are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A method for calculating distribution line equipment parameters using 3D point clouds, characterized in that, It includes the following steps: Step 1: Use advanced visible light imaging devices or other 3D sensing devices to obtain the point cloud data of the distribution line equipment and its surrounding environment. Obtain the detailed geometric shape and spatial distribution of the distribution line equipment from the point cloud data to provide a high-quality data source for subsequent equipment parameter extraction; Step 2: After obtaining the point cloud data, preprocess it to improve the quality of the point cloud data. In addition, use the principal component analysis method to align the point cloud spatially to ensure that the main axis direction of the point cloud data is consistent with the global coordinate system, providing a standardized reference coordinate system for subsequent parameter calculations; Step 3: Use the RANSAC algorithm to fit the ground point cloud, determine the ground plane parameters, and determine the height of the pole tower based on this plane. The height of the pole tower is achieved by measuring the vertical distance between the highest point of the pole tower point cloud and the ground plane, thereby obtaining accurate pole tower height data; Step 4: Extract the length of the crossarm through the minimum circumscribed circle algorithm. First, fit a reference plane and project the crossarm point cloud onto this plane. Then, calculate the length of the crossarm according to the diameter of the minimum circumscribed circle. Through an optimization algorithm, minimize the projection error to provide accurate crossarm length data; Step 5: Calculate the structural parameters of the insulator through multiple algorithms. Specifically: First, use the oriented bounding box method to determine the direction and position of the insulator point cloud and calculate its structural height; then, project the point cloud onto the horizontal plane and use the minimum circumscribed circle to extract the disk diameter of the insulator; finally, use the convex hull and concave hull algorithms to calculate the volume of the insulator to obtain the geometric shape information of the insulator.

2. The method for calculating the parameters of distribution line equipment using three-dimensional point clouds according to claim 1, wherein The acquisition of the 3D point cloud data in Step 1 is jointly collected by scanning with a visible light imaging device and a high-precision camera to ensure more complete and high-resolution point cloud data is obtained under different environmental conditions. The combined use of the visible light imaging device scanning and the camera can effectively improve the density and accuracy of the point cloud data.

3. A method for calculating distribution line equipment parameters using 3D point clouds according to claim 1, characterized in that, The preprocessing of the point cloud data in Step 2 further includes using the Gaussian filtering algorithm to smooth the point cloud, removing false points, and performing hierarchical processing on the point cloud data through a grid method to improve the data density uniformity and the stability of subsequent processing.

4. A method for calculating distribution line equipment parameters using three-dimensional point clouds according to claim 1, characterized in that The extraction of the pole tower height in Step 3 improves the accuracy through multi-stage processing. The specific steps are as follows: Use the RANSAC algorithm to roughly fit the ground point cloud to obtain the preliminary ground plane parameters; Perform further refined fitting on the pole tower point cloud to eliminate the errors between the ground and the pole tower; Through the maximum vertical distance calculation method, extract the vertical distance between the highest point of the pole tower and the fitted ground plane to obtain the accurate height of the pole tower.

5. A method for calculating distribution line equipment parameters using 3D point clouds according to claim 1, characterized in that The extraction of the crossarm length in Step 4 is achieved through the following detailed steps: Based on the 3D data of the crossarm point cloud, fit a reference plane as the projection reference plane for the crossarm point cloud; Vertically project the crossarm point cloud onto the reference plane and use the minimum circumscribed circle algorithm to extract the diameter of the minimum circumscribed circle of the crossarm point cloud as the length of the crossarm; To eliminate the influence of uneven point cloud density or noise, use the weighted average method to comprehensively process multiple fitting results to optimize the calculation accuracy of the crossarm length.

6. A method for calculating distribution line equipment parameters using three-dimensional point clouds according to claim 1, characterized in that, The extraction of the insulator structure parameters in Step 5 is realized in detail through the following steps: The insulator point cloud is preliminarily enclosed using the oriented bounding box method, and the axial direction of the insulator is extracted and its position is determined; By rotating and projecting the insulator point cloud, a suitable projection plane is selected, and the nominal disc diameter of the insulator is calculated using the minimum circumscribed circle algorithm; Based on the high-density regions of the point cloud, the convex hull volume and the concave hull volume of the insulator are extracted using the convex hull and concave hull algorithms respectively; For high-resolution point clouds, a voxel-based block processing method is adopted to improve the accuracy and robustness of the concave hull volume calculation.

7. A method for calculating distribution line equipment parameters using 3D point clouds according to claim 1, characterized in that, The preprocessing of the point cloud data in Step 2 optimizes the density distribution of the point cloud data by combining multi-scale filtering algorithms. The multi-scale filtering algorithms filter at different scales, dynamically adjust the density of the point cloud, adapt to the requirements of different devices and environments, provide more processing capabilities for high-density point cloud regions, and filter out noise in low-density regions.

8. A method for calculating distribution line equipment parameters using three-dimensional point clouds according to claim 1, characterized in that The extraction of the insulator volume in Step 5 is carried out by recursively dividing the insulator point cloud to ensure the accuracy of the volume calculation. The recursive division method gradually divides the insulator point cloud into multiple voxel blocks, and uses the octree data structure for spatial division and adjacent point search, and finally accurately calculates the insulator volume, reducing errors and improving the calculation efficiency.

Images