Position and direction positioning method of power transmission line tower

Abstract

The invention discloses a method for positioning the position and the trend of a power transmission line tower. Firstly, the method does not depend on prior information, but directly utilizes point cloud data to carry out supervised learning, so that the applicability and flexibility are remarkably improved, the method can adapt to different types of electric power towers and structural changes in an actual environment, the recognition precision is improved, and the automation process of electric power inspection is promoted. And secondly, the three-dimensional point cloud data is directly processed, so that the complex operation of converting the point cloud into a two-dimensional image is avoided, and the information loss is reduced. The point cloud is segmented by adopting the RandLA-Net deep learning network, the point cloud of the electric power tower is accurately extracted, and the diversity characteristics of the electric power tower are effectively identified, so that misjudgment or omission is avoided, the calculation efficiency is improved, and the resource occupation is reduced. And thirdly, principal component analysis and automatic threshold judgment are introduced, so that the overall trend of the electric power tower is accurately identified, the automation level and the real-time adaptability of processing are improved, and the efficiency and reliability of electric power inspection are enhanced.

Description

Technical Field

[0001] The invention relates to the technical field of power transmission line towers, and in particular to a method for locating the position and direction of power transmission line towers. Background Art

[0002] As a basic industry and key energy source of the national economy, the stable and high-quality services of electricity are the basic guarantee for local economic development and residents' daily lives. In recent years, the power industry has developed rapidly, and the number and scale of power grid equipment have continued to expand. Traditional manual inspection methods can no longer meet the growing demand for efficiency and precision in power inspections due to their high labor costs, time-consuming and labor-intensive characteristics, as well as their proneness to errors and limited coverage. At present, the introduction of modern technologies such as drones and lidar has significantly improved the efficiency and quality of power inspections, and created new opportunities for the sustainable development of the digital power industry.

[0003] Drones equipped with LiDAR for power inspections can not only quickly obtain massive point cloud data of transmission line corridors, but also achieve efficient spatial coverage. However, point cloud data of transmission line corridors usually contain a large number of non-focus points and redundant information, making data processing extremely complicated. Therefore, how to efficiently detect power towers in transmission corridors in these huge point cloud data and automatically locate their precise position and direction information has become a key issue that needs to be solved urgently. In terms of positioning the position and direction of transmission line towers based on LiDAR point clouds, there are relatively few methods for positioning the position of transmission line corridor towers based on LiDAR point clouds, and the technology is not yet mature. The existing technology mainly has the following defects:

[0004] 1. Some methods rely on CAD drawings of power tower design and prior information such as preset tower coordinates to extract standard transmission tower structures. This strategy has significant limitations in applicability. CAD drawings are usually based on idealized designs and are difficult to adapt to new towers or structural changes in actual environments. This may lead to a significant reduction in matching accuracy, restricting the automation process of power inspections.

[0005] 2. When existing methods use point cloud two-dimensionalization to generate height difference feature accumulation maps, they use morphological opening operations to extract candidate areas of power towers, which not only leads to misjudgment or omission of important features, but also reduces computational efficiency and occupies a large amount of computing resources. Since morphological operations rely on fixed structural elements and fail to fully consider the diversity of tower shapes and structures, they may lead to misidentification, thus affecting the accuracy of recognition;

[0006] 3. The algorithm processing process relies on manually set thresholds and empirical rules, which further limits the level of automation of power inspection. Moreover, this reliance makes the processing process lack flexibility and cannot adapt to changes in different environments and conditions in real time. Summary of the invention

[0007] The present invention solves the technical problems existing in the prior art by providing a method for locating the position and direction of a transmission line tower.

[0008] The present invention provides a method for locating the position and direction of a transmission line tower, comprising:

[0009] The original disordered point cloud data of the power transmission line is filtered by using progressively encrypted triangulated network filtering to obtain point cloud data of non-ground points.

[0010] Using the RandLA-Net deep learning network to classify the point cloud data of the non-ground points into transmission corridor point cloud data, and extracting all point cloud data belonging to the power towers;

[0011] Clustering and segmenting all the point cloud data belonging to the power poles and towers to generate independent power pole tower point clusters, and obtaining the three-dimensional geometric center coordinates of each independent power pole tower point cluster;

[0012] Extracting the eigenvector of the first principal component from the set of the three-dimensional geometric center coordinates based on principal component analysis PCA, and determining the point cloud direction of the power tower of the transmission line according to the angle between the eigenvector of the first principal component and the X-axis and the Y-axis;

[0013] Performing two-dimensional coordinate axis projection on the three-dimensional geometric center coordinates of each power tower point cluster according to the direction of the power tower point cloud to obtain the two-dimensional projection coordinates of the center point of each power tower point cluster;

[0014] The two-dimensional projection coordinates of the center points of the power tower point clusters are sequentially connected according to the sizes of the two-dimensional projection coordinates of the center points to obtain the geometric topological orientation of the transmission line towers.

[0015] Specifically, before the method uses the progressively encrypted triangulated network filtering to filter the acquired original disordered point cloud data of the power transmission line, it also includes:

[0016] Performing statistical filtering on the acquired original unordered point cloud data of the power transmission line based on the distribution of distances from a point to its neighboring points to remove point cloud noise;

[0017] The method of filtering the acquired original disordered point cloud data of the power transmission line by using the progressively encrypted triangulated network filter comprises:

[0018] The point cloud data after removing the point cloud noise is filtered using the progressively encrypted triangulated network filter.

[0019] Specifically, clustering and segmenting all point cloud data belonging to the power poles and towers to generate independent power pole tower point clusters includes:

[0020] Create a KD-Tree data structure for all the point cloud data belonging to the power pole tower, and initialize an empty clustering set Q and an empty processing queue;

[0021] Using KD-Tree to traverse all point cloud data of the KD-Tree data structure, adding the current point to the processing queue, and marking it as processed;

[0022] For each point in the processing queue, perform a KD-Tree neighborhood search based on Euclidean distance to find all neighboring points whose distance to the current point is less than a preset threshold d, and cluster the neighboring points into the cluster set Q;

[0023] If the number of points in the cluster set Q does not increase any more, the clustering process ends, and the current cluster set Q is regarded as an independent power pole tower point cloud cluster, and then each independent power pole tower point cluster is obtained.

[0024] Specifically, the three-dimensional geometric center coordinates of each independent power tower point cluster are obtained, including:

[0025] For a single power tower point cloud cluster P = {(x i ,y i ,z i )|i=1,2,…,N}, through the formula Calculate the three-dimensional geometric center coordinates (C x ,C y ,C z ), and then obtain the three-dimensional geometric center coordinates of each power tower point cluster; where N is the total number of points in a single power tower point cloud cluster P, (x i ,y i ,z i ) is the three-dimensional coordinate of the midpoint of a single power tower point cloud cluster P.

[0026] Specifically, the extracting the eigenvector of the first principal component from the set of the three-dimensional geometric center coordinates based on principal component analysis PCA, and determining the point cloud direction of the power tower of the transmission line according to the angle between the eigenvector of the first principal component on the X-axis and the Y-axis, includes:

[0027] For the set of three-dimensional geometric center coordinates C = {(C xi ,C yi ,C zi )|i=1,2,…,K}, where K is the number of power towers in the transmission line corridor, and the formula Calculate the mean vector of the three-dimensional geometric center coordinate set Among them, (Cxi ,C yi ,C zi ) are the three-dimensional geometric center coordinates of each power tower point cluster;

[0028] By formula The de-shifted coordinate set C′ is calculated;

[0029] By formula Calculate and obtain the covariance matrix ∑ of the de-shifted coordinate set C′; wherein T represents a transposition operation;

[0030] The eigenvalue λ of the covariance matrix ∑ is calculated by the formula det(∑-λI)=0; wherein det represents the determinant, and I is the unit matrix;

[0031] The corresponding eigenvector v is obtained by solving the equation Σv=λv;

[0032] The eigenvector v1 corresponding to the maximum eigenvalue λ1 is 1x ,v 1y ,v 1z ) as the eigenvector of the first principal component, through the formula Calculate the angle θ between the eigenvector of the first principal component and the X-axis and Y-axis X and θ Y ; where v 1x and v 1y are the components of the eigenvector of the first principal component on the X-axis and the Y-axis respectively;

[0033] If θ X ∈[0,∈], it is determined that the direction of the power tower point cloud of the transmission line is distributed along the X-axis direction; wherein ∈ is the angle determination threshold;

[0034] If θ Y ∈[0,∈], it is determined that the direction of the power tower point cloud of the transmission line is distributed along the Y-axis direction;

[0035] if It is determined that the direction of the power tower point cloud of the transmission line can be distributed along the X-axis direction or the Y-axis direction.

[0036] Specifically, the three-dimensional geometric center coordinates of each power tower point cluster are projected on a two-dimensional coordinate axis according to the direction of the power tower point cloud to obtain the two-dimensional projection coordinates of the center point of each power tower point cluster, including:

[0037] If the direction of the power tower point cloud of the transmission line is distributed along the X-axis direction, the three-dimensional geometric center coordinates are projected onto the X-axis, and the projection coordinates on the X-axis are obtained by setting the Y-axis and Z-axis components to zero;

[0038] If the direction of the power tower point cloud of the transmission line is distributed along the Y-axis direction, the three-dimensional geometric center coordinates are projected to the Y-axis, and the projection coordinates on the Y-axis are obtained by setting the X-axis and Z-axis components to zero.

[0039] Specifically, the two-dimensional projection coordinates of the center points of the power tower point clusters are sequentially connected according to the size of the two-dimensional projection coordinates of the center points to obtain the geometric topological orientation of the transmission line towers, including:

[0040] Sorting the sizes of the two-dimensional projection coordinates of the center points of the power tower point clusters;

[0041] The two-dimensional projection coordinates of the center points of the power tower point clusters are sequentially connected from small to large according to the size of the two-dimensional projection coordinates of the center points, so as to obtain the geometric topological direction of the transmission line tower.

[0042] Specifically, it also includes:

[0043] The three-dimensional geometric center coordinates of each power tower point cluster are converted into the WGS84 longitude and latitude geographic coordinate system, and a vector KML file of the power tower direction of the transmission line is output.

[0044] One or more technical solutions provided in the present invention have at least the following technical effects or advantages:

[0045] The original unordered point cloud data of the transmission line is filtered by progressively encrypted triangulated network filtering to obtain the point cloud data of non-ground points; the point cloud data of non-ground points is classified into transmission corridor point cloud by RandLA-Net deep learning network to extract all point cloud data belonging to power towers; all point cloud data belonging to power towers are clustered and segmented to generate independent power tower point clusters, and the three-dimensional geometric center coordinates of each independent power tower point cluster are obtained; the eigenvector of the first principal component is extracted from the set of three-dimensional geometric center coordinates based on principal component analysis PCA, and the point cloud direction of the power tower of the transmission line is determined according to the angle between the eigenvector of the first principal component on the X-axis and the Y-axis; the three-dimensional geometric center coordinates of each power tower point cluster are projected on the two-dimensional coordinate axis according to the direction of the power tower point cloud to obtain the two-dimensional projection coordinates of the center point of each power tower point cluster; the two-dimensional projection coordinates of each center point are connected in sequence according to the size of the two-dimensional projection coordinates of the center point of each power tower point cluster to obtain the geometric topological direction of the transmission line tower. First, the present invention does not rely on prior information such as CAD drawings and preset coordinates, but directly uses the laser radar point cloud data of the transmission line corridor for supervised learning. This method significantly improves the applicability and flexibility, and can adapt to different types of power towers and structural changes in the actual environment, thereby improving the recognition accuracy and promoting the automation process of power inspection. Secondly, the present invention directly processes three-dimensional point cloud data, avoids the complex operation of converting point clouds into two-dimensional images, and reduces information loss. The RandLA-Net deep learning network based on deep learning is used to segment the original disordered transmission corridor point cloud, accurately extract the power tower point cloud, effectively identify and extract the diversity characteristics of the power tower, avoid relying on prior design information, not only avoid misjudgment or omission, enhance the adaptability of the algorithm, but also improve the calculation efficiency and reduce resource occupation. Thirdly, by introducing principal component analysis PCA and automated threshold judgment, not only the overall direction of the power tower is accurately identified, but also the automation level and real-time adaptability of the processing are improved, so that the algorithm can dynamically respond to changes in different environments and conditions, enhance the efficiency and reliability of power inspection, and significantly improve the automation level. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 A flow chart of a method for locating the position and direction of a transmission line tower provided by an embodiment of the present invention;

[0047] Figure 2 A schematic diagram of a method for locating the position and orientation of a transmission line tower provided by an embodiment of the present invention;

[0048] Figure 3 The original transmission line corridor point cloud data collected in the embodiment of the present invention;

[0049] Figure 4 Statistical filtering results of the transmission line corridor point cloud and ground point filtering results in an embodiment of the present invention (grey represents non-ground points, and brown represents ground points);

[0050] Figure 5 The result of using the RandLA-Net deep learning network to classify the transmission line corridor point cloud in an embodiment of the present invention;

[0051] Figure 6 The result of performing point cloud cluster segmentation of power towers based on the three-dimensional Euclidean clustering algorithm accelerated by KD-Tree in an embodiment of the present invention;

[0052] Figure 7 It is a vector KML file of the transmission line tower orientation output in the embodiment of the present invention. DETAILED DESCRIPTION

[0053] The embodiment of the present invention solves the technical problems existing in the prior art by providing a method for locating the position and direction of a transmission line tower.

[0054] In order to better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0055] like Figure 1 and Figure 2 As shown, the method for locating the position and direction of a transmission line tower provided by an embodiment of the present invention includes:

[0056] Step S110: using progressively encrypted triangulated network filtering to filter the acquired original disordered point cloud data of the power transmission line to obtain point cloud data of non-ground points;

[0057] In order to remove the outlier noise points in the original point cloud data, before using the progressive encryption triangulation filter to filter the original disordered point cloud data of the transmission line, it also includes:

[0058] The original disordered point cloud data of the power transmission line is statistically filtered based on the distance distribution from the point to its neighboring points to remove the point cloud noise;

[0059] In this case, the original disordered point cloud data of the power transmission line is filtered using progressively encrypted triangulated network filtering, including:

[0060] The point cloud data after removing the point cloud noise is filtered using progressively encrypted triangulated network filtering.

[0061] Step S120: using the RandLA-Net deep learning network to classify the point cloud data of non-ground points into transmission corridor point cloud data, and extracting all point cloud data belonging to power towers;

[0062] Step S130: performing clustering and segmentation on all point cloud data belonging to the power poles and towers, generating independent power pole tower point clusters, and obtaining the three-dimensional geometric center coordinates of each independent power pole tower point cluster;

[0063] Specifically, all point cloud data belonging to power towers are clustered and segmented to generate independent power tower point clusters, including:

[0064] A KD-Tree data structure is created for all point cloud data belonging to power poles and towers, and an empty clustering set Q and an empty processing queue are initialized; wherein the clustering set Q is used to store the identified clustering results, and the processing queue is used to store the points to be analyzed in the clustering process.

[0065] Use KD-Tree to traverse all point cloud data in the KD-Tree data structure, add the current point to the processing queue, and mark it as processed;

[0066] For each point in the processing queue, perform a KD-Tree neighborhood search based on Euclidean distance to find all neighboring points whose distance to the current point is less than the preset threshold d, and cluster the neighboring points into a cluster set Q;

[0067] If the number of points in the cluster set Q does not increase any more, the clustering process ends, and the current cluster set Q is regarded as an independent power pole tower point cloud cluster, and then each independent power pole tower point cluster is obtained.

[0068] Get the three-dimensional geometric center coordinates of each independent power tower point cluster, including:

[0069] For a single power tower point cloud cluster P = {(x i ,y i , z i )|i=1,2,...,N}, by the formula Calculate the three-dimensional geometric center coordinates (C x , C y , C z ), and then obtain the three-dimensional geometric center coordinates of each power tower point cluster; where N is the total number of points in a single power tower point cloud cluster P, (x i ,y i , z i ) is the three-dimensional coordinate of the midpoint of a single power tower point cloud cluster P.

[0070] Step S140: extracting the eigenvector of the first principal component from the set of three-dimensional geometric center coordinates based on principal component analysis PCA, and determining the point cloud direction of the power tower of the transmission line according to the angle between the eigenvector of the first principal component on the X-axis and the Y-axis;

[0071] This step is specifically described. Based on the principal component analysis PCA, the eigenvector of the first principal component is extracted from the set of three-dimensional geometric center coordinates. The direction of the power tower point cloud of the transmission line is determined according to the angle between the eigenvector of the first principal component on the X-axis and the Y-axis, including:

[0072] For the set of three-dimensional geometric center coordinates C = {(C xi , C yi , C zi )|i=1,2,...,K}, where K is the number of power towers in the transmission line corridor, and the formula Calculate the mean vector of the three-dimensional geometric center coordinate set Among them, (C xi , C yi , C zi ) are the three-dimensional geometric center coordinates of each power tower point cluster;

[0073] By formula The de-shifted coordinate set C′ is calculated;

[0074] By formula The covariance matrix ∑ of the de-shifted coordinate set C′ is calculated, where T represents a transpose operation; the covariance matrix is ​​a 3×3 symmetric matrix, which represents the covariance relationship between the coordinate components.

[0075] The eigenvalue λ of the covariance matrix ∑ is calculated by the formula det(∑-λI)=0; wherein det represents the determinant and I is the unit matrix;

[0076] The corresponding eigenvector v is obtained by solving the equation ∑v=λv; wherein the eigenvalue λ can describe the amount of expansion and contraction of the covariance matrix ∑ in a specific direction, and the eigenvector v describes this direction.

[0077] The calculated eigenvalues ​​are sorted according to their sizes: the first principal component direction v1 corresponds to the maximum eigenvalue λ1, the second principal component direction v2 corresponds to the second largest eigenvalue λ2, and the third principal component direction v3 corresponds to the minimum eigenvalue λ3.

[0078] The eigenvector v1 corresponding to the maximum eigenvalue λ1 is 1x , v 1y , v 1z ) as the eigenvector of the first principal component, through the formula Calculate the angle θ between the eigenvector of the first principal component and the X-axis and Y-axis X and θ Y , and satisfies Among them, v1x and v 1y are the components of the eigenvector of the first principal component on the X-axis and Y-axis respectively;

[0079] If θ X ∈[0,∈], it is determined that the direction of the power tower point cloud of the transmission line is distributed along the X-axis direction, that is, horizontal distribution; where ∈ is the angle determination threshold;

[0080] If θ Y ∈[0,∈], it is determined that the direction of the power tower point cloud of the transmission line is distributed along the Y-axis direction, that is, longitudinal distribution;

[0081] if This indicates that the power tower point cloud of the transmission line has no obvious bias towards the X-axis direction or the Y-axis direction, but is evenly and symmetrically extended on these two axes. Due to the symmetry of the distribution, it is determined that the power tower point cloud of the transmission line can be distributed along the X-axis direction or the Y-axis direction, and is preferentially determined to be distributed along the X-axis direction.

[0082] Step S150: performing two-dimensional coordinate axis projection on the three-dimensional geometric center coordinates of each power tower point cluster according to the direction of the power tower point cloud, to obtain the two-dimensional projection coordinates of the center point of each power tower point cluster;

[0083] This step is specifically described. According to the direction of the power tower point cloud, the three-dimensional geometric center coordinates of each power tower point cluster are projected on the two-dimensional coordinate axis to obtain the two-dimensional projection coordinates of the center point of each power tower point cluster, including:

[0084] If the direction of the power tower point cloud of the transmission line is distributed along the X-axis, the three-dimensional geometric center coordinates are projected onto the X-axis, and the projection coordinates on the X-axis are obtained by setting the Y-axis and Z-axis components to zero;

[0085] If the direction of the power tower point cloud of the transmission line is distributed along the Y-axis direction, the three-dimensional geometric center coordinates are projected to the Y-axis, and the projection coordinates on the Y-axis are obtained by setting the X-axis and Z-axis components to zero.

[0086] Step S160: connecting the two-dimensional projection coordinates of the center points of each power tower point cluster in sequence according to the size of the two-dimensional projection coordinates of the center points to obtain the geometric topological orientation of the transmission line tower.

[0087] This step is specifically described. According to the size of the two-dimensional projection coordinates of the center points of each power tower point cluster, the two-dimensional projection coordinates of each center point are sequentially connected to obtain the geometric topological direction of the transmission line tower, including:

[0088] Sort the sizes of the two-dimensional projection coordinates of the center points of each power tower point cluster;

[0089] According to the size of the two-dimensional projection coordinates of the center points of each power tower point cluster, the two-dimensional projection coordinates of each center point are connected in sequence from small to large to obtain the geometric topological direction of the transmission line tower.

[0090] The embodiment of the present invention is further described, and further includes:

[0091] The three-dimensional geometric center coordinates of each power tower point cluster are converted into the WGS84 longitude and latitude geographic coordinate system, and the vector KML file of the power tower direction of the transmission line is output.

[0092] In order to make the purpose, technical solutions and advantages of the present invention clearer, the laser radar point cloud data obtained by the UAV platform is taken as the actual research object, and the present invention is further described in detail with reference to the drawings in this specification.

[0093] Embodiment 1: Figure 3 The point cloud data is obtained by using a DJI M350 RTK drone equipped with a DJI Zenmuse L2 laser radar. The common raw point cloud data of the power corridor of the transmission line used in the embodiment of the present invention is mainly for the 500KV high-voltage AC transmission line corridor scene. The types of towers in the selected transmission line power corridor scene include straight towers, tension towers and corner towers.

[0094] Step 1: Perform statistical filtering on the originally acquired disordered LiDAR point cloud based on the distance distribution from the point to its neighboring points to remove outlier noise points in the original point cloud data and ensure that the proportion of noise points is controlled below 3%.

[0095] Step 2: The point cloud filtering method based on the progressively encrypted triangulated network is used for the denoised transmission corridor point cloud to obtain the point cloud of non-ground points such as power towers in the transmission line corridor, such as Figure 4 shown.

[0096] Step 3: Use the RandLA-Net deep learning network to classify the non-ground point cloud of the transmission line corridor, and obtain all point clouds belonging to the power tower category in the transmission line corridor. The classification accuracy rate reaches more than 90%, such as Figure 5 shown.

[0097] Step 4: Use the three-dimensional Euclidean clustering method based on KD-Tree acceleration to cluster and segment all point cloud data belonging to the power tower category in the transmission line corridor.

[0098] Specifically, a KD-Tree data structure is created for all point cloud datasets belonging to the power tower category in the transmission line corridor to improve the efficiency of point cloud neighborhood search. An empty clustering set Q and processing queue are initialized.

[0099] Use KD-Tree to traverse each point in the point cloud dataset, add the current point to the processing queue, and mark it as processed.

[0100] For each point in the processing queue, a KD-Tree neighborhood search based on Euclidean distance is performed to find all neighboring points whose distance to the current point is less than a preset threshold d.

[0101] Furthermore, the current cluster set Q is regarded as a complete power tower point cloud cluster, and the number of cluster points should reach at least 1000 points, such as Figure 6 shown.

[0102] Step 5: Extract the three-dimensional geometric center point coordinates of the single power tower point cloud cluster after clustering segmentation.

[0103] Step 6: Determine the point cloud direction (including lateral distribution and longitudinal distribution) of the power tower of the transmission line based on principal component analysis PCA, and perform two-dimensional coordinate axis (i.e., X-axis or Y-axis on the XOY plane) projection processing on the three-dimensional geometric center point coordinates of each power tower point cluster.

[0104] Specifically, principal component analysis (PCA) is applied to the three-dimensional geometric center point coordinate set of the power tower in the transmission corridor to calculate the eigenvalues ​​and corresponding eigenvectors of the de-shifted covariance matrix of the three-dimensional geometric center point coordinates of the power tower.

[0105] According to the PCA analysis results, the angle between the X-axis and Y-axis of the eigenvector of the first principal component eigenvalue is calculated, and the overall direction of the point cloud is determined by setting a threshold. Two-dimensional coordinate projection is performed according to the determined distribution direction.

[0106] Step 7: Sort the projection coordinates of the center points of the power tower point cluster according to the projection coordinate values ​​on the projection two-dimensional coordinate axis.

[0107] Step 8: According to the sorted distribution of power poles and towers, connect them in sequence to infer the geometric topological direction of the transmission line towers.

[0108] Step 9: Convert the three-dimensional geometric center point coordinates of the towers in the transmission line corridor into the WGS84 longitude and latitude geographic coordinate system, and output the vector KML file of the transmission line tower direction with an accuracy of ±1 meter. Figure 7 shown.

[0109] The embodiment of the present invention provides a method for locating the position and direction of a transmission line tower based on a laser radar point cloud, which significantly improves the accuracy and efficiency of data processing. Compared with the existing technical methods, statistical filtering and deep learning classification are used to effectively reduce the interference of outlier noise. At the same time, a three-dimensional Euclidean clustering method based on KD-Tree is adopted to accelerate the clustering and segmentation of point clouds. Combined with principal component analysis PCA and automated threshold judgment, the overall direction of the power tower is accurately identified, and the coordinate projection and sorting process is simplified. Finally, the three-dimensional geometric center coordinates are successfully converted into longitude and latitude coordinates, and a standardized vector KML file is generated, which provides strong technical support for the application of geographic information system GIS, and enhances the process of intelligent operation and maintenance of transmission lines, which is of great significance to promoting the development of drone power inspection technology.

[0110] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0111] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0112] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0113] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0114] The parts not described in detail in the embodiments of the present invention are all known technologies to those skilled in the art. Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention is described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solutions of the present invention, which should be included in the scope of the claims of the present invention.

Claims

1. A method for locating the position and direction of a transmission line tower, characterized in that: include: The original disordered point cloud data of the power transmission line is filtered by using progressively encrypted triangulated network filtering to obtain point cloud data of non-ground points. Using the RandLA-Net deep learning network to classify the point cloud data of the non-ground points into transmission corridor point cloud data, and extracting all point cloud data belonging to the power towers; Clustering and segmenting all the point cloud data belonging to the power poles and towers to generate independent power pole tower point clusters, and obtaining the three-dimensional geometric center coordinates of each independent power pole tower point cluster; Extracting the eigenvector of the first principal component from the set of the three-dimensional geometric center coordinates based on principal component analysis PCA, and determining the point cloud direction of the power tower of the transmission line according to the angle between the eigenvector of the first principal component and the X-axis and the Y-axis; Performing two-dimensional coordinate axis projection on the three-dimensional geometric center coordinates of each power tower point cluster according to the direction of the power tower point cloud to obtain the two-dimensional projection coordinates of the center point of each power tower point cluster; The two-dimensional projection coordinates of the center points of the power tower point clusters are sequentially connected according to the sizes of the two-dimensional projection coordinates of the center points to obtain the geometric topological orientation of the transmission line towers.

2. The method for locating the position and direction of a transmission line tower according to claim 1, characterized in that: Before the method uses the progressively encrypted triangulated network filtering to filter the acquired original disordered point cloud data of the power transmission line, it also includes: Performing statistical filtering on the acquired original unordered point cloud data of the power transmission line based on the distribution of distances from a point to its neighboring points to remove point cloud noise; The method of filtering the acquired original disordered point cloud data of the power transmission line by using the progressively encrypted triangulated network filter comprises: The point cloud data after removing the point cloud noise is filtered using the progressively encrypted triangulated network filter.

3. The method for locating the position and direction of a transmission line tower according to claim 1, characterized in that: The step of clustering and segmenting all the point cloud data belonging to the power poles and towers to generate independent power pole tower point clusters includes: Create a KD-Tree data structure for all the point cloud data belonging to the power pole tower, and initialize an empty clustering set Q and an empty processing queue; Using KD-Tree to traverse all point cloud data of the KD-Tree data structure, adding the current point to the processing queue, and marking it as processed; For each point in the processing queue, perform a KD-Tree neighborhood search based on Euclidean distance to find all neighboring points whose distance to the current point is less than a preset threshold d, and cluster the neighboring points into the cluster set Q; If the number of points in the cluster set Q does not increase any more, the clustering process ends, and the current cluster set Q is regarded as an independent power pole tower point cloud cluster, and then each independent power pole tower point cluster is obtained.

4. The method for locating the position and direction of a transmission line tower according to claim 1, characterized in that: The method of obtaining the three-dimensional geometric center coordinates of each independent power tower point cluster includes: For a single power tower point cloud cluster P = {(x i ,y i ,z i )|i=1,2,…,N}, through the formula Calculate the three-dimensional geometric center coordinates (C x ,C y ,C z ), and then obtain the three-dimensional geometric center coordinates of each power tower point cluster; where N is the total number of points in a single power tower point cloud cluster P, (x i ,y i ,z i ) is the three-dimensional coordinate of the midpoint of a single power tower point cloud cluster P.

5. The method for locating the position and direction of a transmission line tower according to claim 1, characterized in that: The method of extracting the eigenvector of the first principal component from the set of the three-dimensional geometric center coordinates based on principal component analysis PCA, and determining the point cloud direction of the power tower of the transmission line according to the angle between the eigenvector of the first principal component on the X-axis and the Y-axis, includes: For the set of three-dimensional geometric center coordinates C = {(C xi ,C yi ,C zi )|i=1,2,…,K}, where K is the number of power towers in the transmission line corridor, and the formula Calculate the mean vector of the three-dimensional geometric center coordinate set Among them, (C xi ,C yi ,C zi ) are the three-dimensional geometric center coordinates of each power tower point cluster; By formula The de-shifted coordinate set C′ is calculated; By formula Calculate and obtain the covariance matrix ∑ of the de-shifted coordinate set C′; wherein t represents a transposition operation; The eigenvalue λ of the covariance matrix Σ is calculated by the formula det(Σ-λI)=0; wherein det represents the determinant, and I is the unit matrix; The corresponding eigenvector v is obtained by solving the equation Σv=λv; The eigenvector v1 corresponding to the maximum eigenvalue λ1 is 1x ,v 1y ,v 1z ) as the eigenvector of the first principal component, through the formula Calculate the angle θ between the eigenvector of the first principal component and the X-axis and Y-axis X and θ Y ; where v 1x and v 1y are the components of the eigenvector of the first principal component on the X-axis and the Y-axis respectively; If θ X ∈[0,∈], it is determined that the direction of the power tower point cloud of the transmission line is distributed along the X-axis direction; wherein v is the angle determination threshold; If θ Y ∈[0,∈], it is determined that the direction of the power tower point cloud of the transmission line is distributed along the Y-axis direction; if It is determined that the direction of the power tower point cloud of the transmission line can be distributed along the X-axis direction or the Y-axis direction.

6. The method for locating the position and direction of a transmission line tower according to claim 5, characterized in that: The two-dimensional coordinate axis projection is performed on the three-dimensional geometric center coordinates of each power tower point cluster according to the direction of the power tower point cloud to obtain the two-dimensional projection coordinates of the center point of each power tower point cluster, including: If the direction of the power tower point cloud of the transmission line is distributed along the X-axis direction, the three-dimensional geometric center coordinates are projected onto the X-axis, and the projection coordinates on the X-axis are obtained by setting the Y-axis and Z-axis components to zero; If the direction of the power tower point cloud of the transmission line is distributed along the Y-axis direction, the three-dimensional geometric center coordinates are projected to the Y-axis, and the projection coordinates on the Y-axis are obtained by setting the X-axis and Z-axis components to zero.

7. The method for locating the position and direction of a transmission line tower according to claim 1, characterized in that: The step of sequentially connecting the two-dimensional projection coordinates of the center points of the power tower point clusters according to the size of the two-dimensional projection coordinates to obtain the geometric topological orientation of the transmission line towers includes: Sorting the sizes of the two-dimensional projection coordinates of the center points of the power tower point clusters; The two-dimensional projection coordinates of the center points of the power tower point clusters are sequentially connected from small to large according to the size of the two-dimensional projection coordinates of the center points, so as to obtain the geometric topological direction of the transmission line tower.

8. The method for locating the position and direction of a transmission line tower according to any one of claims 1 to 7, characterized in that: Also includes: The three-dimensional geometric center coordinates of each power tower point cluster are converted into the WGS84 longitude and latitude geographic coordinate system, and a vector KML file of the power tower direction of the transmission line is output.

Images