Line tree obstacle risk hidden danger analysis system and method based on laser radar point cloud

Abstract

The invention discloses a line tree obstacle risk hidden danger analysis system and method based on a laser radar point cloud, and relates to the field of power transmission line safety inspection, and the key points of the technical scheme are that the method comprises the steps: obtaining point cloud data of a power transmission line corridor region along a power transmission line corridor; extracting conductor point cloud data of the power transmission line; performing ground point segmentation of an RANSAC algorithm on the point cloud data and performing clustering analysis and identification on a suspension point to extract the conductor point cloud data of the power transmission line; extracting offline ground feature data of the power transmission line; using a random forest classifier to identify and classify ground feature point cloud data including the ground, trees and towers in a corridor of the power transmission line; calculating the distance between the conductor and the ground object based on the conductor point cloud data and the ground object point cloud data, evaluating the potential risk, and outputting a potential risk point; and screening potential risk points of the trees to the power transmission line, determining that the potential risk points of the trees to the power transmission line are output as target risk points, and outputting a risk assessment report based on the target risk points.

Description

Technical Field

[0001] The present invention relates to the field of power transmission line safety inspection, and more specifically, to a system and method for analyzing tree obstacle risk hazards of power transmission lines based on laser radar point cloud. Background Art

[0002] In the existing technology, machine learning algorithms are often used to automatically separate power lines, poles, trees, poles and other categories. With the key objects in the power corridor - power lines and power towers as the core, a columnar exploration space is formed to calculate the current distance between the trees and the power lines, and quickly obtain the location information of tree obstacle hazards.

[0003] However, the study found that the extraction of power line data and tree point clouds, as well as the collection of the main factors affecting the growth of tree species in the study area and the development of a tree growth prediction model database for transmission line channels are all very complex, resulting in a large amount of data processing, and the inability to efficiently complete tree hazard analysis and early warning processing. Summary of the invention

[0004] The purpose of the present invention is to provide a system and method for analyzing line tree obstacle risk hazards based on laser radar point cloud, so as to solve the problem of low execution efficiency of tree hazard analysis in the prior art.

[0005] In a first aspect of the present application, a system for analyzing the risk of tree obstacles on a line based on a laser radar point cloud is provided, the system comprising:

[0006] A point cloud acquisition module is used to acquire point cloud data along the transmission line corridor in the transmission line corridor area;

[0007] A cluster processing module is used to extract the conductor point cloud data of the transmission line: the point cloud data is segmented by the RANSAC algorithm for ground points and clustered for the suspended points for identification, so as to extract the conductor point cloud data of the transmission line;

[0008] The extraction module is used to extract offline ground object data of the transmission line: using the random forest classifier to identify and classify the ground object point cloud data including the ground, trees, and towers in the transmission line corridor;

[0009] A calculation and evaluation module is used to calculate the distance between the conductor and the ground object based on the conductor point cloud data and the ground object point cloud data, evaluate the potential risk and output the potential risk point;

[0010] The report output module is used to screen potential risk points of trees to transmission lines, determine the potential risk points of trees to transmission lines and output them as target risk points, and output a risk assessment report based on the target risk points.

[0011] In one implementation, the cluster processing module specifically includes:

[0012] The ground point and non-ground point segmentation module is used to segment the point cloud data using the RANSAC algorithm: identify and extract ground points and non-ground points;

[0013] A suspended point extraction module is used to separate the point cloud located above the ground through a height threshold method and determine it as a suspended point to extract the suspended point;

[0014] A point cluster identification module is used to apply a density-based spatial clustering algorithm to the extracted dangling points to identify the initial wire point clusters;

[0015] The wire fitting module is used to fit the wire point cluster using the wire catenary equation: the wire catenary equation is applied to each identified point cluster for fitting to obtain the wire curve equation;

[0016] The verification output module is used to use the fitted conductor curve equation to perform secondary extraction of conductor points, and then verify the positions of multiple conductor points currently extracted for the second time. If the verification result of the conductor point is that the proportion of the correct conductor point cluster is higher than the standard proportion threshold, the conductor point output on the current conductor curve equation is determined to be the target conductor point cluster, and the target conductor point cluster is fitted and connected to form the conductor point cloud data of the transmission line.

[0017] In one implementation, the density-based spatial clustering algorithm is a DBSCAN clustering algorithm.

[0018] In one implementation, the catenary equation of the conductor is: y=a*cosh((xb) / a)+c; wherein y is the vertical coordinate of the conductor, x is the horizontal coordinate of the conductor, a is the catenary parameter, b is the horizontal coordinate of the axis of symmetry of the curve, and c is the vertical coordinate of the lowest point.

[0019] In one implementation, the extraction module includes:

[0020] A digital elevation model generation module is used to generate a digital elevation model using segmented ground points through a triangular irregular network interpolation method;

[0021] The non-ground classification module is used to classify non-ground points: the random forest classifier is used to classify non-ground points and identify ground objects such as towers and trees;

[0022] The tree extraction module is used to extract tree point clouds: separate tree point clouds from tree categories;

[0023] A single tree segmentation module, used for performing single tree segmentation: performing single tree segmentation on the tree point cloud;

[0024] The single tree attribute information extraction module is used to extract the attribute information of each tree based on the single tree segmentation result of the tree point cloud; wherein the attribute information of each tree includes the position, height, crown width and breast diameter of each tree.

[0025] In one implementation, the single wood segmentation module specifically includes:

[0026] Single tree slicing module, used to slice the tree point cloud horizontally: slice the tree point cloud at equal intervals in the vertical direction;

[0027] Single tree outline recognition module, which is used to apply the convex hull algorithm within each slice: identify the tree outline in each slice;

[0028] The module for calculating inter-layer connectivity of single tree contours is used to determine the vertical structure of trees by analyzing the overlap of contours between adjacent slices using inter-layer connectivity analysis;

[0029] The single tree growth module is used to apply the region growing algorithm: taking the crown vertex as the seed point and growing downward to achieve complete single tree segmentation.

[0030] In one implementation, the calculation and evaluation module specifically includes:

[0031] The spatial index module is used to establish a KD tree index for the wire point cloud data, so as to facilitate the search for other wire points for the nearest neighbors of the current wire point; and to establish an R tree index for the tree point cloud data, so as to facilitate the search for trees in a specific target area;

[0032] The threshold acquisition module is used to determine and set the corresponding safety distance threshold according to the voltage level type of the current conductor;

[0033] The minimum distance calculation module is used to traverse the point cloud data of each tree; for each tree, use the KD tree to find the nearest wire point; calculate the Euclidean distance between the tree point cloud and the nearest wire point; record the minimum Euclidean distance between each tree point cloud and the wire where each tree is located as the minimum distance;

[0034] The risk assessment module is used to compare the calculated minimum distance with the safety distance threshold; if the minimum distance is less than the safety distance threshold, the tree is marked as a potential risk point.

[0035] In one implementation, the calculation and evaluation module further includes a recording module:

[0036] The recording module is used to import the wire point cloud data and the ground object point cloud data, and load the attribute information of each tree.

[0037] In one implementation, the system also includes a risk level grading module; the risk level grading module is used to calculate the risk level; and to grade potential risk points according to the risk level; wherein the potential risk points include high, medium, and low risk levels.

[0038] In a second aspect of the present application, a method for analyzing the risk of tree obstacles on a line based on a laser radar point cloud is provided, the method comprising:

[0039] Obtain point cloud data along the transmission line corridor in the transmission line corridor area;

[0040] Extracting the conductor point cloud data of the transmission line: performing ground point segmentation of the point cloud data using the RANSAC algorithm and clustering analysis and identification of the suspended points to extract the conductor point cloud data of the transmission line;

[0041] Extract offline feature data of transmission lines: Use random forest classifier to identify and classify feature point cloud data including ground, trees, and towers within the transmission line corridor;

[0042] Calculate the distance between the wire and the ground object based on the wire point cloud data and the ground object point cloud data, evaluate the potential risk and output the potential risk points;

[0043] Screen potential risk points of trees to transmission lines, determine the potential risk points of trees to transmission lines and output them as target risk points, and output a risk assessment report based on the target risk points.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] The present invention uses a random sampling consensus algorithm (RANSAC) to segment ground points, helping to identify and remove noise and unnecessary points in the data. The present invention also uses a random forest classifier in machine learning to identify and classify point cloud data, and extracts ground, tree, tower and other ground object data within the transmission line corridor. This step can improve the accuracy and efficiency of ground object identification. Then, the distance between the conductor and the surrounding ground objects is calculated based on the conductor point cloud data and the ground object point cloud data. This calculation can help identify potential risk points, such as when trees are too close to transmission lines. Trees that pose potential risks to transmission lines are screened out, and these trees are identified as target risk points. Finally, a detailed risk assessment report is generated based on the target risk points to provide data support for guiding subsequent management and maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, constitute a part of this application, and do not constitute a limitation of the embodiments of the present invention. In the drawings:

[0047] Figure 1A schematic diagram of the architecture principle of a system for analyzing tree obstacle risks on a line based on a laser radar point cloud provided in an embodiment of the present invention;

[0048] Figure 2 A schematic flow chart of a method for analyzing tree obstacle risks on a line based on a laser radar point cloud provided in an embodiment of the present invention;

[0049] Figure 3 A schematic diagram of the operation steps of extracting conductor point cloud data of a transmission line in a method for analyzing tree obstacle risk hazards of a line based on a laser radar point cloud provided by an embodiment of the present invention;

[0050] Figure 4 A schematic diagram of the operation steps of using a random forest classifier to identify and classify point cloud data of ground objects including ground, trees, and towers in a transmission line corridor in a method for analyzing tree obstacle risks based on laser radar point clouds provided by an embodiment of the present invention;

[0051] Figure 5 A schematic diagram of the operation steps of single tree segmentation in a method for analyzing the risk of tree obstacles on a line based on a laser radar point cloud provided in an embodiment of the present invention;

[0052] Figure 6 A schematic diagram of the operation steps of evaluating potential risks and outputting potential risk points in a method for analyzing potential risks of tree obstacles on a line based on a laser radar point cloud provided in an embodiment of the present invention;

[0053] Figure 7 A schematic diagram of the operational steps for outputting a warning message at a future time point in a method for analyzing line tree obstacle risks based on a laser radar point cloud provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0054] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with embodiments and drawings. The exemplary embodiments of the present invention and their description are only used to explain the present invention and are not intended to limit the present invention.

[0055] It should be noted that the terms "include" or "may include" used in various embodiments of the present application indicate the presence of the function, operation or element applied for, and do not limit the addition of one or more functions, operations or elements. In addition, as used in various embodiments of the present application, the terms "include", "have" and their cognates are only intended to indicate specific features, numbers, steps, operations, elements, components or a combination of the foregoing items, and should not be understood as first excluding the presence of one or more other features, numbers, steps, operations, elements, components or a combination of the foregoing items or the possibility of adding one or more features, numbers, steps, operations, elements, components or a combination of the foregoing items.

[0056] like Figure 1 As shown, an embodiment of the present invention provides a system for analyzing the risk of tree obstacles on a line based on a laser radar point cloud, and the system includes:

[0057] The point cloud acquisition module 10 is used to acquire point cloud data along the transmission line corridor in the transmission line corridor area;

[0058] The cluster processing module 20 is used to extract the conductor point cloud data of the power transmission line: the ground point segmentation of the point cloud data by the RANSAC algorithm and the cluster analysis and identification of the suspended points are performed to extract the conductor point cloud data of the power transmission line;

[0059] The extraction module 30 is used to extract offline ground object data of the transmission line: using a random forest classifier to identify and classify ground object point cloud data including ground, trees, and towers in the transmission line corridor;

[0060] A calculation and evaluation module 40 is used to calculate the distance between the conductor and the ground object based on the conductor point cloud data and the ground object point cloud data, evaluate the potential risk and output the potential risk point;

[0061] The report output module 50 is used to screen potential risk points of trees to transmission lines, determine the potential risk points of trees to transmission lines and output them as target risk points, and output a risk assessment report based on the target risk points.

[0062] Calculate the distance between the conductor and the ground object based on the conductor point cloud data and the ground object point cloud data, evaluate the potential risk and output the potential risk point; screen the potential risk points of the trees to the transmission line, determine the potential risk points of the trees to the transmission line and output them as target risk points, and then output a final risk assessment report based on the target risk points;

[0063] The random sampling consensus algorithm (RANSAC) is used to segment ground points to help identify and remove noise and unnecessary points in the data. The present invention also uses the random forest classifier in machine learning to identify and classify point cloud data, and extracts ground, trees, towers and other ground object data within the transmission line corridor. This step can improve the accuracy and efficiency of ground object identification. Then, the distance between the conductor and the surrounding ground objects is calculated based on the conductor point cloud data and the ground object point cloud data. This calculation can help identify potential risk points, such as when trees are too close to transmission lines. Trees that pose potential risks to transmission lines are screened out, and these trees are identified as target risk points. Finally, a detailed risk assessment report is generated based on the target risk points to provide data support for guiding subsequent management and maintenance.

[0064] Preferably, as an implementable method, the cluster processing module performs "segmentation of ground points of the point cloud data by the RANSAC algorithm and cluster analysis and identification of suspended points to extract the conductor point cloud data of the transmission line", and specifically includes a denoising module, a ground point and non-ground point segmentation module, a suspended point extraction module, a point cluster identification module, a conductor fitting module, and a verification output module:

[0065] The denoising module is used to perform spatial filtering on the original point cloud data: using a statistical outlier filtering algorithm to remove noise points and improve data quality;

[0066] The ground point and non-ground point segmentation module is used to perform ground point segmentation on the point cloud data using the RANSAC algorithm: identifying and extracting ground points and non-ground points;

[0067] The suspended point extraction module is used to separate the point cloud located above the ground by using a height threshold method and determine it as a suspended point to extract the suspended point;

[0068] The point cluster identification module is used to apply a density-based spatial clustering algorithm (such as DBSCAN) to the extracted dangling points to identify initial wire point clusters;

[0069] The conductor fitting module is used to fit the conductor point clusters using the conductor catenary equation: the conductor catenary equation is applied to each identified point cluster for fitting to obtain the conductor curve equation;

[0070] The verification output module is used to use the fitted conductor curve equation to perform secondary extraction of conductor points, and then verify the positions of multiple conductor points currently extracted for the second time. If the verification result of the conductor point is that the proportion of the correct conductor point cluster is higher than the standard proportion threshold, the conductor point output on the current conductor curve equation is determined to be the target conductor point cluster, and the target conductor point cluster is fitted and connected to form the conductor point cloud data of the transmission line.

[0071] Preferably, as an implementable method, the density-based spatial clustering algorithm is a DBSCAN clustering algorithm.

[0072] Preferably, as an implementable method, the catenary equation of the conductor is:

[0073] y = a*cosh((xb) / a)+c;

[0074] Among them, y is the vertical coordinate of the wire, x is the horizontal coordinate of the wire, a is the catenary parameter, b is the horizontal coordinate of the axis of symmetry of the curve, and c is the vertical coordinate of the lowest point.

[0075] Preferably, as an implementable method, the extraction module, when executing "using a random forest classifier to identify and classify ground point cloud data including ground, trees, and towers in the transmission line corridor", specifically includes a digital elevation model generation module, a non-ground classification module, a tree extraction module, a single tree segmentation module, and a single tree attribute information extraction module:

[0076] The digital elevation model generation module is used to generate a digital elevation model (DEM) by using the segmented ground points through a triangulated irregular network (TIN) interpolation method;

[0077] The non-ground classification module is used to classify non-ground points: using a random forest classifier to classify non-ground points and identify ground objects such as towers and trees;

[0078] The tree extraction module is used to extract tree point clouds: separate tree point clouds from tree categories;

[0079] The single tree segmentation module is used to perform single tree segmentation: performing single tree segmentation on the tree point cloud;

[0080] The single tree attribute information extraction module is used to extract the attribute information of each tree based on the single tree segmentation result of the tree point cloud.

[0081] The attribute information of each tree includes the attribute information of the position (center of gravity coordinates), height, crown width, and breast diameter of each tree.

[0082] Preferably, as an implementable method, the single tree segmentation module specifically includes a single tree slicing module, a single tree contour recognition module, a single tree contour inter-layer connectivity calculation module, and a single tree growth module:

[0083] The single tree slicing module is used to slice the tree point cloud horizontally: slice the tree point cloud at equal intervals in the vertical direction;

[0084] The single tree outline recognition module is used to apply a convex hull algorithm in each slice to: identify the tree outline in each slice;

[0085] The single tree outline interlayer connectivity calculation module is used to determine the vertical structure of the tree by analyzing the overlap of the outlines between adjacent slices using interlayer connectivity analysis;

[0086] The single tree growth module is used to apply a regional growing algorithm: taking the crown vertex as a seed point and growing downward to achieve complete single tree segmentation.

[0087] Preferably, as an implementable method, the calculation and evaluation module specifically includes a spatial index module, a threshold acquisition module, a minimum distance calculation module and a risk assessment module:

[0088] The spatial index module is used to establish a KD tree index for the wire point cloud data, so as to facilitate the nearest neighbor search for other wire points on the current wire point; establish an R tree index for the tree point cloud data, so as to facilitate the search for trees in a specific target area;

[0089] The threshold acquisition module is used to determine and set a corresponding safety distance threshold according to the voltage level type of the current conductor;

[0090] The minimum distance calculation module is used to traverse the point cloud data of each tree; for each tree, use the KD tree to find the nearest wire point; calculate the Euclidean distance between the tree point cloud and the nearest wire point; record the minimum Euclidean distance between each tree point cloud and the wire where each tree is located as the minimum distance;

[0091] The risk assessment module is used to compare the calculated minimum distance with the safety distance threshold; if the minimum distance is less than the safety distance threshold, the tree is marked as a potential risk point.

[0092] Preferably, as an implementable method, the calculation and evaluation module further includes a recording module:

[0093] The recording module is used to import the extracted wire point cloud data; import the identified ground feature point cloud data; and load the attribute information of each tree calculated previously.

[0094] Preferably, as an implementable method, after marking the tree as a potential risk point, it also includes calculating the risk level and grading the potential risk points according to the current risk level:

[0095] The risk level is calculated using the following formula: risk level = (safety distance threshold - minimum distance) / safety distance threshold; potential risk points are graded according to the risk level; the potential risk points include high, medium and low risk levels.

[0096] like Figure 2As shown, an embodiment of the present invention provides a method for assessing tree barrier risk hazards in a transmission line corridor area based on a laser radar point cloud, including the following steps:

[0097] S1. Obtaining LiDAR point cloud data: Use the UAV-mounted LiDAR system to scan the transmission line corridor area along the transmission line corridor to obtain high-precision point cloud data;

[0098] S2, extracting the conductor point cloud data of the transmission line: performing ground point segmentation of the point cloud data using the RANSAC algorithm and clustering analysis and identification of the suspended points to extract the conductor point cloud data of the transmission line;

[0099] S3. Extract offline ground feature data of the transmission line: Use the random forest classifier to identify and classify the ground feature point cloud data including the ground, trees, and towers in the transmission line corridor;

[0100] S4, calculating the distance between the conductor and the ground object based on the conductor point cloud data and the ground object point cloud data, evaluating the potential risk and outputting the potential risk point;

[0101] S5. Screen potential risk points of trees to transmission lines, determine the potential risk points of trees to transmission lines and output them as target risk points, and then output a final risk assessment report based on the target risk points.

[0102] like Figure 3 As shown, the step S2 of performing ground point segmentation of the point cloud data using the RANSAC algorithm and clustering analysis and identification of the suspended points to extract the conductor point cloud data of the transmission line specifically includes:

[0103] S21. Perform spatial filtering on the original point cloud data: Use statistical outlier filtering algorithm to remove noise points and improve data quality;

[0104] S22, performing ground point segmentation on the point cloud data using the RANSAC algorithm: identifying and extracting ground points and non-ground points;

[0105] S23, separating the point cloud located above the ground by a height threshold method, and determining the point cloud as a suspended point to extract the suspended point;

[0106] S24, applying a density-based spatial clustering algorithm (such as DBSCAN) to the extracted dangling points to identify initial wire point clusters;

[0107] S25, fitting the conductor point clusters using the conductor catenary equation: fitting each identified point cluster using the conductor catenary equation to obtain a conductor curve equation;

[0108] S26. Perform secondary extraction of conductor points using the fitted conductor curve equation, and then verify the positions of multiple conductor points currently extracted for the second time. If the verification result of the conductor points is that the proportion of the correct conductor point cluster is higher than the standard proportion threshold, the conductor points output on the current conductor curve equation are determined to be the target conductor point cluster, and the target conductor point cluster is fitted and connected to form conductor point cloud data of the transmission line.

[0109] Ground points and non-ground points refer to point cloud data that are divided into ground and non-ground points according to their positions and attributes in point cloud data. Ground points usually refer to points on the surface, including various terrain features such as plains and hillsides, while non-ground points refer to all objects other than the ground, such as poles, trees, road signs, etc. In point cloud processing, separating ground points from non-ground points is the basis for further analysis and application. This step is crucial for subsequent point cloud classification, 3D modeling, terrain mapping and other applications. By separating ground points and non-ground points, information on non-ground objects such as poles, vegetation, etc. can be more accurately extracted, thereby improving the accuracy and efficiency of analysis and application. There are many ways to separate ground points and non-ground points, including mathematical morphology algorithm, slope filtering algorithm, irregular triangulation progressive encryption algorithm, moving surface fitting algorithm, etc.

[0110] In this embodiment, a height threshold is directly set to identify all points above the threshold. This simple and effective filtering step can quickly identify suspended objects, reduce the number of point clouds to be processed later, and improve efficiency. S24: Density-based spatial clustering algorithm (DBSCAN); DBSCAN can effectively identify clusters of arbitrary shapes and has good tolerance to noise. By setting the minimum number of points and radius parameters, the points related to the wire can be quickly clustered to reduce redundant calculations. S25: The wire catenary equation fits the wire point cluster, and the identified point cluster is fitted with the wire catenary equation. The catenary equation can accurately describe the shape of the wire and provide a mathematical model to enhance the structuring of point cloud data. The wire points are secondary extracted using the fitted wire equation, and the accuracy of the extraction results is verified. By setting a standard ratio threshold, the extracted wire point cloud is ensured to have a certain degree of accuracy. The feedback mechanism of this step can continuously optimize the extraction algorithm and improve the overall efficiency. Therefore, the above scheme significantly improves the extraction efficiency of the wire point cloud.

[0111] Preferably, as an implementable method, the density-based spatial clustering algorithm is a DBSCAN clustering algorithm.

[0112] Applying a density-based spatial clustering algorithm (such as DBSCAN) to the extracted dangling points to identify initial wire point clusters, specifically including:

[0113] Prepare data: First, this case has a set of extracted suspended points. These points represent all objects in the air, including wires, branches, etc. Example: Assume that this case has 1,000 suspended points, and their coordinates are distributed in a space of 300 meters long, 50 meters wide, and 30 meters high.

[0114] Set DBSCAN parameters: The DBSCAN algorithm requires two main parameters: ε (Epsilon, neighborhood radius) and MinPts (minimum number of points). Example: Based on the characteristics of the wire, this example may set ε = 0.5 meters (considering the density of the point cloud) and MinPts = 10 (to ensure that each cluster has at least 10 points).

[0115] Apply DBSCAN algorithm: Run DBSCAN algorithm on all dangling points. The algorithm will traverse each point and check whether there are at least MinPts points in its ε neighborhood. Example: The algorithm starts running and checks point by point. For example, it finds that there are 15 other points within 0.5 meters of point A (10, 5, 20), thus forming the core of a cluster.

[0116] Forming point clusters: The algorithm will group points that are densely accessible into one category to form different point clusters. Example: After processing, the algorithm may identify 5 main point clusters, 3 of which may be power lines and 2 may be large tree crowns.

[0117] Filtering and optimization: Perform preliminary screening based on the characteristics of point clusters (such as shape, size, density, etc.) to remove point clusters that are obviously not wires. Example: In this case, one of the point clusters was found to be spherical, which was obviously a tree crown, so it was excluded. The remaining 4 elongated point clusters are likely to be wires.

[0118] Output results: The preliminary identified wire point clusters are finally output to prepare for subsequent fine processing. Example: Four point clusters are finally output, containing 150, 180, 160, and 140 points respectively. They are roughly arranged in a straight line with a length of about 100 meters.

[0119] This process effectively separates possible wire points from other dangling points through the density clustering characteristics of the DBSCAN algorithm. This algorithm does not require high computing power and has a faster processing speed.

[0120] Preferably, as an implementable method, the catenary equation of the wire in step S25 is:

[0121] y = a*cosh((xb) / a)+c;

[0122] Among them, y is the coordinate of the wire in the vertical direction, x is the coordinate of the wire in the horizontal direction, a is the catenary parameter, b is the horizontal coordinate of the symmetry axis of the curve, and c is the vertical coordinate of the lowest point; the point cluster is fitted by the least squares method to obtain the optimal a, b, and c parameter values.

[0123] It should be noted that y: the vertical coordinate of the wire; x: the horizontal coordinate;

[0124] a: catenary parameters, which are related to the tension and line density of the conductor; b: horizontal coordinate of the axis of symmetry of the curve; c: vertical coordinate of the lowest point;

[0125] The catenary equation of the conductor is:

[0126] y = a*cosh((xb) / a)+c;

[0127] in:

[0128] y: The vertical coordinate of the wire

[0129] x: horizontal coordinate

[0130] a: Catenary parameters, related to the tension and line density of the conductor

[0131] b: horizontal coordinate of the curve's symmetry axis

[0132] c: vertical coordinate of the lowest point

[0133] This embodiment uses this equation and its application in transmission line modeling: The catenary equation describes the shape of a soft chain or rope with both ends fixed and drooping naturally under the action of gravity. For transmission lines, although the conductor has a certain rigidity, its shape is very close to the ideal catenary under the action of its own weight.

[0134] Parameter explanation:

[0135] a: This parameter determines the "tightness" of the curve. A larger a value means the curve is more "loose", while a smaller a value makes the curve more "tight". In physics, a = H / w, where H is the horizontal tension and w is the weight per unit length.

[0136] b: determines the horizontal position of the curve, that is, the x-coordinate of the axis of symmetry.

[0137] c: determines the vertical position of the curve, that is, the y coordinate of the lowest point.

[0138] Theoretical explanation: b represents the position of the symmetry axis of the catenary curve on the x-axis. The catenary is a symmetrical curve, and the parameter b determines the horizontal translation of the curve.

[0139] Practical understanding: In a transmission line, b usually corresponds to the midpoint between two towers. If b = 0, the axis of symmetry of the curve is exactly on the y-axis of the coordinate system. If b>0, the curve shifts to the right; if b<0, the curve shifts to the left.

[0140] The embodiment of the present invention removes noise points and ground points through spatial filtering and RANSAC algorithm, thereby improving the quality and accuracy of data for subsequent processing. The height threshold method is used to quickly separate the suspended points, greatly reducing the amount of data to be processed and improving processing efficiency.

[0141] Subsequently, density clustering algorithms such as DBSCAN were used to effectively identify the initial wire point clusters, providing a good foundation for subsequent fitting. Finally, the accuracy of wire identification was improved by using precise fitting and verification of fitting results. That is, precise fitting: the wire catenary equation was used for fitting to obtain an accurate wire curve equation, which is crucial for subsequent wire state analysis. Self-correction mechanism: Through the secondary extraction and verification process, a self-correction mechanism was established to effectively reduce misidentification and improve the accuracy of wire identification.

[0142] like Figure 4 As shown, the step S3 uses a random forest classifier to identify and classify point cloud data of objects including ground, trees, and towers in the transmission line corridor, specifically including:

[0143] S31, using the segmented ground points, generating a digital elevation model (DEM) through a triangulated irregular network (TIN) interpolation method;

[0144] S32, classifying non-ground points: using a random forest classifier to classify non-ground points and identify ground objects such as towers and trees;

[0145] S33, extracting tree point cloud: separating tree point cloud from tree category;

[0146] S34, performing single tree segmentation: performing single tree segmentation on the tree point cloud;

[0147] S35. Extracting attribute information of each tree based on the single tree segmentation result of the tree point cloud.

[0148] The attribute information of each tree includes the attribute information of the position (center of gravity coordinates), height, crown width, and breast diameter of each tree.

[0149] During the execution of S31: Generate Digital Elevation Model (DEM) Use the segmented ground points to generate DEM through the triangulated irregular network (TIN) interpolation method. Function: TIN is an effective interpolation method that can quickly generate terrain models. The generated DEM provides an accurate ground reference for non-ground point classification, helping the subsequent steps to more accurately identify ground objects.

[0150] During the execution of S32: Classification of non-ground points uses a random forest classifier to classify non-ground points and identify objects such as towers and trees. Random forest is an integrated learning method that can process high-dimensional data and improve classification accuracy. Its training process can use existing labeled data to quickly adapt to different types of objects and reduce manual intervention. During the execution of S33: Extracting tree point clouds, this step quickly obtains tree-related point clouds through simple data filtering, reducing the amount of data for subsequent processing, thereby improving efficiency.

[0151] like Figure 5 As shown, the single tree segmentation in step S34 adopts the following steps:

[0152] S341, horizontally slicing the tree point cloud: slicing the tree point cloud at equal intervals along the vertical direction;

[0153] S342, applying a convex hull algorithm within each slice: identifying tree outlines in each slice;

[0154] S343, using inter-layer connectivity analysis: by analyzing the overlap of contours between adjacent slices, the vertical structure of the tree is determined;

[0155] S344. Apply the region growing algorithm: take the crown vertex as the seed point and grow downward to achieve complete single tree segmentation.

[0156] Assume that this embodiment has point cloud data of a small forest, including 10 trees, with tree heights between 10-15 meters and crown diameters between 3-5 meters. The goal of this embodiment is to perform single tree segmentation on these trees.

[0157] S341, horizontal slices: This embodiment decides to perform horizontal slices every 0.5 meters, starting from 0 meters on the ground and up to 15 meters in height. Result: This embodiment obtains 30 horizontal slices. In the lower slices (such as 0-3 meters), this embodiment mainly sees the cross section of the trunk. In the higher slices (such as 5-15 meters), this embodiment sees the cross section of the crown.

[0158] S342. Apply convex hull algorithm: apply the convex hull algorithm to each slice.

[0159] Result: In the lower slices, the present embodiment obtains a nearly circular outline (trunk). In the higher slices, the present embodiment obtains an irregular polygonal outline (crown). For example, in a slice at a height of 10 meters, the present embodiment may obtain 10 different polygonal outlines, each representing a cross section of a tree.

[0160] S343. Inter-layer connectivity analysis: This embodiment compares the overlap of the outlines of adjacent slices. Results: This embodiment finds that the outline gradually increases from the bottom to the top (trunk to crown), and then begins to decrease at the top of the crown. For example, the outline of tree 1 remains relatively stable at a height of 0-5 meters (trunk), then rapidly expands at 5-12 meters (crown), and finally begins to decrease at 12-15 meters (top of the crown).

[0161] S344, Region Growing Algorithm: This implementation method identifies the highest point of each tree as a seed point. In this example, this implementation method finds 10 seed points with heights between 13-15 meters. Starting from each seed point, this implementation method "grows" downward, adding connected points to the same tree. Result: This implementation method successfully splits the point cloud data into 10 independent tree entities.

[0162] Final result: This implementation successfully segmented 10 independent trees, each with complete point cloud data from trunk to crown.

[0163] For example, tree 1: height 14.5 meters, maximum crown diameter 4.2 meters; tree 2: height 13.8 meters, maximum crown diameter 3.9 meters;

[0164] This process allows the present embodiment to accurately segment each tree, even when the tree crowns may partially overlap. In this way, the present embodiment can not only identify individual trees, but also obtain detailed structural information of each tree, providing accurate basic data for subsequent tree analysis and risk assessment.

[0165] like Figure 6 As shown, S4 calculates the distance between the wire and the ground object based on the wire point cloud data and the ground object point cloud data, evaluates the potential risk and outputs the potential risk point, specifically including:

[0166] S42, establishing spatial index: establishing a KD tree index for the wire point cloud data, so as to facilitate searching other wire points for the nearest neighbors on the current wire point; establishing an R tree index for the tree point cloud data, so as to facilitate searching for trees in a specific target area;

[0167] S43, obtaining a preset safety distance threshold: the voltage level type of the current conductor, and determining and setting a corresponding safety distance threshold;

[0168] S44, calculating the minimum distance between the wire and the tree: traversing the point cloud data of each tree; for each tree, using the KD tree to find the nearest wire point; calculating the Euclidean distance between the tree point cloud and the nearest wire point; recording the minimum value of the Euclidean distance between each tree point cloud and the wire where each tree is located as the minimum distance;

[0169] S45, risk assessment: compare the calculated minimum distance with the safety distance threshold; if the minimum distance is less than the safety distance threshold, mark the tree as a potential risk point.

[0170] Continue to see Figure 6 , before creating the spatial index, it also includes:

[0171] S41, import the wire point cloud data extracted in S2; import the ground object point cloud data identified in S3, especially the tree point cloud data; load the attribute information (position, height, crown width, etc.) of each tree calculated in S3.

[0172] Preferably, as an implementable method, after marking the tree as a potential risk point, it also includes calculating the risk level and grading the potential risk points according to the current risk level:

[0173] The risk level is calculated using the following formula: risk level = (safety distance threshold - minimum distance) / safety distance threshold; potential risk points are graded according to the risk level; the potential risk points include high, medium and low risk levels.

[0174] like Figure 7 As shown, Figure 7 The present invention is a schematic diagram of the operation steps for outputting warning messages at future time points in a method for assessing tree barrier risk hazards in a transmission line corridor area based on a lidar point cloud.

[0175] After the risk points are graded, they also include:

[0176] S61, obtaining tree point clouds on trees with non-potential risk points (that is, trees excluding potential risk points are recorded as trees with non-potential risk points);

[0177] S62, obtaining the value of the terrain slope where the tree that is not a potential risk point is located, the current tree height, the maximum tree height of the same type of trees within a preset radius, and the original minimum distance to calculate the corrected minimum distance to obtain a target correction value;

[0178] S63, then based on the target correction value and the safety distance threshold, the average annual growth rate of trees of the same species, the annual precipitation, the multi-year average precipitation, the first influencing parameter and the second influencing parameter, calculate and output the time required to predict the future risk point (i.e., the time required to reach the risk state), so as to propose an early warning message at a future time point:

[0179] Corrected minimum distance calculation formula

[0180] D min : original minimum distance; θ: terrain slope; H tree : Tree height; H max : The maximum tree height of the same tree within the preset radius; α, β, γ are all correction coefficients;

[0181] Calculation formula for predicting future risk points Among them, T risk The time required to reach the risk state; D safe is the safety distance threshold; G rate is the average annual growth rate of trees of the same species; P annual is the annual precipitation;

[0182] P avg is the multi-year average precipitation; λ and ω are the first and second influencing parameters respectively;

[0183] Example: Setting parameters:

[0184] Original minimum distance Dmin: 50 meters;

[0185] Terrain slope θ: 30 degrees;

[0186] Tree height Htree: 10 meters;

[0187] Maximum tree height in the area Hmax: 25 meters;

[0188] Correction coefficient: α = 0.2, β = 0.5, γ = 1.0;

[0189] Safety distance threshold Dsafe: 70 meters;

[0190] Average annual growth rate of trees: 0.5 m / year;

[0191] Annual precipitation Pannual: 800 mm;

[0192] Average annual precipitation Pavg: 600 mm;

[0193] Influence coefficient: λ = 0.3; ω = 1.2;

[0194] Step 1: Calculate the corrected minimum distance D corrected :

[0195]

[0196] Calculate sin(30°)=0.5;

[0197] Substitute into the calculation: D corrected =50×(1+0.2·0.5+0.5·0.4)=50×(1+0.1+0.2)=50×1.3=65 meters;

[0198] Step 2: Predict future risk points T risk :

[0199]

[0200] calculate

[0201] Calculate (1.333) 1.2 (i.e. approximately 1.5);

[0202]

[0203] Result analysis Based on the above calculations, the manager concluded that the revised minimum distance is 65 meters, which is lower than the safe distance of 70 meters. The time required to reach the risk state is about 6.9 years.

[0204] The present invention realizes the prediction of tree species growth cycle and the height information of trees after a specified time period. It combines the LiDAR point cloud data to determine when and where tree hidden dangers will appear. The accuracy and reliability of the model are verified during the application process, thus improving the ability of operation and maintenance units to predict tree obstacle hazards.

[0205] In summary, the system and method for evaluating tree barrier risk hazards in the corridor area of ​​a transmission line based on lidar point cloud proposed in the example of the present invention can improve the safety of transmission lines, reduce the risk of accidents caused by trees and other landforms, and has important practical value for early warning of tree barrier risks of transmission lines.

[0206] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A system for analyzing the risk of tree obstacles on lines based on laser radar point cloud, characterized in that: The system includes: A point cloud acquisition module is used to acquire point cloud data along the transmission line corridor in the transmission line corridor area; A cluster processing module is used to extract the conductor point cloud data of the transmission line: the point cloud data is segmented by the RANSAC algorithm for ground points and clustered for the suspended points for identification, so as to extract the conductor point cloud data of the transmission line; The extraction module is used to extract offline ground object data of the transmission line: using the random forest classifier to identify and classify the ground object point cloud data including the ground, trees, and towers in the transmission line corridor; A calculation and evaluation module is used to calculate the distance between the conductor and the ground object based on the conductor point cloud data and the ground object point cloud data, evaluate the potential risk and output the potential risk point; The report output module is used to screen potential risk points of trees to transmission lines, determine the potential risk points of trees to transmission lines and output them as target risk points, and output a risk assessment report based on the target risk points.

2. The system according to claim 1, characterized in that The cluster processing module specifically includes: The ground point and non-ground point segmentation module is used to segment the point cloud data using the RANSAC algorithm: identify and extract ground points and non-ground points; A suspended point extraction module is used to separate the point cloud located above the ground through a height threshold method and determine it as a suspended point to extract the suspended point; A point cluster identification module is used to apply a density-based spatial clustering algorithm to the extracted dangling points to identify the initial wire point clusters; The wire fitting module is used to fit the wire point cluster using the wire catenary equation: the wire catenary equation is applied to each identified point cluster for fitting to obtain the wire curve equation; The verification output module is used to use the fitted conductor curve equation to perform secondary extraction of conductor points, and then verify the positions of multiple conductor points currently extracted for the second time. If the verification result of the conductor point is that the proportion of the correct conductor point cluster is higher than the standard proportion threshold, the conductor point output on the current conductor curve equation is determined to be the target conductor point cluster, and the target conductor point cluster is fitted and connected to form the conductor point cloud data of the transmission line.

3. The system according to claim 2, characterized in that The density-based spatial clustering algorithm is the DBSCAN clustering algorithm.

4. The system according to claim 3, characterized in that The catenary equation of the conductor is: y=a*cosh((xb) / a)+c; wherein y is the vertical coordinate of the conductor, x is the horizontal coordinate of the conductor, a is the catenary parameter, b is the horizontal coordinate of the axis of symmetry of the curve, and c is the vertical coordinate of the lowest point.

5. The system according to claim 4, characterized in that Extraction modules, including: A digital elevation model generation module is used to generate a digital elevation model using segmented ground points through a triangular irregular network interpolation method; The non-ground classification module is used to classify non-ground points: the random forest classifier is used to classify non-ground points and identify ground objects such as towers and trees; The tree extraction module is used to extract tree point clouds: separate tree point clouds from tree categories; A single tree segmentation module, used for performing single tree segmentation: performing single tree segmentation on the tree point cloud; The single tree attribute information extraction module is used to extract the attribute information of each tree based on the single tree segmentation result of the tree point cloud; wherein the attribute information of each tree includes the position, height, crown width and breast diameter of each tree.

6. The system according to claim 5, characterized in that The single wood splitting module specifically includes: Single tree slicing module, used to slice the tree point cloud horizontally: slice the tree point cloud at equal intervals in the vertical direction; Single tree outline recognition module, which is used to apply the convex hull algorithm within each slice: identify the tree outline in each slice; The module for calculating inter-layer connectivity of single tree contours is used to determine the vertical structure of trees by analyzing the overlap of contours between adjacent slices using inter-layer connectivity analysis; The single tree growth module is used to apply the region growing algorithm: taking the crown vertex as the seed point and growing downward to achieve complete single tree segmentation.

7. The system according to claim 6, characterized in that The calculation and evaluation module specifically includes: The spatial index module is used to establish a KD tree index for the wire point cloud data, so as to facilitate the search for other wire points for the nearest neighbors of the current wire point; and to establish an R tree index for the tree point cloud data, so as to facilitate the search for trees in a specific target area; The threshold acquisition module is used to determine and set the corresponding safety distance threshold according to the voltage level type of the current conductor; The minimum distance calculation module is used to traverse the point cloud data of each tree; for each tree, use the KD tree to find the nearest wire point; calculate the Euclidean distance between the tree point cloud and the nearest wire point; record the minimum Euclidean distance between each tree point cloud and the wire where each tree is located as the minimum distance; The risk assessment module is used to compare the calculated minimum distance with the safety distance threshold; if the minimum distance is less than the safety distance threshold, the tree is marked as a potential risk point.

8. The system according to claim 7, characterized in that The calculation and evaluation module also includes a recording module: The recording module is used to import the wire point cloud data and the ground object point cloud data, and load the attribute information of each tree.

9. The system according to claim 7, characterized in that The system also includes a risk level grading module; the risk level grading module is used to calculate the risk level; and to grade potential risk points according to the risk level; wherein the potential risk points include high, medium, and low risk levels.

10. A method for analyzing the risk of tree obstacles on a line based on laser radar point cloud, characterized in that: Methods include: Obtain point cloud data along the transmission line corridor in the transmission line corridor area; Extracting the conductor point cloud data of the transmission line: performing ground point segmentation of the point cloud data using the RANSAC algorithm and clustering analysis and identification of the suspended points to extract the conductor point cloud data of the transmission line; Extract offline feature data of transmission lines: Use random forest classifier to identify and classify feature point cloud data including ground, trees, and towers within the transmission line corridor; Calculate the distance between the wire and the ground object based on the wire point cloud data and the ground object point cloud data, evaluate the potential risk and output the potential risk points; Screen potential risk points of trees to transmission lines, determine the potential risk points of trees to transmission lines and output them as target risk points, and output a risk assessment report based on the target risk points.

Images