Tree obstacle risk monitoring method and system for overhead transmission line

Abstract

The invention provides a tree obstacle risk monitoring method and system for an overhead transmission line. The method comprises the following steps: acquiring satellite remote sensing stereo image pair data of a monitoring area and unmanned aerial vehicle point cloud data of a local area in the monitoring area; obtaining a first digital surface model and a first digital elevation model of the local area based on the unmanned aerial vehicle point cloud data; obtaining a second digital surface model and a second digital elevation model of the monitoring area based on the satellite remote sensing stereo image pair data; correcting the second digital surface model and the second digital elevation model by using the first digital surface model and the first digital elevation model to obtain a target digital surface model and a target digital elevation model of the monitoring area; and completing tree obstacle risk monitoring of the overhead transmission line in the monitoring area according to the target digital surface model and the target digital elevation model. According to the invention, the tree obstacle risk monitoring precision of the large-range overhead transmission line can be improved.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a method and system for monitoring tree obstruction risks of overhead transmission lines. Background Technology

[0002] Tree obstruction monitoring is crucial for the safety of power transmission lines. Contact or proximity between trees and transmission lines can lead to short circuits, fires, power outages, and even equipment damage, affecting the stability and reliability of power supply. Tree obstruction hazards are a key focus during transmission line safety inspections. Because the distance between transmission lines and trees is often less than the safe distance, and because wildfires can cause trees to burn, transmission lines can trip, impacting the power system's operation. To better manage tree obstruction hazards and ensure the stable operation of transmission lines, it is necessary to monitor the height of tree growth and the distance between trees and transmission lines, assess the risk of tree obstruction hazards, and promptly address any hazard information.

[0003] Currently, the most common method for collecting information on tree obstructions along overhead transmission lines is manual inspection. Traditional methods rely heavily on visual inspection, which is labor-intensive, makes it difficult to pinpoint the exact location of faults, and is inefficient.

[0004] Related research mainly focuses on using UAV-mounted LiDAR technology for 3D modeling and tree obstruction analysis of power transmission lines. This method uses lasers to achieve echo ranging and orientation, obtaining information such as the position and radial velocity of targets, thus enabling the monitoring of tree obstruction risks. However, UAV-mounted LiDAR technology is costly and difficult to apply to tree obstruction monitoring of large areas of overhead power transmission lines. While satellite remote sensing technology can achieve wide-area coverage in extracting Digital Surface Models (DSMs) and Digital Elevation Models (DEMs), the accuracy of the data obtained is relatively low and the error is large due to limitations in resolution and sensor characteristics, which cannot meet the accuracy requirements for tree obstruction monitoring. Summary of the Invention

[0005] To address at least one problem in the prior art, this application proposes a method and system for monitoring tree obstruction risks of overhead transmission lines, which can improve the accuracy of monitoring tree obstruction risks of overhead transmission lines over a wide area.

[0006] To address the aforementioned technical problems, this application provides the following technical solution:

[0007] Firstly, this application provides a method for monitoring tree obstruction risk of overhead transmission lines, including:

[0008] Acquire satellite remote sensing stereo image pairs of the monitored area and UAV point cloud data of local areas within the monitored area;

[0009] Based on the UAV point cloud data, a first digital surface model and a first digital elevation model of the local area are obtained;

[0010] Based on the satellite remote sensing stereo image pair data, a second digital surface model and a second digital elevation model of the monitored area are obtained;

[0011] The second digital surface model and the second digital elevation model are corrected by applying the first digital surface model and the first digital elevation model to obtain the target digital surface model and the target digital elevation model of the monitoring area;

[0012] Based on the target digital surface model and the target digital elevation model, tree obstacle risk monitoring of overhead transmission lines in the monitoring area is completed.

[0013] In one embodiment, the step of applying the first digital surface model and the first digital elevation model to correct the second digital surface model and the second digital elevation model to obtain the target digital surface model and the target digital elevation model of the monitoring area includes:

[0014] A batch of training samples and their corresponding labels are collected. Each training sample includes: elevation data of a pixel in the local region in the second digital elevation model and the second digital surface model, first slope information corresponding to the pixel obtained based on the second digital elevation model, and second slope information corresponding to the pixel obtained based on the second digital surface model. The label corresponding to each training sample includes: elevation data of the pixel corresponding to the training sample in the first digital surface model and the first digital elevation model, respectively. The pixels corresponding to each training sample are different. The first slope information and the second slope information both include at least one of slope, aspect, slope rate, and aspect rate.

[0015] The random forest algorithm is trained using a batch of training samples and their corresponding labels to obtain an accuracy correction model;

[0016] By applying the second digital surface model, the second digital elevation model, the accuracy correction model, and the first slope information and the second slope information of each pixel in the monitoring area, the target digital surface model and the target digital elevation model of the monitoring area are obtained.

[0017] In one embodiment, before collecting the batch of training samples and their corresponding labels, the method further includes:

[0018] Based on the second digital elevation model and the first digital elevation model, the elevation error of each pixel in the local area is obtained;

[0019] Determine whether there are any abnormal pixels in the local area whose elevation error does not meet the preset error conditions. If so, delete the elevation data of the abnormal pixel in the second digital elevation model.

[0020] In one embodiment, the preset error condition is a three-standard-deviation rule.

[0021] In one embodiment, the step of monitoring tree obstruction risks of overhead transmission lines in the monitoring area based on the target digital surface model and the target digital elevation model includes:

[0022] The height difference of each pixel in the monitoring area is obtained by performing differential calculations on the target digital surface model and the target digital elevation model.

[0023] By applying the height difference of each pixel and the pre-acquired multispectral image, the height and location of the trees in the monitoring area are determined.

[0024] Based on the height and location of trees in the monitoring area, tree obstacle risk monitoring is completed in the monitoring area.

[0025] In one embodiment, the step of monitoring tree barrier risk in the monitoring area based on tree height and location includes:

[0026] The disturbance range after a tree falls is determined based on the tree height and location in the monitoring area.

[0027] Based on the tree height, tree location, and disturbance range after a tree falls in the monitoring area, determine whether the monitoring area meets the preset safety conditions. If so, the monitoring area is determined to be a safe area; otherwise, the monitoring area is determined to be a tree obstacle risk area.

[0028] The preset distance conditions are as follows: the height of all trees is less than the preset vertical safety distance, the location of the trees is outside the preset minimum horizontal safety distance range, and the disturbance range after the trees fall is outside the preset minimum horizontal safety distance range.

[0029] Secondly, this application provides a tree obstacle risk monitoring system for overhead transmission lines, comprising:

[0030] The first acquisition device is used to acquire satellite remote sensing stereo image pairs of the monitoring area and UAV point cloud data of a local area in the monitoring area;

[0031] The first obtaining device is used to obtain a first digital surface model and a first digital elevation model of the local area based on the UAV point cloud data.

[0032] The second obtaining device is used to obtain a second digital surface model and a second digital elevation model of the monitored area based on the satellite remote sensing stereo image pair data;

[0033] A calibration device is used to apply the first digital surface model and the first digital elevation model to calibrate the second digital surface model and the second digital elevation model, so as to obtain the target digital surface model and the target digital elevation model of the monitoring area.

[0034] The monitoring device is used to monitor the tree obstruction risk of overhead transmission lines in the monitoring area based on the target digital surface model and the target digital elevation model.

[0035] In one embodiment, the correction device includes:

[0036] The acquisition module is used to acquire batch training samples and their corresponding labels. Each training sample includes: elevation data of a pixel in the local region in the second digital elevation model and the second digital surface model, first slope information corresponding to the pixel obtained based on the second digital elevation model, and second slope information corresponding to the pixel obtained based on the second digital surface model. The label corresponding to each training sample includes: elevation data of the pixel corresponding to the training sample in the first digital surface model and the first digital elevation model, respectively. The pixels corresponding to each training sample are different. The first slope information and the second slope information both include at least one of: slope, aspect, slope rate, and aspect rate.

[0037] The training module is used to train the random forest algorithm using a batch of training samples and their corresponding labels to obtain an accuracy correction model.

[0038] The application module is used to apply the slope information of each pixel in the second digital surface model, the second digital elevation model, and the accuracy correction model to obtain the target digital surface model and the target digital elevation model of the monitoring area.

[0039] In one embodiment, the tree obstacle risk monitoring system for overhead transmission lines further includes:

[0040] The second acquisition device is used to obtain the elevation error of each pixel in the local area based on the second digital elevation model and the first digital elevation model.

[0041] The deletion device is used to determine whether there are any abnormal pixels in the local area whose elevation errors do not meet the preset error conditions. If so, the elevation data of the abnormal pixel in the second digital elevation model is deleted.

[0042] In one embodiment, the preset error condition is a three-standard-deviation rule.

[0043] In one embodiment, the monitoring device includes:

[0044] The differential calculation module is used to perform differential calculations on the target digital surface model and the target digital elevation model to obtain the height difference of each pixel in the monitoring area;

[0045] The determination module is used to determine the height and location of trees in the monitoring area by applying the height difference of each pixel and the pre-acquired multispectral image.

[0046] The monitoring module is used to monitor tree obstacle risks in the monitoring area based on the height and location of the trees.

[0047] In one embodiment, the monitoring module includes:

[0048] The determining unit is used to determine the disturbance range after a tree falls, based on the tree height and location in the monitoring area.

[0049] The monitoring unit is used to determine whether the monitoring area meets the preset safety conditions based on the tree height, tree location, and disturbance range after the tree falls. If so, the monitoring area is determined to be a safe area; otherwise, the monitoring area is determined to be a tree obstacle risk area.

[0050] The preset distance conditions are as follows: the height of all trees is less than the preset vertical safety distance, the location of the trees is outside the preset minimum horizontal safety distance range, and the disturbance range after the trees fall is outside the preset minimum horizontal safety distance range.

[0051] Thirdly, this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the tree obstacle risk monitoring method for overhead transmission lines.

[0052] Fourthly, this application provides a computer-readable storage medium storing computer instructions thereon, which, when executed by a processor, implement the tree obstacle risk monitoring method for overhead transmission lines.

[0053] As can be seen from the above technical solution, this application provides a method and system for monitoring tree obstruction risks of overhead transmission lines. The method includes: acquiring satellite remote sensing stereo image pairs of the monitoring area and UAV point cloud data of a local area within the monitoring area; obtaining a first digital surface model and a first digital elevation model of the local area based on the UAV point cloud data; obtaining a second digital surface model and a second digital elevation model of the monitoring area based on the satellite remote sensing stereo image pairs; applying the first digital surface model and the first digital elevation model to correct the second digital surface model and the second digital elevation model, thereby obtaining a target digital surface model and a target digital elevation model of the monitoring area; and completing tree obstruction risk monitoring of the overhead transmission lines in the monitoring area based on the target digital surface model and the target digital elevation model, which can improve the accuracy of tree obstruction risk monitoring for large-scale overhead transmission lines. Specifically, the advantages of this solution include:

[0054] This solution combines UAV point cloud data and satellite imagery to achieve complementary advantages, bringing multi-dimensional and in-depth benefits to tree obstruction monitoring of long-distance overhead power transmission lines. On the one hand, UAV point cloud data can provide high-precision local terrain and overhead power transmission line information, ensuring the accuracy of tree obstruction monitoring; on the other hand, satellite imagery, with its wide-area coverage advantage, can quickly acquire large-scale environmental data around long-distance overhead power transmission lines, thereby enabling comprehensive monitoring of the entire overhead power transmission line and greatly expanding the monitoring range.

[0055] In terms of accuracy improvement, the fusion of the two types of data can extract terrain data more accurately and determine the spatial relationship between trees and overhead power lines more precisely, thereby significantly improving the accuracy of tree obstacle risk monitoring and early warning. Regarding cost reduction, it eliminates the need to rely solely on high-cost drone equipment for comprehensive coverage monitoring, allowing for more efficient resource utilization and effectively reducing production costs.

[0056] In addition, this solution can also achieve dynamic monitoring and early warning of tree obstruction risks on overhead transmission lines, timely detection of potential risks, and has important practical value and broad application prospects, providing a strong guarantee for the safe and stable operation of transmission lines. Attached Figure Description

[0057] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0058] Figure 1This is a first flowchart illustrating the tree obstacle risk monitoring method for overhead transmission lines in this application embodiment;

[0059] Figure 2 This is a second flowchart illustrating the tree obstacle risk monitoring method for overhead transmission lines in this application embodiment;

[0060] Figure 3 This is a schematic diagram of the third process of the tree obstacle risk monitoring method for overhead transmission lines in the embodiments of this application;

[0061] Figure 4 This is a schematic diagram of the fourth process of the tree obstacle risk monitoring method for overhead transmission lines in the embodiments of this application;

[0062] Figure 5 This is a schematic diagram of the fifth process of the tree obstacle risk monitoring method for overhead transmission lines in the embodiments of this application;

[0063] Figure 6 This is a schematic diagram of the structure of the tree obstacle risk monitoring system for overhead transmission lines in the embodiments of this application;

[0064] Figure 7 This is a schematic block diagram of the system configuration of an electronic device according to an embodiment of this application. Detailed Implementation

[0065] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0066] Currently, in the field of tree obstruction monitoring along power transmission lines, the use of unmanned aerial vehicles (UAVs) equipped with LiDAR technology is a significant research direction. This technology, leveraging the UAV platform, enables LiDAR to operate in the vicinity of power transmission lines. By emitting laser pulses and utilizing echo ranging and direction-finding principles, LiDAR accurately acquires key information such as the position and radial velocity of targets (e.g., trees, power transmission lines). Based on this information, three-dimensional modeling of the power transmission lines can be achieved, clearly presenting the three-dimensional spatial structure of the power transmission lines and their surrounding environment. This allows for tree obstruction analysis, accurately determining the spatial relationship between trees and power transmission lines, and providing strong support for tree obstruction risk monitoring.

[0067] Power lines are extracted based on the elevation distribution, density, and tilt angle characteristics of the laser point cloud. The 3D reconstructed vectors of the power lines and the 3D convex hulls of each vegetation point cloud segment are obtained. By calculating the distances between the vegetation convex hull points and the power line vectors, the presence of potential tree obstruction hazards along the transmission lines is detected. Further calculations of the distances between the power line vectors and all vegetation points determine the existence of such hazards.

[0068] The existing rapid detection method for tree obstruction hazards based on laser point clouds follows this process: First, power line points are accurately extracted using the point cloud density, elevation distribution, and tilt angle features of the transmission line. Then, tower points are extracted using a region search method based on tower location information. Ground points are extracted using a cloth-based simulated filtering algorithm. Finally, the remaining vegetation point cloud is manually denoised. A power line striping extraction algorithm based on random sample consensus (RANSAC) is used to separate individual power lines. Three-dimensional reconstruction of each power line is achieved using straight line and parabolic models, generating power line vectors. Finally, three-dimensional convex hulls are generated for the vegetation point cloud segments of each power line segment. The distances between the vegetation convex hull points and the power line vectors within that segment are calculated to quickly detect tree obstruction hazard points that do not meet safety distance requirements. Based on the rapid detection results, the search radius is expanded to obtain detailed tree obstruction hazard information, completing the rapid detection and analysis of tree obstruction hazards based on transmission line laser point cloud data.

[0069] While LiDAR technology on drones offers numerous advantages for monitoring tree obstructions along power transmission lines, its high cost is a significant drawback. From equipment purchase and routine maintenance to flight operations, substantial investment is required, hindering its widespread application in monitoring tree obstructions along large power transmission corridors and limiting its scope for large-scale power facility maintenance. Satellite remote sensing, while capable of wide-area coverage, suffers from significant accuracy limitations. Due to satellite resolution and sensor characteristics, data acquired through satellite remote sensing has relatively low precision and large errors. In scenarios like tree obstruction monitoring, where high accuracy is crucial, this low-precision data fails to meet practical needs, making it difficult to accurately determine the precise distance and spatial relationship between trees and power transmission lines, thus impacting the accuracy and reliability of tree obstruction monitoring.

[0070] In summary, while traditional methods can extract power lines based on features such as elevation, density, and tilt angle of laser point clouds and detect tree obstruction hazards by calculating the distance between vegetation convex hull points and power line vectors, the calculation process is relatively complex. Furthermore, when dealing with long-distance overhead power lines, relying solely on lidar technology carried by UAVs limits the data acquisition range, making it difficult to achieve large-area, wide-area monitoring. In addition, the high cost of this technology restricts its large-scale application.

[0071] Based on this, and to address the problems existing in the prior art, this application provides a method and system for monitoring tree obstruction risks on overhead transmission lines. This method combines satellite remote sensing and UAV point cloud technology, uses machine learning models to correct the DSM and DEM obtained from satellite data, calculates tree height and lodging disturbance range, and combines multispectral imagery to identify the canopy range, thereby achieving dynamic early warning of tree obstructions. This can improve the accuracy and efficiency of tree obstruction monitoring, reduce costs, and is of significant value to the safe operation of power systems.

[0072] The following examples illustrate this in detail.

[0073] To improve the accuracy of tree obstruction risk monitoring for large-scale overhead transmission lines, this embodiment provides a method for monitoring tree obstruction risk on overhead transmission lines, wherein the execution subject is an overhead transmission line tree obstruction risk monitoring system, which includes, but is not limited to, a server, such as... Figure 1 As shown, this method specifically includes the following:

[0074] Step 100: Acquire satellite remote sensing stereo image pairs of the monitoring area and UAV point cloud data of local areas within the monitoring area.

[0075] Specifically, the monitoring area can be an area containing overhead power lines; this embodiment is particularly suitable for large-area, long-distance monitoring areas. The local area can be a portion of the monitoring area, and its location and size can be set according to actual conditions. The satellite remote sensing stereo image pair data can be high-resolution optical satellite remote sensing stereo image pair data. Before step 200, the satellite remote sensing stereo image pair data and the UAV point cloud data can be preprocessed separately.

[0076] Step 200: Based on the UAV point cloud data, obtain the first digital surface model and the first digital elevation model of the local area.

[0077] Specifically, the three-dimensional coordinates (i.e., latitude and longitude coordinates and elevation values) of each pixel can be extracted from UAV point cloud data using a multi-view stereo matching method, thereby obtaining a first digital surface model and a first digital elevation model; the first digital surface model represents the digital surface model DSM obtained based on UAV point cloud data, and the first digital elevation model represents the digital elevation model DEM obtained based on UAV point cloud data.

[0078] Specifically, DSM represents the top elevation data of all features on the Earth's surface, including buildings, vegetation, etc.; that is, when there is vegetation and buildings on the surface, DSM represents the elevation data of the vegetation and buildings. DEM represents the elevation data of the Earth's surface (topographic relief), without considering the height of objects such as buildings and vegetation on the ground; that is, it is DSM with the elevation data of surface buildings and vegetation removed.

[0079] Step 300: Based on the satellite remote sensing stereo image pair data, obtain the second digital surface model and the second digital elevation model of the monitoring area.

[0080] Specifically, the three-dimensional coordinates of each pixel can be extracted from satellite remote sensing stereo image pairs using a multi-view stereo matching method, thereby obtaining a second digital surface model and a second digital elevation model. The second digital surface model represents the digital surface model obtained based on the satellite remote sensing stereo image pairs, and the second digital elevation model represents the digital elevation model obtained based on the satellite remote sensing stereo image pairs. The accuracy of the first digital surface model is higher than that of the second digital surface model, and the accuracy of the first digital elevation model is higher than that of the second digital elevation model.

[0081] Step 400: Apply the first digital surface model and the first digital elevation model to correct the second digital surface model and the second digital elevation model to obtain the target digital surface model and the target digital elevation model of the monitoring area.

[0082] Specifically, the target digital surface model can represent the corrected digital surface model of the entire monitoring area; the target digital elevation model can represent the corrected digital elevation model of the entire monitoring area. The accuracy of the target digital surface model is higher than that of the second digital surface model, and the accuracy of the target digital elevation model is higher than that of the second digital elevation model.

[0083] Step 500: Based on the target digital surface model and the target digital elevation model, complete the tree obstacle risk monitoring of overhead transmission lines in the monitoring area.

[0084] As described above, the tree obstacle risk monitoring method for overhead transmission lines provided in this embodiment can improve the accuracy of tree obstacle risk monitoring for large-scale overhead transmission lines; it can achieve high-precision terrain data extraction around long-distance overhead transmission lines, and fully utilize the wide-area advantages of satellite imagery to realize dynamic monitoring and early warning of tree obstacles risk for overhead transmission lines. It can not only significantly improve the accuracy of tree obstacle risk monitoring and early warning, but also effectively reduce production costs, and has important practical value and broad application prospects.

[0085] To improve the accuracy of the digital surface model and digital elevation model of the monitoring area, such as Figure 2As shown, in one embodiment, step 400 includes:

[0086] Step 401: Collect batch training samples and their corresponding labels. Each training sample includes: elevation data of a pixel in the local region in the second digital elevation model and the second digital surface model, first slope information corresponding to the pixel obtained based on the second digital elevation model, and second slope information corresponding to the pixel obtained based on the second digital surface model. The label corresponding to each training sample includes: elevation data of the pixel corresponding to the training sample in the first digital surface model and the first digital elevation model, respectively. The pixels corresponding to each training sample are different. The first slope information and the second slope information both include at least one of slope, aspect, slope rate, and aspect rate.

[0087] Specifically, the monitoring area can be divided into multiple grids according to the actual situation, and each grid can be equivalent to a pixel. The elevation data of each pixel can include: the latitude and longitude coordinates and the elevation value of the pixel. The first slope information corresponding to any pixel is the slope information of the pixel obtained based on the second digital elevation model, and the second slope information corresponding to any pixel is the slope information of the pixel obtained based on the second digital surface model.

[0088] For example, assuming the pixel corresponding to training sample 1 is pixel 1, training sample 1 can be shown in Table 1. The information obtained based on the DEM value in the second digital elevation model includes: second slope, second aspect, second slope rate, and second aspect rate; the information obtained based on the DSM value in the second digital surface model includes: second slope, second aspect, second slope rate, and second aspect rate. The labels corresponding to training sample 1 can be shown in Table 2.

[0089] Table 1

[0090]

[0091] Table 2

[0092]

[0093] Specifically, the first slope information and the second slope information of pixel i can be obtained based on the second digital elevation model, the second digital surface model, and the following formula:

[0094]

[0095]

[0096] Among them, Z i To monitor the elevation value of pixel i in the target model, X iTo determine the east-west spatial coordinates of pixel i in the monitoring area, Y i Let i be the spatial coordinates of pixel i in the north-south direction within the monitoring area. The rate of change of elevation along the east-west direction. This represents the rate of change of elevation along the north-south direction. for n is the number of pixels within a local window centered on pixel i; θ i To determine the slope of pixel i in the monitoring area based on the target model, yes The size of the local window can be set according to the actual terrain conditions and accuracy requirements. For example, in areas with relatively gentle terrain changes or lower accuracy requirements, a larger window can be used to capture the overall trend; while in areas with drastic terrain changes or higher accuracy requirements, the window size needs to be reduced to more accurately reflect local changes. For instance, n is a 3x3 window formed by taking the eight neighboring pixels around pixel i as the center. The slope rate and aspect rate of pixel i are affected by the surrounding pixels, and the size of the local window can be set according to the actual situation; this application does not impose any restrictions on this.

[0097] When the target model is a second digital elevation model, the slope, aspect, slope rate, and aspect rate in the first slope information can be obtained through the above formula; when the target model is a second digital surface model, the slope, aspect, slope rate, and aspect rate in the second slope information can be obtained through the above formula.

[0098] Step 402: Train the random forest algorithm using a batch of training samples and their corresponding labels to obtain the accuracy correction model.

[0099] Step 403: Apply the second digital surface model, the second digital elevation model, the accuracy correction model, the first slope information and the second slope information of each pixel in the monitoring area to obtain the target digital surface model and the target digital elevation model of the monitoring area.

[0100] Specifically, the elevation data of the same pixel in the second digital surface model and the second digital elevation model, the first slope information corresponding to the pixel obtained based on the second digital elevation model, and the second slope information corresponding to the pixel obtained based on the second digital surface model can be input into the accuracy correction model. The output of the accuracy correction model is the corrected DEM data and DSM data of the pixel. The DEM data is the elevation data in the digital elevation model, and the DSM data is the elevation data in the digital surface model. The corrected DSM data corresponding to each pixel in the monitoring area can be used to form the target digital surface model, and the corrected DEM data corresponding to each pixel in the monitoring area can be used to form the target digital elevation model.

[0101] To remove abnormal elevation data and improve data reliability, such as Figure 3 As shown, in one embodiment, prior to step 401, the method further includes:

[0102] Step 001: Based on the second digital elevation model and the first digital elevation model, obtain the elevation error of each pixel in the local area.

[0103] For example, the local area includes: pixels 1 to 3; in the second digital elevation model, the elevation values ​​of pixels 1 to 3 are X1 to X3 respectively, and in the first digital elevation model, the elevation values ​​of pixels 1 to 3 are Y1 to Y3 respectively. Then the elevation errors of pixels 1 to 3 are Y1-X1, Y2-X2 and Y3-X3 respectively.

[0104] Step 002: Determine whether there are any abnormal pixels in the local area whose elevation error does not meet the preset error conditions. If so, delete the elevation data of the abnormal pixel in the second digital elevation model.

[0105] Preferably, the preset error condition is the three-standard-deviation rule.

[0106] To improve the accuracy of tree barrier risk monitoring, such as Figure 4 As shown, in one embodiment, step 500 includes:

[0107] Step 501: Perform differential calculation on the target digital surface model and the target digital elevation model to obtain the height difference of each pixel in the monitoring area.

[0108] Step 502: Using the height difference of each pixel and the pre-acquired multispectral image, determine the height and location of the trees in the monitoring area.

[0109] Specifically, satellite remote sensing stereo image pairs, in addition to panchromatic data from both forward and backward views, typically also contain a set of multispectral data. Here, multispectral data refers to optical image data containing at least four bands: red, green, blue, and near-infrared. That is, multispectral data can be extracted from the satellite remote sensing stereo image pairs to form the multispectral image. Multispectral image NDVI calculations often only distinguish whether an area is vegetated, not whether the vegetation is grassland, shrubs, or trees. Elevation difference can only determine if there is surface cover, not whether it is vegetation cover. This process can be automated using machine learning, using elevation difference and NDVI values ​​as sample features, and whether it is a tree as a label. Alternatively, the location of pixels whose elevation difference meets a preset tree height range and is vegetated in the pre-acquired multispectral image can be determined as the tree location, and the elevation difference of those pixels can be determined as the tree height. The preset tree height range can be set according to actual conditions, and this application does not impose any restrictions on it.

[0110] Step 503: Based on the tree height and location in the monitoring area, complete the tree obstacle risk monitoring of the monitoring area.

[0111] To further improve the accuracy of tree barrier risk monitoring, in one embodiment, step 503 includes:

[0112] Step 5031: Determine the disturbance range after the tree falls based on the tree height and location in the monitoring area.

[0113] Specifically, the disturbance range after the tree falls can be a circular range with the tree's location as the center and the tree's height as the radius.

[0114] Step 5032: Based on the tree height, tree location, and disturbance range after a tree falls in the monitoring area, determine whether the monitoring area meets the preset safety conditions. If so, the monitoring area is determined to be a safe area; otherwise, the monitoring area is determined to be a tree obstacle risk area. The preset distance conditions are: the tree height is less than the preset vertical safety distance, the tree location is outside the preset minimum horizontal safety distance range, and the disturbance range after a tree falls is outside the preset minimum horizontal safety distance range.

[0115] Specifically, the minimum horizontal safety distance range can represent the range where the distance to the overhead transmission line is less than or equal to a preset horizontal safety distance. The preset vertical safety distance, horizontal safety distance, and minimum horizontal safety distance range can all be set according to actual conditions, and this application does not impose any restrictions on them.

[0116] To further illustrate this solution, this application provides an application example of a tree obstacle risk monitoring method for overhead transmission lines, such as... Figure 5 As shown, the specific description is as follows:

[0117] Step 1: Acquire high-resolution optical satellite remote sensing stereo image pairs and UAV point cloud data, and perform data preprocessing.

[0118] In step 1, high-resolution optical satellite remote sensing stereo image pairs and UAV point cloud data are first acquired and preprocessed. Satellite stereo image pair data preprocessing includes steps such as geometric correction, radiometric correction, atmospheric correction, image stitching and cropping. UAV point cloud data preprocessing includes steps such as data filtering, filtering, coordinate transformation, accuracy checking, noise reduction, color assignment, and feature extraction.

[0119] Step 2: Based on the preprocessed point cloud data, obtain high-precision local DSM and DEM data of the monitoring area. Based on the preprocessed stereo image pairs, obtain lower-precision DSM and DEM data of the entire monitoring area. Use the random forest algorithm to apply the former to the adjustment and correction of the latter, thus obtaining high-precision, large-scale DSM and DEM data for the entire monitoring area. Step 2 includes:

[0120] Step 21: Based on the preprocessed point cloud data, obtain high-precision DSM and DEM data of the local monitoring area.

[0121] Specifically, multi-angle images captured by the camera on the drone can be used to generate point cloud data. After obtaining preprocessed drone point cloud data, the three-dimensional coordinates of each pixel can be extracted from these images through multi-view stereo matching technology, thereby constructing a high-precision DSM and DEM. Here, the high-precision DSM can be equivalent to the first digital surface model mentioned above, and the high-precision DEM can be equivalent to the first digital elevation model mentioned above.

[0122] Step 22: Based on the preprocessed stereo image pair data, obtain low-precision DSM and DEM data for the entire monitoring area. Here, the low-precision DSM can be equivalent to the second digital surface model mentioned above, and the low-precision DEM can be equivalent to the second digital elevation model mentioned above.

[0123] Furthermore, outlier elimination is performed according to the three-standard-deviation rule (3σ criterion). Based on previous research, the elevation error between the DEM obtained from stereo image pair data and the reference DEM follows a normal distribution. For DEM data obtained from stereo image pair data within a certain area, if the average elevation difference between it and the reference DEM to be processed (i.e., the DEM data extracted from UAV point clouds) is μ and the standard deviation is σ, then according to the 3σ criterion, the elevation error distribution should satisfy the following formula:

[0124] P(μ-3σ≤d≤μ+3σ)≈0.9973

[0125] Therefore, elevation data in the DEM obtained based on stereo image pairs whose corresponding elevation errors are not within the range of [μ-3σ,μ+3σ] are discarded as gross errors.

[0126] Step 23: Using the random forest algorithm, the former (i.e., locally high-precision DSM and DEM data) is used to adjust and correct the latter (i.e., lower-precision DSM and DEM data for the entire monitoring area) to obtain a high-precision, large-scale DSM and DEM for the entire monitoring area. Here, the high-precision, large-scale DSM can be equivalent to the aforementioned target digital surface model, and the high-precision, large-scale DEM can be equivalent to the aforementioned target digital elevation model. Step 23 mainly includes:

[0127] Step 231: Extract high-precision DSM and DEM from UAV point cloud data. These data are characterized by high density and high resolution.

[0128] Step 232: Use these high-precision DSMs and DEMs as control data and compare and analyze them with DSMs and DEMs acquired based on satellite stereo image pairs.

[0129] Step 233: Establish a relationship model between satellite data elevation errors and influencing factors such as terrain factors using the random forest algorithm. This model can be used to predict and correct elevation errors in satellite data, thereby improving its accuracy. Due to its good applicability and robustness, the random forest algorithm can effectively handle this complex regression problem and provide relatively accurate correction results. The adjusted and corrected satellite DSM and DEM data will have higher accuracy and reliability, making them suitable for large-scale topographic mapping and monitoring.

[0130] Specifically, the random forest model is constructed as follows:

[0131] The following formulas are used to calculate the slope, aspect, topographic relief, slope variability, and aspect variability within the monitoring area:

[0132]

[0133] Where Z is the elevation, and X and Y are the spatial coordinates in the east-west and north-south directions, respectively. and Z represents the rate of change of elevation along the grid; i It is the elevation value of the neighboring pixels. It is the average elevation; n is the number of pixels; θ i Set the slope value of the surrounding pixels. It is the average slope value.

[0134] Furthermore, the latitude and longitude coordinates, slope, aspect, slope rate, aspect rate, DSM and DEM data corresponding to the filtered stereo image pairs are extracted and read into a table as the geospatial features of the dataset. Then, the DSM and DEM values ​​of the UAV point cloud data at the corresponding locations are extracted. Next, the DSM and DEM values ​​of the UAV point cloud data are used as control points to construct a tabular dataset indexed by the latitude and longitude coordinates of the control points. In the tabular dataset, one row of data can represent a training sample and its corresponding label.

[0135] Furthermore, using the regression random forest classification algorithm as a framework, parameter optimization is performed. By combining latitude and longitude coordinates, slope, aspect, slope rate, and aspect rate, the optimal parameters are selected, and the importance of the optimal parameters is calculated. 70% of the features are selected as the training set and 30% as the test set. After the model is trained, DSM and DEM adjustment of the monitoring area are performed to obtain high-precision, large-scale DSM and DEM data.

[0136] Step 3: By performing differential calculations on the corrected DSM and DEM data, the height difference of each pixel is calculated. Combined with multispectral imagery, this effectively identifies tree types and their distribution, and simultaneously obtains the horizontal and vertical distances between the treetops and overhead power lines, as well as the disturbance range after tree fall. Step 3 includes:

[0137] Step 31: By performing differential calculations on the corrected DSM and DEM data, the height difference of each pixel in the monitoring area is calculated. The height difference can be equated to the height of ground features.

[0138] Step 32: By analyzing multispectral imagery, the reflectance characteristics of trees in different spectral bands are extracted. Tree height information from the DSM data is combined with the spectral features of the multispectral imagery. Machine learning algorithms such as Support Vector Machine (SVM) or Random Forest are used to learn the spectral and height characteristics of different tree types, and identification is performed accordingly. Finally, by comparing the actual height of trees with the spectral features in the multispectral imagery, the trees and their distribution can be accurately identified. The tree distribution indicates the location of the trees.

[0139] Step 33: Based on the height and distribution of the trees, obtain the disturbance range after the trees fall.

[0140] Step 4: Finally, based on the height of the trees, the distribution of the trees, and the disturbance range after the trees fall, dynamic monitoring and early warning of tree obstruction risk to overhead transmission lines are achieved.

[0141] Furthermore, based on the aforementioned extracted tree height, tree distribution, and disturbance range after tree fall, the following judgments can be made:

[0142] If the height of a tree exceeds the vertical safety distance, the area is considered to pose a safety hazard.

[0143] Based on the aforementioned extracted tree range distribution, neighborhood analysis is used to calculate the nearest horizontal distance between the overhead transmission line and the tree. Based on the minimum horizontal safety distance standard for tree obstacles of overhead transmission lines, it is determined whether the tree is outside the minimum horizontal safety distance range. If the nearest horizontal distance between the overhead transmission line and the tree is greater than the preset horizontal safety distance, it is determined that the tree is outside the minimum horizontal safety distance range.

[0144] Based on the extracted tree height and distribution data, the impact range of fallen vegetation is calculated. The disturbance range after a tree falls is a circular area centered at the intersection of the tree trunk and the ground, with a radius equal to the tree's height. Neighborhood analysis is used to calculate the closest distance between the overhead power line and this circle to determine if the fallen tree is outside the minimum horizontal safety distance. If the closest distance between the overhead power line and the circle is greater than the preset horizontal safety distance, then the fallen tree is determined to be outside the minimum horizontal safety distance.

[0145] If the height of the trees in the monitoring area is less than the vertical safety distance, the nearest horizontal distance between the overhead transmission line and the trees is greater than the preset horizontal safety distance, and the nearest distance between the overhead transmission line and the circle is greater than the preset horizontal safety distance, then the monitoring area can be determined to be safe.

[0146] To further illustrate this solution, this application provides a specific application example of a tree obstacle risk monitoring method for overhead transmission lines, as described below:

[0147] (1) Acquire satellite stereo image pair data. This application example takes 0.65M high-resolution stereo image pair data as an example. The 0.65M high-resolution stereo image pair data is preprocessed, including radiometric correction, noise reduction, histogram equalization, etc.

[0148] (2) Multi-angle images captured by the camera on the UAV can be used to generate point cloud data. The UAV point cloud data preprocessing includes steps such as data screening, filtering, coordinate transformation, accuracy checking, noise reduction, coloring and feature extraction.

[0149] (3) Based on the preprocessed point cloud data, high-precision DSM and DEM data of the local area of ​​the monitoring area are obtained. The location and size of the local area can be determined according to the actual situation, and this application does not impose any restrictions on this.

[0150] (4) Using a stereo matching algorithm, the corresponding points are automatically or semi-automatically matched on the stereo image pair data to obtain the three-dimensional coordinates of the ground points. These coordinates are then filtered and optimized to remove outliers and points with large errors. A DSM is generated based on the three-dimensional coordinates. Then, the ground points are extracted from the DSM using a ground point extraction algorithm to generate low-precision DEM data for the entire monitoring area.

[0151] (5) Based on the low-precision DSM and DEM data of the monitoring area, calculate the slope, aspect, topographic relief, slope variability, and aspect variability within the monitoring area using the following formulas:

[0152]

[0153] Where Z is the elevation, and X and Y are the spatial coordinates in the east-west and north-south directions, respectively. and Z represents the rate of change of elevation along the grid; i It is the elevation value of the neighboring pixels. It is the average elevation; n is the number of pixels; θ i It is the slope value of the surrounding pixels. It is the average slope value.

[0154] (6) Gross error removal: According to the three-standard-deviation rule (3σ criterion), based on previous research, the elevation error between the DEM obtained from stereo image pair data and the reference DEM follows a normal distribution. For DEM data obtained from stereo image pair data in a certain area, if the average elevation difference between it and the reference DEM to be processed, i.e., the DEM data extracted from UAV point clouds, is μ and the standard deviation is σ, then according to the 3σ criterion, the elevation error distribution should satisfy the following formula:

[0155] P(μ-3σ≤d≤μ+3σ)≈0.9973

[0156] Therefore, elevation data in the DEM obtained based on stereo image pairs whose corresponding elevation errors are not within the range of [μ-3σ, μ+3σ] are discarded as gross errors. The overlapping areas between the point cloud data and the stereo image pairs are used as the comparison areas.

[0157] (7) Extract the latitude and longitude coordinates, slope, slope aspect, slope rate, slope aspect rate, DSM and DEM data corresponding to the filtered stereo image pairs and read them into a table as the geospatial features of the dataset. Then extract the DSM and DEM values ​​of the UAV point cloud data at the corresponding location. Next, use the DSM and DEM values ​​of the UAV point cloud data as control points to construct a table dataset indexed by the latitude and longitude coordinates of the control points. The table dataset may include: DSM feature table and DEM feature table. Each row in the table represents a training sample and its corresponding label. In one example, the DSM feature table is shown in Table 3 and the DEM feature table is shown in Table 4.

[0158] Table 3

[0159]

[0160] Table 4

[0161]

[0162] (8) Using the regression random forest classification algorithm as a framework, parameter optimization is performed. The optimal parameters are selected by combining latitude and longitude coordinates, slope, aspect, slope rate, and aspect rate. The importance of the optimal parameters is calculated. 70% of the features are selected as the training set and 30% as the test set. After training the model, DSM and DEM adjustment of the monitoring area are performed to obtain high-precision and large-scale DSM and DEM data.

[0163] The input data consists of a table containing DSM and DEM values ​​from stereo image pairs, DSM and DEM values ​​corresponding to control points, elevation error between the two, latitude and longitude coordinates, slope, aspect, slope rate, and aspect rate. The Random Forest method first selects one or more features with the highest correlation from the numerous features in the input table (and optimizes parameters). Then, it fits these features to the DSM and DEM data of the point cloud separately. Using 70% of the data in the table as the training set and 30% as the test set, an optimal model is trained. This allows for DSM and DEM adjustment of the monitoring area by inputting raster data of the DSM and DEM of the entire monitoring area. By performing difference calculations on the corrected DSM and DEM data, the height difference of each pixel in the monitoring area is calculated.

[0164] (9) By analyzing multispectral images, reflectance characteristics of trees in different spectral bands are extracted. The height difference of each pixel in the monitoring area is combined with the spectral characteristics of the multispectral images. Machine learning algorithms such as Support Vector Machine (SVM) or Random Forest are used to learn the spectral and height characteristics of different tree types, and identification is performed accordingly. Finally, by comparing the actual height of the trees with the spectral characteristics in the multispectral images, the trees and their distribution range can be accurately identified. Trees are taller than grasslands and shrubs and pose a greater threat to power transmission lines; therefore, subsequent power transmission line disturbance analysis focuses on trees.

[0165] For example: if the height difference is DSM-DEM = 0, then there is no ground cover; if the height difference is DSM-DEM = 10, there is ground cover, and the optical image shows vegetation, then it is highly likely that there are trees; if the optical image shows a building, then there is a 10m high building; if the height difference is DSM-DEM = 0.2m, there is ground cover, and the optical image shows vegetation, then it is highly likely that there is grassland cover.

[0166] (10) Based on the height and distribution of the trees, the disturbance range after the trees fall is obtained.

[0167] (11) Based on the height of the trees, the distribution of the trees, and the disturbance range after the trees fall, complete the tree obstacle risk monitoring of the overhead transmission lines in the monitoring area.

[0168] From a software perspective, in order to improve the accuracy of tree obstruction risk monitoring for large-scale overhead transmission lines, this application provides an embodiment of an overhead transmission line tree obstruction risk monitoring system for implementing all or part of the aforementioned overhead transmission line tree obstruction risk monitoring method. See [link to relevant documentation]. Figure 6 The tree obstacle risk monitoring system for the overhead transmission line specifically includes the following components:

[0169] The first acquisition device 01 is used to acquire satellite remote sensing stereo image pairs of the monitoring area and UAV point cloud data of a local area in the monitoring area.

[0170] The first obtaining device 02 is used to obtain a first digital surface model and a first digital elevation model of the local area based on the UAV point cloud data.

[0171] The second obtaining device 03 is used to obtain a second digital surface model and a second digital elevation model of the monitoring area based on the satellite remote sensing stereo image pair data;

[0172] The calibration device 04 is used to apply the first digital surface model and the first digital elevation model to calibrate the second digital surface model and the second digital elevation model to obtain the target digital surface model and the target digital elevation model of the monitoring area.

[0173] Monitoring device 05 is used to monitor the tree obstruction risk of overhead transmission lines in the monitoring area based on the target digital surface model and the target digital elevation model.

[0174] In one embodiment, the correction device includes:

[0175] The acquisition module is used to acquire batch training samples and their corresponding labels. Each training sample includes: elevation data of a pixel in the local region in the second digital elevation model and the second digital surface model, first slope information corresponding to the pixel obtained based on the second digital elevation model, and second slope information corresponding to the pixel obtained based on the second digital surface model. The label corresponding to each training sample includes: elevation data of the pixel corresponding to the training sample in the first digital surface model and the first digital elevation model, respectively. The pixels corresponding to each training sample are different. The first slope information and the second slope information both include at least one of: slope, aspect, slope rate, and aspect rate.

[0176] The training module is used to train the random forest algorithm using a batch of training samples and their corresponding labels to obtain an accuracy correction model.

[0177] The application module is used to apply the slope information of each pixel in the second digital surface model, the second digital elevation model, and the accuracy correction model to obtain the target digital surface model and the target digital elevation model of the monitoring area.

[0178] In one embodiment, the tree obstacle risk monitoring system for overhead transmission lines further includes:

[0179] The second acquisition device is used to obtain the elevation error of each pixel in the local area based on the second digital elevation model and the first digital elevation model.

[0180] The deletion device is used to determine whether there are any abnormal pixels in the local area whose elevation errors do not meet the preset error conditions. If so, the elevation data of the abnormal pixel in the second digital elevation model is deleted.

[0181] In one embodiment, the preset error condition is a three-standard-deviation rule.

[0182] In one embodiment, the monitoring device includes:

[0183] The differential calculation module is used to perform differential calculations on the target digital surface model and the target digital elevation model to obtain the height difference of each pixel in the monitoring area;

[0184] The determination module is used to determine the height and location of trees in the monitoring area by applying the height difference of each pixel and the pre-acquired multispectral image.

[0185] The monitoring module is used to monitor tree obstacle risks in the monitoring area based on the height and location of the trees.

[0186] In one embodiment, the monitoring module includes:

[0187] The determining unit is used to determine the disturbance range after a tree falls, based on the tree height and location in the monitoring area.

[0188] The monitoring unit is used to determine whether the monitoring area meets the preset safety conditions based on the tree height, tree location, and disturbance range after the tree falls. If so, the monitoring area is determined to be a safe area; otherwise, the monitoring area is determined to be a tree obstacle risk area.

[0189] The preset distance conditions are as follows: the height of all trees is less than the preset vertical safety distance, the location of the trees is outside the preset minimum horizontal safety distance range, and the disturbance range after the trees fall is outside the preset minimum horizontal safety distance range.

[0190] The embodiments of the tree obstacle risk monitoring system for overhead transmission lines provided in this specification can be used to execute the processing flow of the embodiments of the above-described tree obstacle risk monitoring method for overhead transmission lines. Its functions will not be repeated here, but can be referred to the detailed description of the embodiments of the above-described tree obstacle risk monitoring method for overhead transmission lines.

[0191] As described above, the tree obstacle risk monitoring method and system for overhead transmission lines provided in this application can solve the problem that traditional tree obstacle monitoring methods are often limited by a single data source, making it difficult to comprehensively and accurately reflect the surrounding environment of transmission lines. By integrating the high precision of UAV point cloud data with the wide coverage advantage of satellite imagery, high-precision terrain data around long-distance overhead transmission lines can be accurately extracted. This fusion of multi-source data makes the judgment of the spatial relationship between trees and power lines more accurate, thereby significantly improving the accuracy of tree obstacle risk monitoring and early warning, and effectively reducing the risk of safety accidents caused by inaccurate monitoring. It also shows significant effectiveness in cost control, solving the problem of high costs associated with relying on expensive UAVs equipped with lidar technology for large-area monitoring. By introducing satellite imagery and utilizing its wide-area advantage, the high-intensity, large-area flight operations of UAVs are reduced. While ensuring monitoring effectiveness, resources are rationally allocated, unnecessary cost investment is avoided, production costs are effectively reduced, and economic efficiency is improved, making large-scale, routine tree obstacle monitoring possible. From the perspective of application value and application prospects, the embodiments of this application possess powerful dynamic monitoring and early warning capabilities. This system can acquire real-time or periodic information on changes in the environment surrounding power transmission lines, promptly identify potential tree obstruction risks and issue early warnings, buying valuable time for power departments to take preventative measures and effectively ensuring the stability of power supply. Its comprehensive advantages make this solution of significant practical value in the field of power transmission line tree obstruction monitoring, and it can be widely applied in the power industry, promoting technological upgrades and improving safety standards across the entire sector.

[0192] Figure 7 This is a schematic diagram of the physical structure of an electronic device provided in an embodiment of the present invention, such as... Figure 7 As shown, the electronic device includes: a memory 701, a processor 702, and a computer program stored in the memory 701 and executable on the processor 702. When the processor 702 executes the computer program, it implements the following method:

[0193] Acquire satellite remote sensing stereo image pairs of the monitored area and UAV point cloud data of local areas within the monitored area;

[0194] Based on the UAV point cloud data, a first digital surface model and a first digital elevation model of the local area are obtained;

[0195] Based on the satellite remote sensing stereo image pair data, a second digital surface model and a second digital elevation model of the monitored area are obtained;

[0196] The second digital surface model and the second digital elevation model are corrected by applying the first digital surface model and the first digital elevation model to obtain the target digital surface model and the target digital elevation model of the monitoring area;

[0197] Based on the target digital surface model and the target digital elevation model, tree obstacle risk monitoring of overhead transmission lines in the monitoring area is completed.

[0198] This embodiment discloses a computer program product, which includes a computer program that, when executed by a processor, implements the following method:

[0199] Acquire satellite remote sensing stereo image pairs of the monitored area and UAV point cloud data of local areas within the monitored area;

[0200] Based on the UAV point cloud data, a first digital surface model and a first digital elevation model of the local area are obtained;

[0201] Based on the satellite remote sensing stereo image pair data, a second digital surface model and a second digital elevation model of the monitored area are obtained;

[0202] The second digital surface model and the second digital elevation model are corrected by applying the first digital surface model and the first digital elevation model to obtain the target digital surface model and the target digital elevation model of the monitoring area;

[0203] Based on the target digital surface model and the target digital elevation model, tree obstacle risk monitoring of overhead transmission lines in the monitoring area is completed.

[0204] This embodiment provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the following method:

[0205] Acquire satellite remote sensing stereo image pairs of the monitored area and UAV point cloud data of local areas within the monitored area;

[0206] Based on the UAV point cloud data, a first digital surface model and a first digital elevation model of the local area are obtained;

[0207] Based on the satellite remote sensing stereo image pair data, a second digital surface model and a second digital elevation model of the monitored area are obtained;

[0208] The second digital surface model and the second digital elevation model are corrected by applying the first digital surface model and the first digital elevation model to obtain the target digital surface model and the target digital elevation model of the monitoring area;

[0209] Based on the target digital surface model and the target digital elevation model, tree obstacle risk monitoring of overhead transmission lines in the monitoring area is completed.

[0210] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied 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.

[0211] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0212] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0213] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0214] In the description of this specification, the references to terms such as "an embodiment," "a specific embodiment," "some embodiments," "for example," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0215] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are 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 within the scope of protection of the present invention.

Claims

1. A method for monitoring tree obstruction risk of overhead transmission lines, characterized in that, include: Acquire satellite remote sensing stereo image pairs of the monitored area and UAV point cloud data of local areas within the monitored area; Based on the UAV point cloud data, a first digital surface model and a first digital elevation model of the local area are obtained; Based on the satellite remote sensing stereo image pair data, a second digital surface model and a second digital elevation model of the monitored area are obtained; The second digital surface model and the second digital elevation model are corrected by applying the first digital surface model and the first digital elevation model to obtain the target digital surface model and the target digital elevation model of the monitoring area; Based on the target digital surface model and the target digital elevation model, tree obstacle risk monitoring of overhead transmission lines in the monitoring area is completed.

2. The tree obstacle risk monitoring method for overhead transmission lines according to claim 1, characterized in that, The step of applying the first digital surface model and the first digital elevation model to correct the second digital surface model and the second digital elevation model to obtain the target digital surface model and the target digital elevation model of the monitoring area includes: A batch of training samples and their corresponding labels are collected. Each training sample includes: elevation data of a pixel in the local region in the second digital elevation model and the second digital surface model, first slope information corresponding to the pixel obtained based on the second digital elevation model, and second slope information corresponding to the pixel obtained based on the second digital surface model. The label corresponding to each training sample includes: elevation data of the pixel corresponding to the training sample in the first digital surface model and the first digital elevation model, respectively. The pixels corresponding to each training sample are different. The first slope information and the second slope information both include at least one of slope, aspect, slope rate, and aspect rate. The random forest algorithm is trained using a batch of training samples and their corresponding labels to obtain an accuracy correction model; By applying the second digital surface model, the second digital elevation model, the accuracy correction model, and the first slope information and the second slope information of each pixel in the monitoring area, the target digital surface model and the target digital elevation model of the monitoring area are obtained.

3. The tree obstacle risk monitoring method for overhead transmission lines according to claim 2, characterized in that, Before collecting batch training samples and their corresponding labels, the process also includes: Based on the second digital elevation model and the first digital elevation model, the elevation error of each pixel in the local area is obtained; Determine whether there are any abnormal pixels in the local area whose elevation error does not meet the preset error conditions. If so, delete the elevation data of the abnormal pixel in the second digital elevation model.

4. The tree obstacle risk monitoring method for overhead transmission lines according to claim 3, characterized in that, The preset error condition is the three-standard-deviation rule.

5. The tree obstacle risk monitoring method for overhead transmission lines according to claim 1, characterized in that, The step of monitoring tree obstruction risks of overhead transmission lines in the monitoring area based on the target digital surface model and the target digital elevation model includes: The height difference of each pixel in the monitoring area is obtained by performing differential calculations on the target digital surface model and the target digital elevation model. By applying the height difference of each pixel and the pre-acquired multispectral image, the height and location of the trees in the monitoring area are determined. Based on the height and location of trees in the monitoring area, tree obstacle risk monitoring is completed in the monitoring area.

6. The tree obstacle risk monitoring method for overhead transmission lines according to claim 5, characterized in that, The process of monitoring tree risk in the monitoring area based on tree height and location includes: The disturbance range after a tree falls is determined based on the tree height and location in the monitoring area. Based on the tree height, tree location, and disturbance range after a tree falls in the monitoring area, determine whether the monitoring area meets the preset safety conditions. If so, the monitoring area is determined to be a safe area; otherwise, the monitoring area is determined to be a tree obstacle risk area. The preset distance conditions are as follows: the height of all trees is less than the preset vertical safety distance, the location of the trees is outside the preset minimum horizontal safety distance range, and the disturbance range after the trees fall is outside the preset minimum horizontal safety distance range.

7. A tree obstruction risk monitoring system for overhead transmission lines, characterized in that, include: The first acquisition device is used to acquire satellite remote sensing stereo image pairs of the monitoring area and UAV point cloud data of a local area in the monitoring area; The first obtaining device is used to obtain a first digital surface model and a first digital elevation model of the local area based on the UAV point cloud data. The second obtaining device is used to obtain a second digital surface model and a second digital elevation model of the monitored area based on the satellite remote sensing stereo image pair data; A calibration device is used to apply the first digital surface model and the first digital elevation model to calibrate the second digital surface model and the second digital elevation model, so as to obtain the target digital surface model and the target digital elevation model of the monitoring area. The monitoring device is used to monitor the tree obstruction risk of overhead transmission lines in the monitoring area based on the target digital surface model and the target digital elevation model.

8. The tree obstacle risk monitoring system for overhead transmission lines according to claim 7, characterized in that, The correction device includes: The acquisition module is used to acquire batch training samples and their corresponding labels. Each training sample includes: elevation data of a pixel in the local region in the second digital elevation model and the second digital surface model, first slope information corresponding to the pixel obtained based on the second digital elevation model, and second slope information corresponding to the pixel obtained based on the second digital surface model. The label corresponding to each training sample includes: elevation data of the pixel corresponding to the training sample in the first digital surface model and the first digital elevation model, respectively. The pixels corresponding to each training sample are different. The first slope information and the second slope information both include at least one of: slope, aspect, slope rate, and aspect rate. The training module is used to train the random forest algorithm using a batch of training samples and their corresponding labels to obtain an accuracy correction model. The application module is used to apply the slope information of each pixel in the second digital surface model, the second digital elevation model, and the accuracy correction model to obtain the target digital surface model and the target digital elevation model of the monitoring area.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the tree obstacle risk monitoring method for overhead transmission lines as described in any one of claims 1 to 6.

10. A computer-readable storage medium storing computer instructions thereon, characterized in that, When the instruction is executed by the processor, it implements the tree obstacle risk monitoring method for overhead transmission lines as described in any one of claims 1 to 6.

Images