Automatic identification system and method for tree obstacles in distribution network lines based on satellite remote sensing

Abstract

This invention discloses a system and method for automatically identifying tree obstacles on distribution network lines based on satellite remote sensing. This system, which belongs to the field of power system operation and maintenance technology, includes a satellite remote sensing data acquisition module, a data processing module, a tree obstacle identification and analysis module, an early warning and decision support module, and a user interaction interface module. This invention can significantly improve the efficiency of tree obstacle management on distribution network lines and reduce the risk of power failures caused by tree obstacles, thus having important significance for ensuring the stable operation of power systems.

Description

Technical Field

[0001] The present invention relates to the technical field of power system operation and maintenance, and more specifically, to a system and method for automatically identifying tree obstacles in distribution network lines based on satellite remote sensing. Background Art

[0002] With the continuous development of power systems, the safe and stable operation of distribution lines is becoming increasingly important for ensuring the reliability of power supply. Distribution lines often pass through forested areas, where rapid tree growth can easily cause tree barrier failures, posing a significant risk to the stable operation of the power grid. Traditional tree barrier monitoring methods rely primarily on manual inspections and drone patrols. Manual inspections are inefficient, time-consuming, and labor-intensive, while drone patrols also present some problems. First, the operating environment of distribution lines is complex, and some areas are restricted by factors such as trust and airspace, making drone operations impossible. Second, channel modeling based on drone patrol data is expensive, requiring at least a year to create a model. The current maximum modeling frequency for distribution line tree barriers cannot keep up with the growth cycle of these barriers. Therefore, large-scale surveys of trees along distribution line corridors are hindered, and the development of an efficient and automatic tree barrier identification system has important practical application value.

[0003] Satellite remote sensing technology has the advantages of wide coverage, stable update cycle, no restrictions on environmental conditions, less manpower, and low cost. Compared with drone inspections, it can quickly realize large-scale and business-oriented inspections of distribution channels, greatly improving the efficiency and pertinence of tree barrier screening. Summary of the Invention

[0004] The purpose of the present invention is to overcome the shortcomings of the existing technology and provide a distribution network line tree barrier automatic identification system and method based on satellite remote sensing, which can realize automatic monitoring and identification of tree growth along the distribution network line, and improve the efficiency and response speed of tree barrier management.

[0005] The object of the present invention is achieved through the following solutions:

[0006] A distribution network line tree obstacle automatic identification system based on satellite remote sensing, including a satellite remote sensing data acquisition module, a data processing module, a tree obstacle identification and analysis module, an early warning and decision support module and a user interaction interface module;

[0007] The satellite remote sensing data acquisition module is used to collect distribution network line image data and distribution network line tower latitude and longitude precise coordinate data using remote sensing satellites;

[0008] The data processing module is used to pre-process the data collected by the satellite remote sensing data collection module and extract tree information related to the distribution network line;

[0009] The tree barrier identification and analysis module is used to identify and evaluate potential tree barriers based on the growth characteristics of trees and the safety distance requirements of distribution network lines;

[0010] The early warning and decision support module is used to summarize the tree obstacle information identified by the tree obstacle identification and analysis module, classify the tree obstacles according to their severity, and analyze them in association with distribution network equipment and facilities;

[0011] The user interaction interface module is used to provide a user interface to display the distribution of tree barriers on the distribution network lines, so that users can query, browse and export tree barrier information.

[0012] Furthermore, the satellite remote sensing data acquisition module is also used to collect vegetation coverage along the distribution network line. The image acquisition range needs to cover the designated distribution line channel, with the line as the center and a buffer zone formed by a set distance extending to the left and right.

[0013] Furthermore, the preprocessing includes image denoising, enhancement and classification, and a conventional data processing process is formulated with reference to national standards. Based on the acquired satellite original images, satellite standard products that have passed quality inspection are obtained through radiometric calibration, geometric correction, atmospheric correction and image mosaicking.

[0014] Furthermore, the extraction of tree information related to the distribution network lines specifically includes: using a module for rapid pixel labeling of regions of interest (ROIs) of arbitrary shapes to construct a labeling specification for tree distribution identification; after obtaining the labeling specification, performing data labeling based on the specification and establishing a tree distribution identification sample database; wherein, using a visual interpretation method and combining existing ground survey data, a tree sample set is constructed on the collected high-resolution satellite remote sensing images, and the image features of the trees are extracted as training samples, and the image features of the trees include spectral features and texture features.

[0015] Furthermore, the module for quickly marking pixels of ROI of arbitrary shape specifically includes:

[0016] The region of interest (ROI) selection module is used to determine the boundaries based on the region geometry, giving priority to regions with clear outlines and separation from other regions;

[0017] Arbitrary shape ROI decomposition module, used to quickly detect concave points on the boundary, and decompose the arbitrary shape polygon area into multiple convex hull areas according to the detected concave points, and then calculate each convex hull area separately;

[0018] The single-category pixel labeling module is used to perform the following process: a digital image containing K objects is represented by the following model:

[0019]

[0020] Where z represents the image pixel, P(ω i ) and p(z|ω i ) is the prior probability and conditional probability density of the i-th type of target; for RGB images, pixel z=(r,g,b)'; the color components on each channel are independent of each other, that is:

[0021]

[0022] To facilitate visualization, the pixel distribution is spread over the three RGB color channels;

[0023] By extracting pixels within ROI of arbitrary shape and training them through SVDD classifier, classification and recognition of the entire image can be achieved.

[0024] Furthermore, the spectral features include normalized vegetation index, difference vegetation index, enhanced vegetation index, ratio vegetation index, green ratio vegetation index, greenness index, adjusted soil-adjusted vegetation index and red-shortwave infrared band difference; the texture features include angular second moment, contrast, correlation, heterogeneity, homogeneity, mean, standard deviation and entropy.

[0025] Furthermore, the tree barrier identification and analysis module specifically includes a tree barrier identification and evaluation module based on a machine learning algorithm.

[0026] Furthermore, the user interaction interface module is specifically used to output the identification and evaluation results in the form of graphics and reports, and provide them to power grid operation and maintenance personnel for decision support.

[0027] A method for automatically identifying tree obstacles on distribution network lines based on satellite remote sensing, based on the system described in any one of the above items, wherein the tree obstacle identification and analysis module performs the following steps:

[0028] S11, based on high-resolution satellite remote sensing images, build a tree distribution recognition model: using a classification method based on support vector machine (SVM),

[0029] The radial basis function kernel function RBF is selected as the kernel function of SVM. The SVM classifier with RBF as the kernel function has two adjustable parameters, namely the kernel function parameter σ2 and the error penalty factor C. The kernel function parameter σ2 affects the complexity of the sample data distribution in the high-dimensional feature space. The error penalty factor C adjusts the confidence interval and empirical risk ratio of the learning machine in the determined feature space, and uses σ2 to map the data to the high-dimensional feature space. Then, the C value is found in this high-dimensional feature space to make the confidence range and empirical risk ratio of the SVM at the optimal level, thereby obtaining the global optimal SVM classifier.

[0030] S12, perform iterative model training and optimization verification: input the constructed training sample set into the SVM classifier, and use the loss function to iteratively train the classifier. When the loss value converges, retain the final model; then input the feature set of the high-resolution image into the trained model, use the SVM classifier to determine the tree distribution range, and output the tree recognition result based on the high-resolution image.

[0031] A method for automatically identifying tree obstacles on distribution network lines based on satellite remote sensing, based on the system described in any one of the above items, wherein the early warning and decision support module performs the following steps:

[0032] S21, Distribution Channel Tree Distribution Risk Level Classification: Based on the distance between trees and distribution lines, the distribution channel tree distribution risk is classified into three levels: high, medium, and low, and the risk assessment criteria are determined;

[0033] S22, determining the tree distribution identification results and overlaying them with the distribution network line data in the GIS: the extracted results reflect the number and distribution of trees, and the tree distribution identification results are associated with the distribution network equipment and facility information by defining associated fields;

[0034] S23, overlay analysis of tree identification results with distribution network line data in GIS: Perform spatial overlay and association analysis on the tree extraction results and power grid GIS data, associate each tree extraction result with a line tower, and calculate the closest distance to the conductor and tower. Use fields to represent the relative position of the tree and the nearest tower, and finally display the association results.

[0035] The beneficial effects of the present invention include:

[0036] The system and method provided by the present invention can greatly improve the efficiency of tree barrier management of distribution network lines, reduce the risk of power failures caused by tree barriers, and are of great significance to ensuring the stable operation of the power system.

[0037] The present invention provides a novel automatic identification system and method for tree barriers on distribution network lines. By integrating satellite remote sensing technology and machine learning algorithms, it realizes efficient and automatic monitoring and identification of tree barriers on distribution network lines. It has broad application prospects and important social and economic value. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0039] Figure 1 is a pixel labeling diagram in an embodiment of the present invention;

[0040] Figure 2 This is a distribution diagram of line hidden dangers in an embodiment of the present invention. DETAILED DESCRIPTION

[0041] All features disclosed in all embodiments in this specification, or steps in all methods or processes implicitly disclosed, except for mutually exclusive features and / or steps, can be combined and / or expanded or replaced in any manner.

[0042] The specific implementation process of the present invention includes the following parts:

[0043] The satellite remote sensing data acquisition module is used to perform the following processes: (1) Select a specific line and provide the precise coordinates of the longitude and latitude of the distribution line tower. (2) Optimize the remote sensing satellite. Compared with the common multi-spectral bands, the imaging payload spectrum of Beijing-3N satellite covers the wavelength range from the deep blue band (minimum 0.416μm) to the near infrared band (maximum 0.888μm). Among them, the deep blue spectrum has strong water penetration and higher atmospheric scattering characteristics, which is conducive to distinguishing different ground objects in the shadow. The red edge spectrum is suitable for vegetation monitoring and agricultural-related remote sensing applications. When the vegetation canopy is closed, the NDVI will be saturated. The red edge spectrum can distinguish different vegetation canopy densities. (3) Use high-resolution satellite images to obtain the vegetation coverage along the distribution network line. The image acquisition range must cover the designated distribution line channel, with the line as the center and a buffer zone extending 1 km to the left and right.

[0044] The data processing module is used to perform the following processes: pre-processing the collected remote sensing data, including image denoising, enhancement and classification, and extracting tree information related to the distribution network line. Among them, (1) remote sensing image pre-processing: referring to national standards to formulate a conventional data processing process, based on the acquired satellite original image, through radiometric calibration, geometric correction, atmospheric correction, image mosaicking and other processing, obtain satellite standard products that have passed quality inspection. (2) tree image feature sample set construction: using arbitrary shape region of interest (ROI) pixel fast labeling algorithm to construct the annotation specification for tree distribution identification. The production of the data set requires: 1) labeling the tree distribution area data in the remote sensing image; 2) converting the labeling results into a format that can be recognized by deep learning. Therefore, after obtaining the labeling specification, data labeling is carried out based on the above specifications and a tree distribution identification sample database is established. Specifically, there can be many annotation shapes, commonly used are Polygons, Rectangles, and Circles, which can appear in the same image at the same time. The appropriate annotation shape is selected according to the shape of the target. Polygons are used for annotation. First, using visual interpretation and existing ground survey data, a tree sample set was constructed from high-resolution satellite remote sensing imagery. Image features of the trees were extracted as training samples. The tree image feature set primarily includes the following: 1) Spectral features: These include the Normalized Difference Vegetation Index (NDVI), Difference Vegetation Index (DVI), Enhanced Vegetation Index (EVI), Ratio Vegetation Index (RVI), Green Ratio Vegetation Index (GRVI), Greenness Index (GI), Adjusted Soil Adjusted Vegetation Index (MSAVI), and RedSWIR1. 2) Texture features: A gray-level co-occurrence matrix was used to calculate the relationship between pixel values, including angular second moment, contrast, correlation, heterogeneity, homogeneity, mean, standard deviation (StdDev), and entropy.

[0045] Among them, the fast labeling process of pixels in the region of interest (ROI) of arbitrary shape, such as Figure 1 Specifically, it includes the following contents:

[0046] (1) Region of interest (ROI) selection: ROI selection is an important step in image preprocessing. The quality and speed of ROI selection will directly affect the demarcation of image boundaries, the segmentation of image regions, and the scale of subsequent data processing. There are many ROI selection methods. The method of determining the boundary based on the geometric shape of the region is the most intuitive. Regular shapes such as circles, triangles, rectangles, and polygons are commonly used. According to function approximation theory, the boundary of an arbitrary shape region can always be approximated by a broken line, and when the boundary is closed, the broken line just forms a polygon. Considering the non-fixed shape of vegetation targets, regions with clear outlines and separated from other regions are preferred.

[0047] (2) Arbitrary shape ROI decomposition: The decomposition process of arbitrary shape ROI includes the following two points: one is to quickly detect the concave points on the boundary; the other is to decompose the arbitrary shape polygon area into multiple convex hull areas according to the detected concave points, and then calculate each convex hull area separately.

[0048] (4) Single-category pixel annotation: A digital image containing K objects can be represented by the following model:

[0049]

[0050] Where z represents the image pixel, P(ωi) and p(z|ωi) are the prior probability and conditional probability density of the i-th type of target. For RGB images, pixel z = (r, g, b)'. According to the principle of image formation, the color components on each channel are independent of each other, that is:

[0051]

[0052] To facilitate visualization, the pixel distribution is spread over the three RGB color channels.

[0053] By extracting pixels within an arbitrary ROI, the SVDD classifier is trained to achieve full image classification and recognition. The idea of ​​this algorithm is to determine whether the pixel value of any point is close to the pixel value within the ROI and then determine whether they belong to the same category. The judgment method is to determine the positional relationship between the point and the convex hull.

[0054] The tree barrier identification and analysis module is used to perform the following process: based on the growth characteristics of trees and the safety distance requirements of distribution network lines, it uses machine learning algorithms to identify and evaluate potential tree barriers. In the method provided by the present invention, the tree barrier identification and analysis module specifically includes the following process:

[0055] (1) A tree distribution recognition model is constructed based on high-resolution satellite remote sensing images. A classification method based on support vector machines (SVM) is used. SVM is a supervised classification algorithm. Its general idea is: assuming there are two types of points in the sample space, we want to find a dividing hyperplane to separate the two types of samples. The dividing hyperplane should be selected with the best generalization ability, that is, it can maximize the distance between the sample points closest to it in the two types of samples.

[0056] The core of SVM lies in the choice of kernel function. Due to its superior performance, the radial basis function (RBF) kernel function is often chosen as the kernel function of SVM in practical applications, as shown below.

[0057]

[0058] As can be seen, the SVM classifier with the RBF kernel function has two adjustable parameters: the kernel parameter σ2 and the error penalty factor C. The kernel parameter σ2 primarily affects the complexity of the sample data distribution in the high-dimensional feature space, while the error penalty factor C adjusts the confidence interval and empirical risk ratio of the learning machine in a given feature space. Therefore, to obtain an SVM classifier with optimal generalization ability, it is first necessary to use an appropriate σ2 to map the data to a suitable high-dimensional feature space. Then, within this high-dimensional feature space, find a suitable C value to optimize the SVM's confidence interval and empirical risk ratio, thereby obtaining the globally optimal SVM classifier.

[0059] (2) Perform iterative model training and optimization verification. Based on the constructed training sample set, input it into the SVM classifier and iteratively train the classifier using the loss function. When the loss value converges, retain the final model. Input the feature set of the high-resolution image into the trained model. The SVM classifier automatically determines the tree distribution range and outputs the tree recognition results based on the high-resolution image.

[0060] The early warning and decision support module specifically includes the following processes: summarizing the identified tree obstacle information, classifying it according to the severity of the tree obstacle, and automatically correlating and analyzing it with distribution network equipment and facilities to provide decision support for operation and maintenance personnel, and achieve focused monitoring and rapid response to key areas.

[0061] In the method provided by the present invention, the early warning and decision support module specifically includes the following processes:

[0062] Step 1: Classify the risk level of tree distribution in distribution corridors. Based on the distance between trees and distribution lines, classify the risk level of tree distribution in distribution corridors into three levels: high, medium, and low, and determine the risk assessment criteria.

[0063] Table 1 Risk levels of tree distribution in distribution networks

[0064]

[0065] Step 2: Determine the tree distribution identification results and perform overlay analysis with the distribution network line data in GIS. The results extracted intelligently in the previous section can reflect the number and distribution of trees, but the results have not yet been associated with the distribution network data. However, for the distribution network tree census, associating the tree census results with the distribution network data is the focus of the tree census and can better reflect the significance of the census. Therefore, it is necessary to associate the tree distribution identification results with the distribution network equipment and facilities information to facilitate the verification and proofreading of the front-line team members. In order to fully realize the automatic integration of the two, the definition of the association fields between the tree distribution identification results and the distribution network equipment and facilities information is the key to the problem. The project plans to define the associated fields as shown in the following table:

[0066] Table 2 Related Field Description

[0067]

[0068] The following is a detailed description of the associated fields in Table 2:

[0069] 1) Line name: refers to the name of the line closest to the geometric center point of the hidden danger map. This field binds the hidden danger map to the line.

[0070] 2) Nearest tower number: refers to the tower number from the geometric center point of the hidden danger map to the nearest tower. This field binds the hidden danger map to the specific tower point on the line.

[0071] 3) Line distance: This refers to the shortest distance to the nearest line. If the intersection of the perpendicular line from the geometric center of the pattern to the line falls on the line, the shortest distance is the vertical distance. If the intersection of the perpendicular line from the geometric center of the pattern to the line falls on the extension line of the line, the line distance is the distance from the geometric center of the pattern to the nearest tower.

[0072] 4) Pole tower distance: refers to the straight-line distance between the geometric center point of the hidden danger map and the nearest pole tower.

[0073] 5) Risk level: Based on the line distance field and combined with the tree risk level standard, the risk level of each hidden danger map is automatically calculated.

[0074] 6) Large and small sides: With the nearest tower as the center, if the hidden danger map is between the nearest tower and the next large tower, it belongs to the small side; if the hidden danger map is between the nearest tower and the previous small tower, it belongs to the large side.

[0075] Step 3: If Figure 2As shown, the tree identification results are overlaid and analyzed with the distribution network line data in the GIS. Spatially overlaying and analyzing the tree extraction results and the power grid GIS data, each tree extraction result is associated with a line tower. The closest distance to the conductor and tower is automatically calculated, and the relative position of the tree to the nearest tower is indicated using the large and small side fields. Finally, the association results are displayed in a table.

[0076] The user interface module provides an intuitive user interface that displays the distribution of tree barriers along distribution lines. Users can query, browse, and export tree barrier information, facilitating management review and decision-making. The module assesses the potential risk of tree barriers to distribution lines based on factors such as tree canopy width, species, and distance from the lines. Identification and assessment results are output as graphs and reports for decision-making support by grid operations and maintenance personnel.

[0077] The above description is merely the technical principles and preferred embodiments used in the present invention. Those skilled in the art will appreciate that the present invention is not limited to the specific embodiments described herein. It is obvious that various changes, adjustments, and substitutions can be made by those skilled in the art without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in more detail through the above embodiments, the present invention is not limited to the above embodiments. Without departing from the principles and concepts of the present invention, it may also include more other equivalent embodiments, and the scope of the present invention is determined by the scope of the appended claims.

[0078] The units involved in the embodiments of the present invention are implemented based on hardware, and some functions can be implemented by software. The units described can also be set in a processor. In some cases, the names of these units do not constitute limitations on the units themselves. The embodiments of the present invention can provide corresponding computer program products or computer programs, which include computer instructions, and the computer instructions are stored in corresponding hardware devices. The corresponding hardware devices read and execute the computer instructions, so that the hardware devices perform the methods provided in the various optional implementations described above.

Claims

1. A distribution network line tree obstacle automatic identification system based on satellite remote sensing, characterized in that: It includes satellite remote sensing data acquisition module, data processing module, tree obstacle identification and analysis module, early warning and decision support module and user interaction interface module; The satellite remote sensing data acquisition module is used to collect distribution network line image data and distribution network line tower latitude and longitude precise coordinate data using remote sensing satellites; The data processing module is used to pre-process the data collected by the satellite remote sensing data collection module and extract tree information related to the distribution network line; The tree barrier identification and analysis module is used to identify and evaluate potential tree barriers based on the growth characteristics of trees and the safety distance requirements of distribution network lines. Specifically, it includes: constructing a tree distribution identification model based on high-resolution satellite remote sensing images: adopting a classification method based on support vector machine (SVM), selecting radial basis function kernel function (RBF) as the kernel function of SVM, and the SVM classifier with RBF as the kernel function has two adjustable parameters, namely kernel function parameter σ2 and error penalty factor C. The kernel function parameter σ2 affects the complexity of sample data distribution in high-dimensional feature space, and the error penalty factor C adjusts the confidence interval and empirical risk ratio of the learning machine in a determined feature space, and uses σ2 to map data to the high-dimensional feature space. Then, the C value is found in this high-dimensional feature space to optimize the confidence range and empirical risk ratio of the SVM, thereby obtaining a globally optimal SVM classifier. The model is then iteratively trained and optimized for verification: the constructed training sample set is input into the SVM classifier and iteratively trained using the loss function. When the loss value converges, the final model is retained. The feature set of the high-resolution image is then input into the trained model, and the SVM classifier is used to determine the tree distribution range and output the tree recognition results based on the high-resolution image. The early warning and decision support module is used to summarize the tree obstacle information identified by the tree obstacle identification and analysis module, classify the tree obstacles according to their severity, and analyze them in association with distribution network equipment and facilities; The user interaction interface module is used to provide a user interface to display the distribution of tree barriers on the distribution network lines, so that users can query, browse and export tree barrier information.

2. The distribution network line tree obstacle automatic identification system based on satellite remote sensing according to claim 1 is characterized in that: The satellite remote sensing data acquisition module is also used to collect vegetation coverage along the distribution network line. The image acquisition range needs to cover the designated distribution line channel, with the line as the center and a buffer zone formed by a set distance extending to the left and right.

3. The distribution network line tree obstacle automatic identification system based on satellite remote sensing according to claim 1 is characterized in that: The preprocessing includes image denoising, enhancement and classification, and a conventional data processing process is formulated with reference to national standards. Based on the acquired satellite original images, radiometric calibration, geometric correction, atmospheric correction and image mosaicking are performed to obtain satellite standard products that have passed quality inspection.

4. The distribution network line tree obstacle automatic identification system based on satellite remote sensing according to claim 1 is characterized in that: The method of extracting tree information related to distribution network lines specifically includes: using a module for rapid pixel labeling of regions of interest (ROIs) of arbitrary shapes to construct a labeling specification for tree distribution identification; after obtaining the labeling specification, performing data labeling based on the specification and establishing a tree distribution identification sample database; wherein, using a visual interpretation method and combining existing ground survey data, a tree sample set is constructed on collected high-resolution satellite remote sensing images, and image features of the trees are extracted as training samples, wherein the image features of the trees include spectral features and texture features.

5. The distribution network line tree obstacle automatic identification system based on satellite remote sensing according to claim 4 is characterized in that: The module for quickly marking pixels of the ROI of any shape specifically includes: The region of interest (ROI) selection module is used to determine the boundaries based on the region geometry, giving priority to regions with clear outlines and separation from other regions; Arbitrary shape ROI decomposition module, used to quickly detect concave points on the boundary, and decompose the arbitrary shape polygon area into multiple convex hull areas according to the detected concave points, and then calculate each convex hull area separately; The single-category pixel labeling module is used to perform the following process: a digital image containing K objects is represented by the following model: Where z represents the image pixel, P(ω i ) and p(z|ω i ) is the prior probability and conditional probability density of the i-th type target; for RGB images, pixel z = (r,g,b)'; the color components on each channel are independent of each other, that is: To facilitate visualization, the pixel distribution is spread over the three RGB color channels; By extracting pixels within ROI of arbitrary shape and training them through SVDD classifier, classification and recognition of the entire image can be achieved.

6. The distribution network line tree obstacle automatic identification system based on satellite remote sensing according to claim 4 is characterized in that: The spectral features include normalized difference vegetation index, difference vegetation index, enhanced vegetation index, ratio vegetation index, green ratio vegetation index, greenness index, adjusted soil adjusted vegetation index and red-shortwave infrared band difference; the texture features include angular second moment, contrast, correlation, heterogeneity, homogeneity, mean, standard deviation and entropy.

7. The distribution network line tree obstacle automatic identification system based on satellite remote sensing according to claim 1 is characterized in that: The tree barrier identification and analysis module specifically includes a tree barrier identification and evaluation module based on a machine learning algorithm.

8. The distribution network line tree obstacle automatic identification system based on satellite remote sensing according to claim 1 is characterized in that: The user interaction interface module is specifically used to output the identification and evaluation results in the form of graphics and reports, and provide them to power grid operation and maintenance personnel for decision support.

9. A method for automatically identifying tree obstacles on distribution network lines based on satellite remote sensing, characterized in that: Based on the system of any one of claims 1 to 8, the early warning and decision support module performs the following steps: S21, Distribution Channel Tree Distribution Risk Level Classification: Based on the distance between trees and distribution lines, the distribution channel tree distribution risk is classified into three levels: high, medium, and low, and the risk assessment criteria are determined; S22, determining the tree distribution identification results and overlaying them with the distribution network line data in the GIS: the extracted results reflect the number and distribution of trees, and the tree distribution identification results are associated with the distribution network equipment and facility information by defining associated fields; S23, overlay analysis of tree identification results with distribution network line data in GIS: Perform spatial overlay and association analysis on the tree extraction results and power grid GIS data, associate each tree extraction result with a line tower, and calculate the closest distance to the conductor and tower. Use fields to represent the relative position of the tree and the nearest tower, and finally display the association results.

Images