Three-dimensional laser point cloud-based method and system for in-depth vegetation management in transmission corridors

Abstract

Disclosed are a three-dimensional laser point cloud-based method and system for in-depth vegetation management in transmission corridors. The method comprises: collecting multi-source data by means of an unmanned aerial vehicle- and helicopter-mounted multi-sensor system, and preprocessing the multi-source data; using the processed multi-source data to create a digitalized power grid corridor and construct a vegetation management analysis model; on the basis of data analyzed by the vegetation management analysis model and geofencing technology, designing a hazard vegetation clearing-cost-compensation prediction model; and developing a mobile application for analyzing hazard vegetation data flows and the closed-loop management of vegetation management service flows. The described method achieves high-precision and high-coverage data collection, precisely identifies transmission line digital twins and hazard vegetation, standardizes and automates the management of hazard vegetation clearing and related compensation, and develops a mobile application for analyzing hazard vegetation data flows and the closed-loop management of vegetation management service flows, thereby achieving the end-to-end digital and mobile management of vegetation management. The organic combination of steps comprehensively improves the intelligence and efficiency of transmission line vegetation management.

Description

A power transmission channel tree barrier deepening method and system based on three-dimensional laser point cloud TECHNICAL FIELD

[0001] The application relates to the technical field of power system operation and maintenance, and particularly relates to a power transmission channel tree barrier deepening method and system based on three-dimensional laser point cloud. BACKGROUND

[0002] The development and application of GIS started relatively late in China, which started in the late 1970s. Although the history is short, the development is still very fast. The development can be roughly divided into four stages, namely the preparation stage, the starting stage, the development stage and the industrialization stage. Since 1996, the industry has been formed and is moving towards the industrialization stage. After the development in the early stage, an independent industry has gradually formed in research and application, and is developing towards industrialization and marketization. During the entire development process, it can be seen that during the period from the "Sixth Five-Year Plan" to the "Eighth Five-Year Plan", the GIS technology in China has developed rapidly. In particular, during the "Ninth Five-Year Plan" period, the original State Science and Technology Commission included GIS as an independent subject in the "top priority" scientific and technological research plan, and gave full attention and support, the technology development speed was significantly accelerated, the basic software technology support was comprehensively strengthened, and a batch of high-level technical achievements and products appeared.

[0003] For example, by using GIS spatial analysis technology, the spatial clustering and historical morphology analysis of the space-time process of Beijing urban land use expansion during 1982-1997 is carried out. By using GIS technology, the connectivity relationship between road sections is analyzed from the characteristics of urban road network, and a shortest path algorithm between two nodes of urban road network is obtained. By using GIS tools and mathematical model method, the present situation of land use and land cover in China is analyzed by using land use degree index and vegetation index, and the conclusion is obtained: the existing land cover situation in China represents a development and utilization level of the total land use degree index of 202 in China. By using GIS technology, the soil erosion amount of Guanxi Township in Taihe County, Jiangxi Province is estimated by using general soil erosion equation. Based on meteorological satellite data as the main information source, land satellite TM data, land use thematic map and meteorological observation data as auxiliary information sources, the spatial analysis technology is used to dynamically monitor and comprehensively analyze the distribution characteristics and change rules of urban heat. In recent years, GIS has developed rapidly and presents new development trends, including the combination of GIS and expert system, neural network, the integration of GIS and CAD software, virtual geographic information system, the integration of remote sensing (RS) and global positioning system (GPS), and open GIS.

[0004] At present, two-dimensional GIS system has been widely used in the production, operation and management of power industry, but two-dimensional GIS system can not give people a sense of being there, can not express the details of the primary and secondary equipment and topology connection inside the substation, also can not express the geographical environment along the transmission and distribution lines and the spatial position relationship between the transmission and distribution line equipment. Therefore, the lack of information performance restricts the further promotion of two-dimensional GIS in the power industry. Compared with two-dimensional GIS system, three-dimensional GIS not only can solve the problem of spatial relationship display, but also can load three-dimensional model of power facilities and equipment into the system, establish the association between unit model object and attribute data, so as to better enable users to understand the overall situation of power grid and the specific situation of each facility and equipment. After years of exploration and application practice, domestic power three-dimensional GIS has gradually entered the data analysis and control stage from the past data management.

[0005] In China, the research on three-dimensional visualization management technology of power line is based on the development of geographic information system. In two-dimensional environment, various ground features and terrain are displayed in layered form, and various actual objects are represented by symbols such as points, lines and surfaces; only planar data can be processed, and the elevation data is processed by projecting it onto the plane first, so it cannot express different elevation points at the same position. Obviously, such two-dimensional system has problems such as poor interactivity and poor visualization, which cannot meet the requirements of deep application.

[0006] Three-dimensional digitization is to obtain the shape data of the object by artificial means, process and splice the obtained data information, arrange through modeling, seamlessly integrate various isolated single-view three-dimensional digital models, and form three-dimensional data file after mapping and rendering. Modeling is a very important step, especially in the face of such large-scale three-dimensional data that need models. Three-dimensional digital model is similar to two-dimensional digital model, but compared with two-dimensional digital model, three-dimensional data model has more advantages and can more comprehensively reflect the objective reality. It can more intuitively express the real appearance and shape of terrain or object by using virtual reality technology, making the abstract point-line-surface symbols of two-dimensional GIS intuitive, and combining with common sense to identify and speed up the analysis and identification speed. Two / three-dimensional integrated digital sharing technology, for users, two-dimensional and three-dimensional interface is no longer different windows of different systems, but only different display options provided by a system to users. Users can choose to apply in two-dimensional environment or three-dimensional environment, and two-dimensional and three-dimensional systems can display the same or different data based on spatial coordinates.

[0007] The video monitoring device has been installed on the tower for tree monitoring in China for a long time. In recent years, the function of the online video monitoring device is becoming more and more powerful. The infrared device can automatically alarm according to the distance of the tree from the conductor, reminding the operation and maintenance personnel to clean in time to avoid causing the line trip. This method has good effect, but the cost of large-scale installation of video monitoring is too high due to the large number of towers. The operation and maintenance of video monitoring is also difficult. In addition, the standards of video monitoring manufacturers are not unified at this stage, and there is no unified access standard, the network security performance is low, causing equipment information security risks. At present, the application of unmanned aerial vehicles greatly helps to improve the operation and maintenance level of the transmission line. The wide range and high accuracy are the advantages of unmanned aerial vehicles. The accuracy of using unmanned aerial vehicles to measure the distance of tree barrier defects in the line protection area is relatively high. The common unmanned aerial vehicle ranging technologies include ultrasonic ranging technology, laser radar scanning ranging, and unmanned aerial vehicle oblique photography ranging technology. The highest accuracy of the unmanned aerial vehicle ranging technology is the laser radar measurement carried by the unmanned aerial vehicle. Since the radar can collect high-precision point cloud data, and is equipped with a high-resolution digital image acquisition device, it can restore the actual situation to the greatest extent. Through data analysis, a three-dimensional model is established, which can effectively analyze the distance. The use of laser radar has broad application prospects in the management of transmission lines.

[0008] The term "geographic information system" (GIS) was first proposed by Canadian surveyor Roger F. Tomlinson in 1963. Due to the great role and broad prospects of GIS, related research and application have made great progress in recent years. Western countries represented by the United States have invested a lot of manpower and material resources in research and development, and have launched software tools such as ARC / INFO, MGI, and MAPINFO. Moreover, with the development of computer-related fields, GIS has also developed, especially the development of virtual reality geographic information systems. The first successful VRGIS appeared in the United States in the early 1990s. Faust and Koller successfully integrated geographic information systems and virtual reality systems, namely the campus environment information system of the American College of Geodesy, and proposed the concept of "virtual reality geographic information system". VRGIS can be seen as a special "traditional" GIS. It moves the three-dimensional visualization module, which was originally only a general position in the two-dimensional geographic information system, to the core position of the entire system, and makes the user and the three-dimensional visual, auditory, and other sensory real-time interaction with the system based on the existence of the system.

[0009] The research on 3D visualization management of power lines abroad is a deep application based on GIS system. The research on 3D visualization abroad started early and is at a high level. There are many mature products, such as the global 3D visualization platform released by the famous company and the visualization platform released by the company. They can realize the visualization of global images and data based on the Internet. The company in the United States adopts multi-resolution technology and virtual reality modeling language to realize the global virtual terrain environment prototype system connected to the Internet. The company has released a 2D visualization module ArcGlobal based on multi-resolution global data in the ArcGIS series products. They all provide secondary development interfaces to meet the specific application requirements of different industries. At present, the application of 3D geographic information is mainly concentrated in the fields of military geographic landforms, agriculture and forestry, etc. Only a few countries have established related 3D power grid geographic information systems, so there is a great development space in this aspect.

[0010] Foreign countries have also made corresponding progress in this regard. The German GAH (Gust Alberts GmbH) company has successfully applied laser radar to the detection of power facilities and developed a system that can be used for power line inspection. The Australian Spatial Information Cooperative Research Center has developed a miniature line patrol unmanned aerial vehicle equipped with a laser range finder, which can directly measure the distance between the power line and the obstacles below it.

[0011] Although there are more studies on power line extraction and analysis, there are currently fewer related studies on tree barrier hazard analysis. The traditional method is still the most common tree barrier detection method, and only some scientific research institutions and power departments use methods based on unmanned aerial vehicle images or point clouds to study and apply power line tree barrier detection. USI (USILAND) power company early designed and developed a system suitable for detecting power lines, and named it Power Dount system. The system mainly detects the sag of the power line through a high-precision temperature sensor. Douglass introduced the DTCR platform designed by the American Electric Power Research Institute, which is also used for power line sag detection. The platform extracts the sag of the power line by using image processing methods on the target position of the power line photographed by the high-resolution camera installed in the system. In 2003, the power transmission network in Canada and the United States failed to supply power due to vegetation, causing the power supply system to be interrupted. Australian researchers tracked and detected the vegetation around the power grid through images obtained by unmanned aerial vehicles, analyzed the spatial position relationship between the power lines and the vegetation based on various image processing algorithms, and judged whether the height of the vegetation would interfere with and damage the use of the power lines. In addition, Chiba University in Japan and Kansai Electric Power Company jointly developed a new unmanned aerial vehicle line inspection system that can detect main defects such as tower material corrosion, tower body tilt, and concrete pole body cracks.

[0012] In recent years, the South Grid and the Cloud Network Company have paid great attention to the close combination of production practice and scientific and technological research and development, and the level of technological innovation has rapidly improved. The research scope covers the focus and hot research fields of unmanned aerial vehicle routine operation, unmanned aerial vehicle special operation, unmanned aerial vehicle inspection, transmission line instantaneous working condition analysis, multi-working condition analysis, line load flow checking, unmanned aerial vehicle automatic driving inspection, transmission line three-dimensional channel construction, digital asset management, etc. In recent years, a number of research results with significant influence have been achieved. However, there is still no mature solution for the lean management demand of tree barriers in key areas such as the Lijiang Power Supply Bureau and the Two Mountains and One Lake Important Scenic Area.

[0013] SUMMARY

[0014] In view of the above problems, the present application is proposed.

[0015] Therefore, the technical problem solved by the present application is that the present application solves the following key technical problems:

[0016] The unmanned aerial vehicle, the helicopter and the multi-sensor system are used to collect high-density and high-precision three-dimensional laser point cloud data, visible light images, infrared images and video data, realize comprehensive coverage and accurate measurement of the power transmission line and the surrounding environment, fuse three-dimensional laser point cloud data, high-resolution images, infrared images, video data, GIS data, environmental data, power grid data and satellite remote sensing data, realize seamless integration and efficient processing of multi-source data, realize real-time early warning and classified processing of tree barrier risks based on tree barrier analysis data and electronic fence technology, verify and archive feedback data, solve the problems of information asymmetry and untimely feedback in the traditional management mode, and utilize the electronic fence technology to accurately mark and manage the trees needing compensation.

[0017] To solve the above technical problems, the application provides the following technical scheme: a power transmission channel tree barrier deepening method based on three-dimensional laser point cloud, comprising:

[0018] The multi-sensor system is carried on the unmanned aerial vehicle and the helicopter to collect multi-source data, and the multi-source data is preprocessed;

[0019] The processed multi-source data is used to establish a digital power grid channel and construct a tree barrier analysis model;

[0020] A tree cutting compensation prediction model is designed based on the data analyzed based on the tree barrier analysis model and the electronic fence technology;

[0021] A mobile application is developed to analyze tree barrier data flow and closed-loop manage tree barrier management business flow.

[0022] As a preferred scheme of the power transmission channel tree barrier deepening method based on three-dimensional laser point cloud, the platform parameters include task information and resource information.

[0023] As a preferred scheme of the power transmission channel tree barrier deepening method based on three-dimensional laser point cloud, the multi-sensor system includes a laser radar system, a high-resolution digital camera, an infrared camera, a high-frequency video camera, an environmental sensor, a device sensor, a GIS platform and a high-resolution remote sensing satellite.

[0024] The multi-source data includes three-dimensional laser point cloud data, high-resolution visible light images, infrared images, video data and GIS data.

[0025] As a preferred embodiment of the power transmission channel tree obstacle deepening method based on three-dimensional laser point clouds described in this invention, the method for establishing a digital power grid channel using processed multi-source data includes: processing three-dimensional laser point cloud data to obtain a three-dimensional point cloud dataset; using a PointNet++ model to construct a classification function to classify the three-dimensional point cloud dataset; and identifying and separating different types of objects such as towers, conductors, ground, and vegetation. The formula is expressed as follows:

[0026] C(x,y,z)=softmax(W·φ(x,y,z)+b)

[0027] Where C(x,y,z) represents the probability distribution of different categories, softmax is the activation function that ensures that the sum of the probabilities of all outputs is 1, W represents the weight matrix, φ(x,y,z) represents the feature vector of the point (x,y,z), b represents the bias term, and (x,y,z) represents the position in space.

[0028] C(x,y,z)=[p 杆塔 ,p 导线 ,p 地面 ,p 植被 ]; If p 杆塔 If p > 0.65, it is classified as a pole or tower. 导线 If p > 0.6, it is classified as a wire. 地面 If p > 0.7, it is classified as ground. 植被 If the value is >0.55, it is classified as vegetation;

[0029] High-resolution visible light images are acquired using a high-resolution digital camera, obtaining clear visual data and converting it into a light reflectance distribution map.

[0030] Infrared images are acquired using an infrared camera to identify areas of thermal anomalies.

[0031] By acquiring video data through high-frequency video cameras, dynamic monitoring information is obtained, and dynamic changing features are extracted.

[0032] GIS data is acquired through a GIS platform, smoothed, and analyzed to extract spatial distribution characteristics.

[0033] The formula for establishing a digital power grid channel is expressed as follows:

[0034] Where D(x,y,z) represents the model of the digital power grid channel, (x,y,z) represents the coordinates in three-dimensional space, (x... i ,y i ,z irepresents the coordinates of the i-th point in the point cloud data, h(x-ξ, y-η) represents the kernel function, f(ξ, η) represents the distribution function of the input data, ξ, η represent the integral variable, R(λ, θ) represents the light reflectivity, λ is the wavelength, θ represents the angle, α(λ) represents the absorption coefficient of light, d represents the distance, σ represents the Stefan-Boltzmann constant, T s represents the surface temperature of the object, T b represents the background temperature, A represents the integral area, N represents the total number of frames of video data, (x i , y i ) represents the coordinates of the i-th frame in the video data, (μ x , μ y ) represents the mean of the coordinates in the video data, σ represents the standard deviation of the coordinates in the video data, exp represents the exponential function, I(x, y) represents the high-resolution visible light image, T(x, y) represents the temperature distribution of the infrared image, V(t) represents the dynamic characteristics in the video data, G(x, y) represents the processed GIS data;

[0035] When 50≤D(x, y, z)≤100, it indicates that the power transmission channel at this position has high spatial density and complexity, high risk level, and needs to be processed and monitored preferentially;

[0036] When 10≤D(x, y, z)<50, it indicates that the power transmission channel at this position has medium complexity, medium risk level, and needs to be regularly monitored and maintained to ensure the safety of the power grid channel;

[0037] When 0<D(x, y, z)<10, it indicates that the power transmission channel at this position is relatively simple or sparse, and has a low risk level, and the power grid channel is in a safe state.

[0038] As a preferred scheme of the power transmission channel tree barrier deepening method based on three-dimensional laser point cloud according to the application, the tree barrier analysis model is constructed by: constructing the tree barrier analysis model, based on the deep learning-based point cloud automatic classification technology, combining the three-dimensional laser point cloud features of the target tree barrier, combining the single tree segmentation technology, and automatically counting the dangerous tree barrier area to construct the tree barrier analysis model;

[0039] The single tree is segmented from the point cloud data:

[0040] By comparing the height T(x, y, z) of each point in the integral region Ω with the dangerous height threshold T threshold , if the height exceeds the threshold, it belongs to the dangerous tree barrier:

[0041] The tree barrier analysis model is constructed by the PointNet++ neural network, and the formula is represented as:

[0042] M = ∫ Ω (S(x,y,z))·C(x,y,z)·I(T(x,y,z)>T threshold )dΩ

[0043] Wherein, M represents the constructed tree barrier analysis model, the value threshold of M is between 0 and 1;Ω represents the integral region, n represents the number of point cloud data points, (x,y,z) represents the coordinates in three-dimensional space, (x i ,y i ,z i ) represents the coordinates of the i-th point, ∈ represents the threshold of single tree segmentation, represents the indicator function, W represents the weight matrix, φ(x,y,z) represents the point cloud feature vector, b represents the bias term, softmax represents the activation function, T(x,y,z) represents the tree barrier height, T threshold represents the dangerous height threshold, S(x,y,z) represents the segmentation result, A represents the dangerous tree barrier area.

[0044] As a preferred scheme of the power transmission channel tree barrier deepening method based on three-dimensional laser point cloud, wherein: the tree felling compensation prediction model is designed based on the data analyzed by the tree barrier analysis model and the electric fence technology, including: by setting a distance threshold, the trees in the electric fence are identified and classified:

[0045] Wherein, represents the logical function;

[0046] By comparing the tree barrier height in the integral region with the dangerous height threshold, combined with the result of the electric fence technology, the compensation priority of each tree is calculated:

[0047] Wherein, P represents the tree felling compensation prediction model, Ω represents the integral region, T(x,y,z) represents the tree barrier height, T threshold represents the dangerous height threshold, is an indicator function, E represents the number of trees in the electric fence, k(x,y,z) represents a Gaussian function, α(λ) represents the light absorption coefficient, d represents the distance, exp represents the exponential function;

[0048] When 0.75<P(x,y,z)≤1, it means that the tree felling and compensation in this area is high risk and needs to be handled in priority;

[0049] When 0<P(x,y,z)<0.75, it means that the tree felling and compensation in this area is low risk and does not need to be handled.

[0050] As a preferred scheme of the power transmission channel tree barrier deepening method based on three-dimensional laser point cloud, the development of mobile application analysis tree barrier data flow and closed-loop management tree barrier management business flow includes: when the value of the tree barrier analysis model M is greater than 0.5, the tree barrier risk level is high, the tree barrier density in the region is high, and there are more dangerous tree barriers, which need to be processed in priority, and the system generates a priority processing task and sends it to the mobile application terminal.

[0051] When the value of the tree barrier analysis model M is less than or equal to 0.5, the tree barrier risk level is low, the tree barrier density in the region is low, and there is no dangerous tree barrier, and the queue processing is performed according to the set priority order.

[0052] The evaluation accuracy of each risk level needs to reach more than 95%, and when it is lower than this standard, the system will optimize and adjust the algorithm.

[0053] As a preferred scheme of the power transmission channel tree barrier deepening method based on three-dimensional laser point cloud, the development of mobile application analysis tree barrier data flow and closed-loop management tree barrier management business flow further includes that the system real-time statistics tree barrier area, and generates a task work order of tree barrier information exceeding the threshold to the mobile terminal of the relevant construction personnel, and when the error exceeds 5%, the system re-counts and calibrates;

[0054] Using GPS and GIS technology, the tree barrier position is real-time positioned, and accurate navigation path is provided for the construction personnel, and when the positioning error exceeds 5 meters, the system automatically adjusts and recalculates the navigation path, so that the construction personnel can accurately arrive at the target position;

[0055] When the construction personnel complete the tree barrier felling task, the photos and on-site situation before and after felling are uploaded, and the system verifies the feedback data; if the feedback data is incomplete or incorrect, the system prompts to resubmit.

[0056] When the tree felling involves compensation, the system automatically generates an electronic fence and marks the trees to be compensated, and the compensation amount and related information are notified to the relevant personnel through the mobile terminal, and the compensation record in the management system is real-time updated;

[0057] The overall performance of the system is evaluated every month, so that the task completion rate reaches more than 95% and the processing time is shortened by more than 10%.

[0058] The power transmission channel tree barrier management deepening system based on three-dimensional laser point cloud, wherein:

[0059] The data acquisition module collects multi-source data through the multi-sensor system carried by the unmanned aerial vehicle and the helicopter, and pre-processes the multi-source data;

[0060] The tree barrier analysis module uses the processed multi-source data to establish a digital power grid channel and construct a tree barrier analysis model;

[0061] The claim settlement prediction module designs a tree cutting claim settlement prediction model based on the analyzed data and the electric fence technology.

[0062] The mobile application module develops a mobile application to analyze tree barrier data flow and closed-loop manage tree barrier management business flow.

[0063] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the method according to any one of the embodiments.

[0064] A computer readable storage medium stores a computer program, wherein the computer program is executed by a processor to implement the steps of the method according to any one of the embodiments.

[0065] The method provided by the application realizes high-precision and high-coverage data collection, accurately identifies the digital twin of the power transmission line and the tree barrier, realizes the standardization and automation of tree cutting and compensation management, develops a mobile application to analyze tree barrier data flow and closed-loop manage tree barrier management business flow, and realizes the full-process digitization and mobile management of tree barrier management. BRIEF DESCRIPTION OF DRAWINGS

[0066] In order to more clearly illustrate the technical solutions of the embodiments of the application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor.

[0067] Fig. 1 is a whole flowchart of a kind of power transmission channel tree barrier deepening method based on three-dimensional laser point cloud provided by the first embodiment of the application;

[0068] Fig. 2 is a mobile application technology architecture diagram in a kind of power transmission channel tree barrier deepening method based on three-dimensional laser point cloud provided by the first embodiment of the application;

[0069] Fig. 3 is the whole technical architecture diagram of a kind of visual display platform of power transmission channel tree barrier deepening method based on three-dimensional laser point cloud provided by the first embodiment of the application. DETAILED DESCRIPTION

[0070] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should belong to the protection scope of the present application.

[0071] Embodiment 1

[0072] Referring to FIG. 1, for an embodiment of the present application, a power transmission corridor tree barrier deepening method based on three-dimensional laser point cloud is provided, comprising:

[0073] S1: Collecting multi-source data by unmanned aerial vehicle and helicopter carrying multi-sensor system, and pre-processing the multi-source data.

[0074] The multi-sensor system includes a laser radar system, a high-resolution digital camera, an infrared camera, a high-frequency video camera, an environmental sensor, a device sensor, a GIS platform, and a high-resolution remote sensing satellite.

[0075] The multi-source data includes three-dimensional laser point cloud data, high-resolution visible light images, infrared images, video data, and GIS data.

[0076] Deep learning and machine learning algorithms are used to classify and denoise the collected multi-source data. The specific steps include: using algorithms to classify laser point cloud data, identify and separate different types of objects such as towers, wires, ground, vegetation, etc. Remove noise points to ensure high precision and high quality of data. Extract and analyze the geometric information of the processed data to provide a reliable data foundation for subsequent tree barrier analysis and three-dimensional visualization display. Data collection by unmanned aerial vehicle and helicopter carrying multi-sensor system can cover a wide area and obtain detailed data at different heights and angles. This method is more efficient than traditional ground inspection, which can significantly reduce the time and cost of manual inspection. At the same time, the multi-sensor system can collect various types of data, including three-dimensional laser point cloud, high-resolution images, infrared images, video data, and GIS data, ensuring the comprehensiveness and diversity of data.

[0077] Through deep learning and machine learning algorithms, the collected multi-source data can be classified and denoised, which can effectively identify and separate different types of objects (such as towers, wires, ground, vegetation, etc.), and remove noise points in the data. This not only ensures the high precision and high quality of the data, but also improves the accuracy of subsequent analysis. The high precision and reliability of data processing provide a solid foundation for tree barrier analysis and three-dimensional visualization display.

[0078] Furthermore, the preprocessed data, after geometric information extraction and analysis, provides a reliable data foundation for tree obstacle analysis and 3D visualization. This data foundation can help to more accurately identify and assess tree obstacle risks, supporting scientific decision-making and efficient management. By constructing a high-precision 3D model, transmission lines and their surrounding environment can be visually displayed, facilitating inspection and maintenance by operation and maintenance personnel, and improving work efficiency and management level.

[0079] Furthermore, this efficient and accurate data acquisition and processing method allows for the timely detection and assessment of potential risks to transmission lines, such as tree obstructions and equipment failures, ensuring the safety and reliability of these lines. High-quality data support enables maintenance personnel to take timely measures to prevent faults, reduce the risk of power outages, and improve the stable operation of the power grid.

[0080] S2: Utilize the processed multi-source data to establish a digital power grid channel and construct a tree obstacle analysis model.

[0081] Establishing a digital power grid channel using processed multi-source data includes processing 3D laser point cloud data to obtain a 3D point cloud dataset, constructing a classification function using a PointNet++ model to classify the 3D point cloud dataset, and identifying and separating different types of objects such as towers, conductors, ground, and vegetation. The formula is expressed as:

[0082] C(x,y,z)=softmax(W·φ(x,y,z)+b)

[0083] Where C(x,y,z) represents the probability distribution of different categories, softmax is the activation function that ensures that the sum of the probabilities of all outputs is 1, W represents the weight matrix, φ(x,y,z) represents the feature vector of the point (x,y,z), b represents the bias term, and (x,y,z) represents the position in space.

[0084] C(x,y,z)=[p 杆塔 ,p 导线 ,p 地面 ,p 植被 ]; If p 杆塔 If p > 0.65, it is classified as a pole or tower. 导线 If p > 0.6, it is classified as a wire. 地面 If p > 0.7, it is classified as ground. 植被 If the value is >0.55, it is classified as vegetation;

[0085] High-resolution visible light images are acquired using a high-resolution digital camera, obtaining clear visual data and converting it into a light reflectance distribution map.

[0086] Infrared images are acquired using an infrared camera to identify areas of thermal anomalies.

[0087] Video data is acquired by high-frequency video cameras to obtain dynamic monitoring information and extract dynamic features:

[0088] GIS data is acquired through the GIS platform, smoothed and analyzed to extract spatial distribution features:

[0089] The digital power grid channel formula is established as:

[0090] where D(x, y, z) represents the model of the digital power grid channel, (x, y, z) represents the coordinates in three-dimensional space, (x i ,y i ,z i ) represents the coordinates of the i-th point in the point cloud data, h(x-ξ, y-η) represents the kernel function, f(ξ,η) represents the distribution function of the input data, ξ,η represents the integral variable, R(λ,θ) represents the light reflectivity, λ is the wavelength, θ represents the angle, α(λ) represents the light absorption coefficient, d represents the distance, σ represents the Stefan-Boltzmann constant, T s represents the surface temperature of the object, T b represents the background temperature, A represents the integral area, N represents the total number of frames of video data, (x i ,y i ) represents the coordinates of the i-th frame of video data, (μ x ,μ y ) represents the mean of the coordinates of the video data, σ represents the standard deviation of the coordinates of the video data, exp represents the exponential function, I(x,y) represents the high-resolution visible light image, T(x,y) represents the temperature distribution of the infrared image, V(t) represents the dynamic features in the video data, and G(x,y) represents the processed GIS data.

[0091] When 50≤D(x,y,z)≤100, it indicates that the power transmission channel at this location has a high spatial density and complexity, with a high risk level, requiring priority processing and enhanced monitoring.

[0092] When 10≤D(x,y,z)<50, it indicates that the power transmission channel at this location has moderate complexity, with a medium risk level, requiring regular monitoring and maintenance to ensure the safety of the power grid channel.

[0093] When 0<D(x,y,z)<10, it indicates that the power transmission channel at this location is relatively simple or sparse, with a low risk level, and the power grid channel is in a safe state.

[0094] The tree barrier analysis model is constructed by using a deep learning-based point cloud automatic classification technology, combining the three-dimensional laser point cloud features of the target tree barrier, and combining a single tree segmentation technology to automatically count the dangerous tree barrier area and construct the tree barrier analysis model.

[0095] The single tree is segmented from the point cloud data:

[0096] The height T(x, y, z) of each point in the integral region Ω is compared with the dangerous height threshold T threshold , and if the height exceeds the threshold, it belongs to the dangerous tree barrier:

[0097] The tree barrier analysis model is constructed by using a PointNet++ neural network, and the formula is:

[0098] M = ∫ Ω (S(x, y, z))·C(x, y, z)·I(T(x, y, z) > T threshold )dΩ

[0099] Where M represents the constructed tree barrier analysis model, the value of M is between 0 and 1; Ω represents the integral region, n represents the number of point cloud data points, (x, y, z) represents the coordinates in three-dimensional space, (x i , y i , z i ) represents the coordinates of the i-th point, ∈ represents the single tree segmentation threshold, represents the indicator function, W represents the weight matrix, φ(x, y, z) represents the point cloud feature vector, b represents the bias term, softmax represents the activation function, T(x, y, z) represents the tree barrier height, T threshold represents the dangerous height threshold, S(x, y, z) represents the segmentation result, and A represents the dangerous tree barrier area.

[0100] By establishing a digital power grid channel through multi-source data, comprehensive and accurate three-dimensional modeling of the transmission line and its surrounding environment can be achieved. This method not only provides high-precision spatial information, but also combines multiple data sources (such as visible light images, infrared images, video data, and GIS data) to achieve comprehensive monitoring and analysis of the transmission line. Through three-dimensional laser point cloud data and other high-resolution data, high-precision three-dimensional modeling of the transmission line can be achieved, providing detailed spatial information. Combined with infrared images and video data, temperature monitoring and dynamic monitoring of transmission line equipment can be achieved, and potential faults and risks can be detected in a timely manner. Through the fusion of GIS data, comprehensive information of the transmission line and the surrounding geographical environment can be provided, supporting more scientific decision-making and management. The fusion processing of multi-source data ensures the diversity and accuracy of the data, providing a reliable data foundation for subsequent analysis and decision-making.

[0101] Based on the processed multi-source data, a tree barrier analysis model is constructed using deep learning and single tree segmentation technology to realize accurate identification and evaluation of tree barriers. Through deep learning algorithm, different objects (such as poles, towers, wires, vegetation, etc.) in point cloud data can be automatically identified and classified, improving the efficiency and accuracy of analysis. Through single tree segmentation technology, the tree barrier area of each tree can be accurately segmented and counted, providing detailed tree barrier information. Based on the tree barrier height and danger threshold, real-time evaluation and warning of tree barrier risk can be realized, and timely measures can be taken to eliminate potential risks. The tree barrier analysis model provides detailed tree barrier information and risk evaluation results, supporting scientific decision-making and efficient management, improving the safety and reliability of the power transmission line.

[0102] S3: Design a tree felling compensation prediction model based on the data analyzed by the tree barrier analysis model and the electronic fence technology.

[0103] Designing a tree felling compensation prediction model based on the data analyzed by the tree barrier analysis model and the electronic fence technology includes identifying and classifying trees within the electronic fence by setting a distance threshold:

[0104] Wherein, represents a logical function;

[0105] By comparing the tree barrier height in the integral region with the dangerous height threshold, combined with the results of the electronic fence technology, the compensation priority of each tree is calculated:

[0106] Wherein, P represents the tree felling compensation prediction model, Ω represents the integral region, T(x,y,z) represents the tree barrier height, T threshold represents the dangerous height threshold, is an indicator function, E represents the number of trees in the electronic fence, k(x,y,z) represents the Gaussian function, α(λ) represents the light absorption coefficient, d represents the distance, and exp represents the exponential function;

[0107] When 0.75 < P(x,y,z) ≤ 1, it means that the tree felling and compensation in this area is high risk and needs to be handled first.

[0108] When 0 < P(x,y,z) < 0.75, it means that the tree felling and compensation in this area is low risk and does not need to be handled.

[0109] By combining the tree barrier analysis model and electronic fence technology, a tree felling and compensation prediction model is designed to achieve comprehensive assessment of tree felling and compensation risks. Through the tree barrier analysis model, the height of the tree barrier and the dangerous threshold are accurately calculated, and combined with the electronic fence technology, the felling risk of each tree is accurately assessed. It can identify high-risk areas and prioritize trees that need to be felled and compensated, avoiding potential risks. The model uses multi-source data and automated algorithms to achieve automated processing of tree felling and compensation management. Reducing manual intervention, improving management efficiency, and ensuring the accuracy of data processing and decision-making. Combined with electronic fence technology, dynamic monitoring of trees and surrounding environment is realized, providing real-time risk feedback. Ensure that the prediction model can be updated in time when the environment or tree conditions change, providing the latest risk assessment and management recommendations. Through precise risk assessment and real-time monitoring, high-risk trees can be identified and handled in time, reducing power line failures caused by tree barriers. Improve the operational safety and reliability of power transmission lines to ensure stable power supply.

[0110] S4: Develop a mobile application to analyze tree barrier data flow and closed-loop manage tree barrier management business flow.

[0111] Developing a mobile application to analyze tree barrier data flow and closed-loop manage tree barrier management business flow includes, when the value of tree barrier analysis model M is greater than 0.5, the tree barrier risk level is high, the tree barrier density in the area is high, and there are more dangerous tree barriers, which need to be handled first. The system automatically generates a priority handling task and sends it to the mobile application terminal;

[0112] When 0 < the value of tree barrier analysis model M ≤ 0.5, the tree barrier risk level is low, the tree barrier density in the area is low and there are no dangerous tree barriers, and according to the set priority order, the system will be queued for processing;

[0113] The assessment accuracy of each risk level needs to reach more than 95%, when it is lower than this standard, the system will optimize and adjust the algorithm.

[0114] The system real-time statistics tree barrier area, and the tree barrier information exceeding the threshold value generates task work order sent to the mobile terminal of the relevant construction personnel, when the error exceeds 5%, the system re-counts and calibrates;

[0115] Using GPS and GIS technology, real-time positioning of tree barrier position, providing accurate navigation path for construction personnel, when positioning error exceeds 5 meters, the system automatically adjusts and recalculates the navigation path, ensuring that construction personnel can accurately arrive at the target position;

[0116] When the construction personnel complete the tree felling task, upload the before and after felling photos and on-site situation explanation, the system verifies the feedback data; if the feedback data is incomplete or incorrect, the system prompts to resubmit.

[0117] When tree felling involves compensation, the system automatically generates an electronic fence and marks the trees to be compensated, and the compensation amount and related information are notified to relevant personnel through a mobile terminal, and the compensation records in the management system are updated in real time;

[0118] The overall performance of the system is evaluated monthly, with a task completion rate of over 95% and a processing time reduction of over 10%.

[0119] Through clear risk level division, high-risk tree barriers are prioritized for processing, reducing potential risks and ensuring the safety of power transmission lines. Meanwhile, the system automatically generates tasks and sends them to mobile application terminals, improving the efficiency and accuracy of task allocation. Through real-time statistics and calibration, high-precision data is ensured, avoiding task allocation and execution problems caused by data errors and improving the reliability of overall management. Through precise positioning and navigation, the work efficiency and accuracy of construction personnel are improved, ensuring that tree barrier processing tasks are completed in a timely and accurate manner, reducing possible delays and errors. Through real-time feedback and task verification, the effectiveness and transparency of task execution are ensured, improving the controllability and reliability of management. Feedback data from construction personnel provides important reference for system optimization and performance evaluation. Through automatic generation of electronic fences and real-time updating of compensation records, the accuracy and transparency of compensation management are improved, reducing disputes and management difficulties. Regular performance evaluation and system optimization ensure efficient operation and continuous improvement of the system, improving overall management level.

[0120] On the other hand, the embodiment also provides a power transmission channel tree barrier management deepening system based on three-dimensional laser point cloud, which includes:

[0121] The data acquisition module collects multi-source data through a multi-sensor system carried by a UAV and a helicopter, and pre-processes the multi-source data;

[0122] The tree barrier analysis module uses the processed multi-source data to establish a digital power grid channel and construct a tree barrier analysis model;

[0123] The compensation prediction module designs a tree felling compensation prediction model based on the data analyzed by the tree barrier analysis model and electronic fence technology;

[0124] The mobile application module develops a mobile application to analyze tree barrier data flow and closed-loop manage tree barrier management business flow.

[0125] If the above functions are realized in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or parts of the present application that make contributions to the prior art or parts of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0126] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a list of executable instructions for implementing logic functions, which can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus or device, such as a computer-based system, a system including a processor or other system that can fetch instructions from an instruction execution system, apparatus or device and execute the instructions, or in conjunction with these instruction execution systems, apparatus or devices. For the purpose of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transport programs for use by or in connection with an instruction execution system, apparatus or device, or in conjunction with these instruction execution systems, apparatus or devices.

[0127] More specific examples (a non-exhaustive list) of the computer-readable medium include the following: an electrical connection having one or more wires (electrical devices), a portable computer diskette (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium can even be paper or other suitable medium on which the program can be printed, because the program can be obtained electronically, for example, by optical scanning of the paper or other medium, followed by editing, interpreting or otherwise processing, if necessary, in other suitable ways to obtain the program electronically, and then storing it in a computer memory.

[0128] It should be understood that various parts of the present application can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, a plurality of steps or methods can be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, and as in another embodiment, it can be implemented using any or a combination of the following technologies, which are well known in the art: discrete logic circuitry having logic gates for implementing logic functions on data signals, application specific integrated circuits having appropriate combinational logic gates, programmable gate arrays (PGA), field programmable gate arrays (FPGA), etc.

[0129] Example 2

[0130] The experimental site is selected in a certain power transmission line area in Lijiang, Yunnan Province, which has complex terrain and dense trees. The multi-sensor system used in the experiment includes a laser radar system, a high-resolution digital camera, an infrared camera, a high-frequency video camera, an environmental sensor, a device sensor, a GIS platform, and a high-resolution remote sensing satellite. Data collection uses unmanned aerial vehicles and helicopters to cover and shoot at different heights and angles.

[0131] The unmanned aerial vehicle is set to fly at a height of 50 meters, and the helicopter is set to fly at a height of 150 meters, covering the entire experimental area. The flight route follows the preset trajectory to ensure full coverage of the area. The laser radar system is used to collect three-dimensional laser point cloud data, the high-resolution digital camera is used to shoot visible light images, the infrared camera is used to obtain infrared images, the high-frequency video camera is used to record video data, the environmental sensor and the device sensor are used to record environmental parameters and device status respectively, and the GIS platform and the high-resolution remote sensing satellite provide geographic information and remote sensing data.

[0132] The collected data is classified and denoised by deep learning and machine learning algorithms. The specific steps include: classifying the laser point cloud data, identifying and separating different types of objects such as towers, conductors, ground, vegetation, etc., removing noise points to ensure high precision and high quality of the data. Then the processed data is extracted and analyzed for geometric information, providing a reliable data basis for subsequent tree barrier analysis and three-dimensional visualization display.

[0133] Using the processed multi-source data, combined with three-dimensional laser point cloud data, high-resolution visible light images, infrared images, video data, GIS data, environmental data, power grid data, and satellite remote sensing data, a digital power grid channel is established. By integrating multi-source data, a high-precision three-dimensional power grid model is formed.

[0134] Based on the digitized power grid channel data, a tree barrier analysis model is constructed using deep learning algorithms. The model automatically classifies point clouds and identifies and classifies dangerous tree barriers using three-dimensional laser point cloud features of target tree barriers. It also automatically calculates the area of dangerous tree barriers using single-tree segmentation technology.

[0135] Based on the data obtained from the tree barrier analysis model, an electronic fence technology is designed to predict tree felling compensation. By setting distance thresholds and risk levels, trees that need to be felled are marked and managed, and potential compensation amounts are predicted.

[0136] A mobile application is developed to realize real-time analysis of tree barrier data streams and closed-loop management of tree barrier management business processes. The mobile application can receive task orders automatically generated by the system and provide accurate navigation paths for construction personnel.

[0137] After completing the tree felling task, the construction personnel provide real-time feedback and task confirmation through the mobile application. The system verifies the feedback data to ensure the effectiveness of the task execution. The experimental results are shown in Table 1.

[0138] Table 1 Experimental data table

[0139] The data shows that when the tree barrier height is high (such as tree barrier 6 at 25 meters), the risk level is also high (risk level 10). This indicates that the system can accurately assess the risk level of tree barriers and generate corresponding priority processing tasks.

[0140] The relationship between tree barrier area and compensation amount shows that the system can reasonably estimate compensation amounts based on the size of the tree barrier area. For example, tree barrier 6 has an area of 80 square meters, corresponding to a compensation amount of 2000 yuan. This precise compensation estimation improves the fairness and transparency of management.

[0141] The relationship between electronic fence distance and risk level shows that by setting an electronic fence distance threshold, high-risk tree barriers can be effectively managed and controlled. For example, tree barrier 6 has an electronic fence distance of 7 meters, ensuring effective management of high-risk tree barriers.

[0142] The data shows that the task completion rate is generally high (such as tree barrier 1 at 98% and tree barrier 7 at 100%), indicating that the system can efficiently allocate tasks and ensure their completion. This reflects the efficiency and reliability of the closed-loop management system.

[0143] The system can receive real-time feedback from construction personnel through the mobile application and verify the feedback data to ensure the effectiveness of task execution.

[0144] It should be noted that the above examples are only used to illustrate the technical solutions of the present application but not limit the present application. Although the present application is described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or equivalently replaced, without departing from the spirit and scope of the technical solutions of the present application, which should be covered in the scope of the claims of the present application.

Claims

1. A method for deepening a tree barrier of a power transmission corridor based on a three-dimensional laser point cloud, characterized in that, The application relates to a tree barrier analysis and management system based on unmanned aerial vehicle (UAV) and helicopter. The system comprises the following steps: collecting multi-source data by means of a multi-sensor system mounted on a UAV and a helicopter, and pre-processing the multi-source data; establishing a digital power grid channel by means of the processed multi-source data and constructing a tree barrier analysis model; designing a tree felling and compensation prediction model based on the data analyzed by the tree barrier analysis model and electronic fence technology; and developing a mobile application for analyzing tree barrier data flow and closed-loop management of tree barrier management business flow. The multi-sensor system comprises a laser radar system, a high-resolution digital camera, an infrared camera, a high-frequency video camera, an environmental sensor, a device sensor, a GIS platform and a high-resolution remote sensing satellite. The multi-source data comprises three-dimensional laser point cloud data, high-resolution visible light images, infrared images, video data and GIS data. The step of establishing a digital power grid channel by means of the processed multi-source data comprises the following steps: processing three-dimensional laser point cloud data to obtain a three-dimensional point cloud data set; using a PointNet++ model to construct a classification function to classify the three-dimensional point cloud data set; and identifying and separating different types of objects such as towers, conductors, ground and vegetation, which is expressed by the formula C (x, y, z) = softmax (W * phi (x, y, z) + b).

2. The method for deepening the understanding of tree obstacles in power transmission channels based on three-dimensional laser point clouds as described in claim 1, characterized in that: In the formula, C (x, y, z) represents the probability distribution of different categories, softmax is an activation function, W represents a weight matrix, phi (x, y, z) represents a feature vector of point (x, y, z), b represents a bias term, and (x, y, z) represents a position in space. When 50 <= D (x, y, z) <= 100, it indicates that the power transmission channel at the position has a high spatial density and complexity, and a high risk level, which needs to be processed and monitored preferentially.

3. The method of claim 2, wherein the method further comprises: determining a three-dimensional (3D) point cloud of the transmission corridor; and determining a 3D point cloud of the tree barrier. When 10 <= D (x, y, z) < 50, it indicates that the power transmission channel at the position has a medium complexity and a medium risk level, which needs to be monitored and maintained regularly to ensure the safety of the power grid channel. When 0 < D (x, y, z) < 10, it indicates that the power transmission channel at the position is relatively simple or sparse, and has a low risk level, and the power grid channel is in a safe state. C(x, y, z) = [p 杆塔 ,p 导线 ,p 地面 ,p 植被 ] ; if p 杆塔 > 0.65 then classify as tower, if p 导线 > 0.6 then classify as conductor, if p 地面 > 0.7 then classify as ground, if p 植被 > 0.55 then classify as vegetation; High resolution visible light imagery is acquired by a high resolution digital camera, acquiring clear visual data and converting to a light reflectance distribution map: Obtain infrared image through infrared camera, identify hot abnormal area: Obtaining video data through a high-frequency video camera, obtaining dynamic monitoring information, and extracting dynamic change features: Obtaining GIS data through a GIS platform, smoothing and analyzing the GIS data, and extracting spatial distribution characteristics: The formula for establishing a digital power grid channel is expressed as: Where D(x,y,z) represents the model of the digital power grid channel, (x,y,z) represents the coordinates in three-dimensional space, (x... i ,y i ,z i Let f(x, ξ, y, η) represent the coordinates of the i-th point in the point cloud data, h(x-ξ, y-η) represent the kernel function, f(ξ, η) represent the distribution function of the input data, ξ and η represent the integration variables, R(λ, θ) represent the light reflectance, λ is the wavelength, θ is the angle, α(λ) represents the light absorption coefficient, d represents the distance, σ represents the Stefan-Boltzmann constant, and T s The surface temperature of an object, T b Let A represent the background temperature, A represent the integral area, and N represent the total number of frames in the video data. i ,y i ) represents the coordinates of the i-th frame in the video data, (μ x ,μ y ) represents the mean of the coordinates in the video data, σ represents the standard deviation of the coordinates in the video data, exp represents the exponential function, I(x,y) represents the high-resolution visible light image, T(x,y) represents the temperature distribution of the infrared image, V(t) represents the dynamic characteristics in the video data, and G(x,y) represents the processed GIS data. The step of constructing a tree barrier analysis model comprises the following steps: based on a deep learning point cloud automatic classification technology, combining three-dimensional laser point cloud features of target tree barriers, and combining single-tree segmentation technology, a dangerous tree barrier area is automatically counted to construct a tree barrier analysis model. The formula represents a logical function. E is an indicator function, k (x, y, z) represents a Gaussian function, alpha (lambda) represents an optical absorption coefficient, d represents a distance, and exp represents an exponential function. When 0.75 < P (x, y, z) <= 1, it indicates that tree felling and compensation in the region are high-risk and need to be processed preferentially.

4. The method of claim 3, wherein the method further comprises: determining a distance between the first point and the second point; and determining a distance between the first point and the third point. When 0 < P (x, y, z) < 0.75, it indicates that tree felling and compensation in the region are low-risk and do not need to be processed. Segmenting individual trees from point cloud data: By integrating the height T(x, y, z) of each point within the area Ω with the dangerous height threshold T threshold Comparison is made, and if the height exceeds the threshold, it belongs to the dangerous hedge: A tree barrier analysis model is constructed by a PointNet++ neural network, and is expressed by a formula: Wherein, M represents the constructed tree barrier analysis model, the value threshold of M is between 0 and 1; Ω represents the integral region, n represents the number of point cloud data points, (x, y, z) represents the coordinates in the three-dimensional space, (x i ,y i ,z i ) represents the coordinates of the i-th point, and ∈ represents the threshold of single tree segmentation. denotes an indicator function, W denotes a weight matrix, φ(x, y, z) denotes a point cloud feature vector, b denotes a bias term, softmax denotes an activation function, T(x, y, z) denotes a tree barrier height, T threshold denotes a dangerous height threshold, S(x, y, z) denotes a segmentation result, A denotes a dangerous tree barrier area.

5. The method of claim 4, wherein the method further comprises: determining a distance between the three-dimensional laser point cloud and the power transmission corridor tree barrier; and determining a distance between the three-dimensional laser point cloud and the power transmission corridor tree barrier based on the distance between the three-dimensional laser point cloud and the power transmission corridor tree barrier. The tree felling compensation prediction model is designed based on the data analyzed by the tree barrier analysis model and the electronic fence technology, including identifying and classifying the trees in the electronic fence by setting a distance threshold: wherein The step of developing a mobile application for analyzing tree barrier data flow and closed-loop management of tree barrier management business flow comprises the following steps: when the value of the tree barrier analysis model M is greater than 0.5, the tree barrier risk level is high, the tree barrier density in the region is high, and there are many dangerous tree barriers, which need to be processed preferentially; the system automatically generates a preferential processing task and sends it to a mobile application terminal. By comparing the height of the trees in the integral area with the dangerous height threshold, combined with the results of the electronic fence technology, the indemnity priority of each tree is calculated: wherein P denotes a tree felling indemnification prediction model, Ω denotes an integration region, T(x, y, z) denotes a tree barrier height, T threshold denotes a dangerous height threshold, ​ ​ ​ 6. The method of claim 5 and 4, wherein the method is characterized in that: ​ When the value of the tree barrier analysis model M is between 0 and 0.5, the tree barrier risk level is low, the tree barrier density in the area is low and there is no dangerous tree barrier, and the queue is processed according to the set priority order; The evaluation accuracy of each risk level needs to reach more than 95%, and when it is lower than this standard, the system will optimize and adjust the algorithm.

7. The method of claim 6, wherein the method further comprises: determining a height of the tree barrier; and determining a height of the power transmission corridor; and determining a distance between the tree barrier and the power transmission corridor based on the height of the tree barrier and the height of the power transmission corridor. The development of mobile application analysis tree barrier data flow and closed-loop management of tree barrier management business flow also includes that the system real-time statistics tree barrier area, and the tree barrier information exceeding the threshold value generates task work order and sends to the mobile terminal of the relevant construction personnel, when the error exceeds 5%, the system re-counts and calibrates; Using GPS and GIS technology, real-time positioning of tree barrier position, providing accurate navigation for construction personnel When the positioning error exceeds 5 meters, the system automatically adjusts and recalculates the navigation path to ensure that the construction personnel can accurately arrive at the target position; When the construction personnel complete the tree cutting task, upload the before and after cutting photos and on-site situation explanation, and the system verifies the feedback data; if the feedback data is incomplete or incorrect, the system prompts to resubmit. When tree cutting involves compensation, the system automatically generates an electronic fence and marks the trees to be compensated, and the compensation amount and related information are notified to the relevant personnel through the mobile terminal, and the compensation record in the management system is updated in real time; The overall performance of the system is evaluated every month to make the task completion rate reach more than 95% and the processing time be shortened by more than 10%.

8. A three-dimensional laser point cloud-based power transmission corridor tree barrier management deepening system using the method of any one of claims 1-7, characterized in that: A data acquisition module collects multi-source data through a multi-sensor system carried by a UAV and a helicopter, and pre-processes the multi-source data; A tree barrier analysis module uses the pre-processed multi-source data to establish a digital power grid corridor and construct a tree barrier analysis model; A clearing and compensation prediction module designs a tree cutting clearing and compensation prediction model based on the data analyzed by the tree barrier analysis model and electronic fence technology; A mobile application module develops a mobile application to analyze tree barrier data flow and closed-loop manage tree barrier management business flow.

9. A computer device comprising: A memory and a processor; The memory stores a computer program, and the processor executes the computer program to implement the steps of the three-dimensional laser point cloud-based power transmission corridor tree barrier deepening method.

10. A computer readable storage medium having stored thereon a computer program, characterized in that: The computer program is executed by the processor to implement the steps of the three-dimensional laser point cloud-based power transmission corridor tree barrier deepening method.

Images