Unmanned aerial vehicle autonomous inspection processing method and system based on distribution network overhead line equipment

Abstract

The invention discloses an unmanned aerial vehicle autonomous inspection processing method and system based on distribution network overhead line equipment, and belongs to the technical field of unmanned aerial vehicle inspection. According to the method, firstly, a laser point cloud inspection information set and an inspection route record of a target unmanned aerial vehicle are obtained, target inspection point cloud element knowledge is generated through point cloud element identification branches in an inspection scheduling processing network, and line crossing information between the unmanned aerial vehicles is considered, so that the inspection safety is improved. The route optimization decision branch is combined with target inspection point cloud element knowledge and inspection route records to generate target route planning suggestion features, and flight safety, inspection efficiency and equipment coverage degree are ensured by integrating multiple factors. The route clustering processing branch groups the target route planning suggestion features, finally determines an unmanned aerial vehicle routing inspection planning cluster through the routing inspection energy efficiency discrimination weight of the grouping result, and improves the routing inspection efficiency and reduces safety accidents in combination with a ball query safety verification technology.

Description

Technical Field

[0001] The present invention belongs to the technical field of unmanned aerial vehicle inspection, and in particular relates to an unmanned aerial vehicle autonomous inspection processing method and system based on distribution network overhead line equipment. Background Art

[0002] Drones have extensive and important applications in the field of distribution network overhead line inspection. With its own mobility and a variety of detection equipment, such as high-definition cameras, laser radars, thermal imagers, etc., it conducts a full range of inspections on the tower body, tower head, tower body, tower base, cross arm, transformer, switch and other equipment in the distribution network overhead line, as well as the line direction, channel and surrounding environment. The high-definition camera can intuitively obtain the appearance of the equipment to check whether there are problems such as damage, corrosion, loose connectors, etc.; the laser point cloud data collected by the laser radar can accurately construct a three-dimensional model of the equipment and the environment, and analyze structural deformation and other conditions; the thermal imager is used to detect the temperature distribution of equipment such as transformers to determine whether there is an overheating anomaly. At the same time, when inspecting the channel, the drone can monitor the distance between the surrounding trees and the line, whether the building is illegally built and its interference with the electromagnetic environment. However, in the actual application of distribution network overhead line inspection, traditional technology has certain shortcomings in the route planning method of drones, making it difficult to ensure inspection efficiency and safety. Summary of the invention

[0003] The present invention provides a method and system for autonomous inspection and processing by unmanned aerial vehicles based on overhead line equipment of a distribution network, which can solve or partially solve the technical problems involved in the above-mentioned background technology.

[0004] The present invention provides an autonomous inspection and processing method for unmanned aerial vehicles based on overhead line equipment of a distribution network, which is applied to an autonomous inspection and processing system for unmanned aerial vehicles. The method comprises: obtaining X laser point cloud inspection information sets and X inspection route records of X target unmanned aerial vehicles, wherein each laser point cloud inspection information set in the X laser point cloud inspection information sets comprises key point detection information and flight status monitoring information, wherein the key point detection information is used to characterize the information obtained after the target unmanned aerial vehicle performs laser point cloud scanning and collection of overhead line equipment of a distribution network within a specified inspection task cycle, and the flight status monitoring information is used to characterize the monitoring result corresponding to the posture change of the target unmanned aerial vehicle within the specified inspection task cycle, and X is greater than or equal to 1; using the X laser point cloud inspection information sets as the input information of a point cloud element recognition branch in an inspection scheduling processing network, and generating X target inspection point cloud element knowledge through the point cloud element recognition branch, wherein the inspection scheduling processing network is debugged based on the initial inspection route record, the first inspection route record sample and the second inspection route record sample of the unmanned aerial vehicle example. , the initial inspection route record is used to characterize the route intersection information between the drone sample and the Y coordinated drones corresponding to the drone sample, the first inspection route record sample is used to characterize the route intersection information between the drone sample and the Z coordinated drones determined from the Y coordinated drones, and the second inspection route record sample is used to characterize the route intersection information corresponding to the drone sample and the conflicting drone, Z is greater than or equal to 1 and less than or equal to Y; the X target inspection point cloud element knowledge and the X inspection route records are used as the input information of the route optimization decision branch in the inspection scheduling processing network, and X target route planning suggestion features are generated through the route optimization decision branch; the X target route planning suggestion features are clustered based on the route clustering processing branch to generate Q clustering results, wherein Q is greater than or equal to 1; the inspection energy efficiency discrimination weights of the Q clustering results are identified, and the drone inspection planning cluster is determined based on the inspection energy efficiency discrimination weights of the Q clustering results, wherein the drone inspection planning cluster includes at least one of the target drones.

[0005] The present invention provides an unmanned aerial vehicle autonomous inspection processing system, comprising at least one processor and a memory; the memory stores computer-executable instructions; the at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes the above method.

[0006] The present invention provides a readable storage medium, on which a program or instruction is stored, and the program or instruction implements the steps of the above method when executed by a processor.

[0007] Traditional technologies make it difficult to fully obtain detailed information on distribution network overhead line equipment and drone flight status, which leads to inaccurate judgments on equipment status and inability to detect potential fault risks in a timely manner. During drone collaborative inspections, route conflicts and other issues are prone to occur due to the inability to fully consider the cross-line information between drones, affecting inspection efficiency and safety. Route planning lacks comprehensive consideration of multiple factors, such as flight safety, inspection efficiency, and equipment coverage, resulting in poor inspection results. In addition, there is a lack of effective classification processing for route planning, which cannot provide an effective basis for the final inspection cluster determination. Traditional methods also lack a cluster determination method combined with safety verification, which cannot improve inspection efficiency and avoid safety accidents.

[0008] Based on this, the present invention realizes the effective planning of the UAV inspection route through a series of steps. First, the laser point cloud inspection information set and inspection route record of the target UAV are obtained, which contain key point detection information and flight status monitoring information, providing the basis for the equipment status and UAV flight status for subsequent processing. Through the point cloud element recognition branch in the inspection scheduling processing network, the target inspection point cloud element knowledge is generated. This is due to the debugging of the network based on a variety of inspection route records, which can fully consider the line crossing information between UAVs and improve the accuracy of the evaluation of the distribution network overhead line equipment. The line optimization decision branch combines the target inspection point cloud element knowledge and the inspection route record to generate the target route planning suggestion features, and comprehensively ensures flight safety, inspection efficiency and equipment coverage. The route clustering processing branch groups the target route planning suggestion features to provide a basis for determining the UAV inspection planning cluster. Finally, the UAV inspection planning cluster is determined by identifying the inspection energy efficiency judgment weight of the clustering result, and combined with the ball query safety verification technology to improve the inspection efficiency and reduce safety accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 The present invention provides a flow chart of a method for autonomous inspection of overhead line equipment using a UAV.

[0010] Figure 2 This is a structural schematic diagram of an unmanned aerial vehicle autonomous inspection and processing system provided by the present invention. DETAILED DESCRIPTION

[0011] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0012] The terms "first", "second", etc. in the present invention are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present invention can be implemented in an order other than those illustrated or described herein, and the objects distinguished by "first", "second", etc. are generally of the same type, and the number of objects is not limited. For example, the first object can be one or more. In addition, in the present invention, "and / or" means at least one of the connected objects, and the character " / " generally indicates that the objects associated with each other are in an "or" relationship.

[0013] Figure 1 A method for autonomous inspection and processing of a UAV based on overhead line equipment of a distribution network is shown, which is applied to an autonomous inspection and processing system of a UAV. The method includes the following steps 110 to 150.

[0014] Step 110: Obtain X laser point cloud inspection information sets and X inspection route records of X target UAVs.

[0015] In the present invention, each of the X laser point cloud inspection information sets includes key point detection information and flight status monitoring information, the key point detection information is used to characterize the information obtained after the laser point cloud scanning and collection of the distribution network overhead line equipment of the target UAV within the specified inspection task cycle, and the flight status monitoring information is used to characterize the monitoring results corresponding to the posture changes of the target UAV in the specified inspection task cycle, and X is greater than or equal to 1.

[0016] Furthermore, the laser point cloud scanning, collection and processing objects of the distribution network overhead line equipment include but are not limited to: tower body, tower head, tower body, tower base, direction, channel, cross arm, transformer, switch and surrounding environment.

[0017] In the present invention, for each target UAV, there is a corresponding laser point cloud inspection information set. Here, X represents the number of target UAVs, and X is greater than or equal to 1. Each laser point cloud inspection information set contains two important information, namely, key point detection information and flight status monitoring information.

[0018] Key point detection information is the information obtained after the target drone performs laser point cloud scanning and collection of distribution network overhead line equipment within the specified inspection mission cycle. The laser point cloud scanning and collection processing objects of distribution network overhead line equipment are very wide, such as the tower body, tower head, tower body, tower base, direction, channel, cross arm, transformer, switch, surrounding environment, etc. For example, when performing laser point cloud scanning and collection on the tower body, the coordinate information and shape information of each part of the tower body may be obtained. For example, the tower body is a cylindrical shape. Through laser point cloud scanning and collection, the coordinates of a large number of points on the surface of the cylinder can be obtained, thereby determining its height, diameter and other key size information. This information is very important for judging the structural integrity of the tower body.

[0019] The flight status monitoring information is based on the monitoring results corresponding to the attitude changes of the target UAV during the specified inspection mission cycle. For example, the attitude of the UAV includes its pitch angle, roll angle, and yaw angle. During the inspection process, if the UAV encounters airflow interference, its attitude will change. For example, at a certain moment, the pitch angle of the UAV suddenly changes from 0° to 5°, and this change will be recorded by the flight status monitoring system. By monitoring these attitude changes, the flight stability of the UAV can be analyzed to determine whether there are abnormal flight conditions, such as whether it is disturbed by strong winds or whether its own flight control system has a fault.

[0020] At the same time, it is also necessary to obtain the inspection route records of each target drone. These inspection route records are of great significance for subsequent route planning and optimization decisions. For example, the inspection route record may contain information such as the drone's take-off point, the coordinates of each checkpoint passed, the flight altitude, the flight speed, and the final landing point.

[0021] Step 120: Use the X laser point cloud inspection information sets as the incoming information of the point cloud element recognition branch in the inspection scheduling processing network, and generate X target inspection point cloud element knowledge through the point cloud element recognition branch.

[0022] In the present invention, the inspection scheduling processing network is debugged based on the initial inspection route record, the first inspection route record sample and the second inspection route record sample of the drone sample. The initial inspection route record is used to characterize the route intersection information between the drone sample and the Y cooperative drones corresponding to the drone sample. The first inspection route record sample is used to characterize the route intersection information between the drone sample and the Z cooperative drones determined from the Y cooperative drones. The second inspection route record sample is used to characterize the route intersection information corresponding to the drone sample and the conflicting drones, and Z is greater than or equal to 1 and less than or equal to Y.

[0023] In the present invention, the initial inspection route record is used to characterize the route intersection information of the drone sample and the Y coordinated drones corresponding to the drone sample. For example, there is a drone sample, and the number of drones cooperating with it is Y=3. The initial inspection route record will record the route intersection of this drone sample and these three coordinated drones during the flight. If the route of the drone sample is straight line A, and the route of the coordinated drone 1 is straight line B, when straight line A and straight line B intersect, the initial inspection route record will record the coordinates of the intersection point, the flight altitude at the time of intersection, and other information.

[0024] The first inspection route record sample is used to characterize the route intersection information between the drone sample and the Z cooperative drones determined from the Y cooperative drones, where Z is greater than or equal to 1 and less than or equal to Y. For example, Y=5, Z=3. This means that 3 of the 5 cooperative drones are selected to determine the route intersection information with the drone sample. For example, if the selected cooperative drones are 1, 3, and 5, the first inspection route record sample will record in detail the route intersection information between the drone sample and these 3 cooperative drones, such as the frequency of intersection, the relative speed at the time of intersection, etc.

[0025] The second inspection route record sample is used to characterize the route intersection information corresponding to the drone sample and the conflicting drone. The conflicting drone may be a drone that has a potential conflict with the drone sample due to unreasonable route planning or unexpected circumstances. For example, when a drone sample is inspecting along a normal route, an unplanned drone suddenly enters its inspection area, and this drone is regarded as a conflicting drone. The second inspection route record sample will record the route intersection between them, including the intersection angle, distance and other information.

[0026] Furthermore, the working principle of the point cloud feature recognition branch is as follows: after X laser point cloud inspection information sets are input into the point cloud feature recognition branch of the inspection dispatch processing network, the point cloud feature recognition branch will process the information. The point cloud feature recognition branch may adopt a convolutional neural network (CNN) model in a deep learning algorithm. For example, in the CNN model, the input laser point cloud inspection information set is processed by a convolution layer, a pooling layer, and the like. For example, the convolution layer uses a 3*3 convolution kernel with a step size of 1, and the features in the laser point cloud inspection information set can be extracted through the convolution operation. The pooling layer uses a 2*2 maximum pooling operation, which can reduce the data dimension and the amount of calculation. After multiple layers of convolution and pooling operations, X target inspection point cloud feature knowledge is finally generated through a fully connected layer. These target inspection point cloud feature knowledge may include more accurate feature descriptions of distribution network overhead line equipment, such as damage assessment of the tower body, transformer working status analysis, and other knowledge based on point cloud data.

[0027] Step 130: Using the X target inspection point cloud element knowledge and the X inspection route records as input information of a route optimization decision branch in the inspection scheduling processing network, and generating X target route planning suggestion features through the route optimization decision branch.

[0028] In this step, the line optimization decision branch receives X target inspection point cloud element knowledge and X inspection route records as input information. These input information include the analysis results of the distribution network overhead line equipment in the previous step and the flight history of the drone. The line optimization decision branch may use a genetic algorithm to generate route planning suggestion features. In the genetic algorithm, the target inspection point cloud element knowledge and the inspection route record are first encoded, for example, the flight altitude, speed, coordinates of the passing points of the drone are encoded into a binary string. For example, the target inspection point cloud element knowledge includes the damage of the tower body (represented by a value between 0 and 1, 0 means no damage, 1 means severe damage) and the working state of the transformer (represented by a value between -1 and 1, -1 means fault, 1 means normal operation), as well as the information in the inspection route record that the flight altitude is 100 meters and the speed is 5 meters / second. After encoding this information, an initial population is formed.

[0029] Then, the advantages and disadvantages of each individual (encoded route plan) are evaluated according to the set fitness function. The fitness function may comprehensively consider factors such as the flight safety, inspection efficiency, and coverage of overhead line equipment of the distribution network of the drone. For example, the fitness function can be expressed as: F = a*S+b*E+c*C, where F is the fitness value, S is the flight safety index (the value range is 0-1, 1 means the safest), E is the inspection efficiency index (such as the number of equipment inspected per unit time), C is the coverage of overhead line equipment of the distribution network (the value range is 0-1, 1 means full coverage), and a, b, c are weight coefficients, which are set according to actual needs.

[0030] Through selection, crossover and mutation operations, the population is continuously evolved, and finally X target route planning suggestion features are obtained. For example, after multiple generations of evolution, the target route planning suggestion features obtained may include new flight altitude, speed adjustment suggestions, and optimization of waypoints, etc. For example, the flight altitude is adjusted from 100 meters to 80 meters, the speed is adjusted from 5 meters / second to 6 meters / second, and a new waypoint is added to more comprehensively inspect a key equipment.

[0031] Step 140: cluster the X target route planning suggestion features based on the route clustering processing branch to generate Q clustering results.

[0032] In the present invention, Q is greater than or equal to 1. The clustering principle of the route clustering processing branch is as follows: the route clustering processing branch performs a clustering operation on X target route planning suggestion features. A clustering algorithm such as a K-Means clustering algorithm may be used. For example, X=10, that is, there are 10 target route planning suggestion features.

[0033] In the K-Means clustering algorithm, the number of clusters Q must be determined first. If Q = 3, the algorithm will randomly select 3 initial cluster centers. For example, 3 of the 10 target route planning suggestion features are randomly selected as the initial cluster centers. Then, the distance from each target route planning suggestion feature to these 3 cluster centers is calculated. The distance can be calculated using the Euclidean distance formula. For each dimension in the target route planning suggestion feature (such as flight altitude, speed, waypoint coordinates, etc.), the sum of the squares of the differences between it and the corresponding dimension of the cluster center is calculated, and then the square root is taken to obtain the distance.

[0034] Assign each target route planning suggestion feature to the class to which the nearest cluster center belongs based on the distance. Then, recalculate the cluster center of each class, that is, calculate the average value of all target route planning suggestion features in each class in each dimension as the new cluster center. Repeat this process until the cluster center no longer changes significantly or reaches the preset number of iterations. Finally, Q = 3 cluster results are obtained, each of which contains several target route planning suggestion features with similar characteristics.

[0035] Step 150: Identify the inspection energy efficiency judgment weights of the Q sub-cluster results, and determine the drone inspection planning cluster based on the inspection energy efficiency judgment weights of the Q sub-cluster results.

[0036] In the present invention, the drone inspection planning cluster includes at least one target drone. Furthermore, in the process of determining the drone inspection planning cluster, the ball query safety verification technology is also combined, so that the route planning of each target drone in the drone inspection planning cluster that needs to perform autonomous route planning can be accurately and safely determined, thereby improving the inspection efficiency and reducing safety accidents during the inspection process.

[0037] In detail, step 150 involves two key steps: identifying inspection energy efficiency judgment weights and determining drone inspection planning clusters.

[0038] Identification of inspection energy efficiency discrimination weights: For Q cluster results, their inspection energy efficiency discrimination weights need to be identified. The determination of inspection energy efficiency discrimination weights may involve multiple factors. For example, if the target route planning suggestion feature in a cluster result can cover more key distribution network overhead line equipment, its inspection energy efficiency discrimination weight may be higher. For example, the target route planning suggestion feature in cluster result 1 can cover 80% of key equipment, and the target route planning suggestion feature in cluster result 2 can only cover 60% of key equipment, then the inspection energy efficiency discrimination weight of cluster result 1 may be higher than that of cluster result 2.

[0039] In addition, the flight safety and efficiency of the drone when executing the route planning recommendation features in the cluster result may also be considered. For example, although the route planning recommendation features in cluster result 3 do not cover the largest number of key equipment, they have the highest flight safety and high flight efficiency (such as the shortest flight path and the most stable flight speed), so its inspection energy efficiency judgment weight will also be correspondingly high.

[0040] Determine the drone inspection planning cluster: In the process of determining the drone inspection planning cluster, the spherical query safety verification technology is also combined. The spherical query safety verification technology can be understood as building a virtual spherical space with the drone as the center. For example, the safety radius of the drone is R = 10 meters. If there are other obstacles or drones within this radius, there may be safety risks.

[0041] According to the inspection energy efficiency judgment weights of the Q cluster results, the target drones in the cluster results with higher weights are included in the drone inspection planning cluster. For example, if the inspection energy efficiency judgment weight of cluster result 1 is the highest and it contains target drones 1, 2, and 3, then target drones 1, 2, and 3 will be determined as members of the drone inspection planning cluster. In this way, the route planning of each target drone in the drone inspection planning cluster that requires autonomous route planning can be accurately and safely determined, thereby improving inspection efficiency and reducing safety accidents during the inspection process.

[0042] It can be understood that the further introduction of the above-mentioned target inspection point cloud element knowledge and target route planning suggestion features is as follows.

[0043] 1. Quantitative characteristics of target inspection point cloud element knowledge

[0044] Quantitative characteristics related to the structure of overhead line equipment in distribution networks

[0045] 1) Quantification of tower structure characteristics

[0046] Quantification of height deviation: For the tower body, the point cloud data collected by laser point cloud scanning can be used to calculate the actual height of the tower. For example, under ideal conditions, the design height of the tower is (H_d=50) meters, and the actual height calculated by point cloud analysis is (H_a). Then the height deviation is (DeltaH=|H_d-H_a|). For example, if (H_a=49.5) meters, then (DeltaH=|50-49.5|=0.5) meters. This quantitative value can be used as part of the target inspection point cloud element knowledge to determine whether the tower has settlement or foundation damage.

[0047] Quantification of tower inclination: Establish a coordinate system with the center of the bottom of the tower as the origin, and measure the horizontal offset of the top of the tower relative to the bottom center. For example, the radius of the bottom of the tower is (r=2) meters, and the offset of the top of the tower in the (x) direction is measured to be (x_o=0.2) meters, and the offset in the (y) direction is (y_o=0.1) meters. Then the inclination of the tower is (theta=arctan(sqrt{x_o^2+y_o^2} / h)), where (h) is the height of the tower. If (h=50) meters, (theta=arctan(sqrt{0.2^2+0.1^2} / 50)approx0.0025) radians. This quantified inclination value is an important indicator of the stability of the tower structure in the target inspection point cloud element knowledge.

[0048] 2) Quantification of cross arm deformation

[0049] Quantification of crossarm length change: The designed crossarm length is (L_d = 5) meters, and the actual crossarm length obtained through laser point cloud analysis is (L_a). The change in crossarm length (DeltaL = |L_d-L_a|). For example, when (L_a = 4.9) meters, (DeltaL = |5-4.9| = 0.1) meters. This quantitative value helps to determine whether the crossarm has undergone tensile or compressive deformation.

[0050] Quantification of crossarm curvature: Consider the crossarm as a straight line segment (ideal state), and fit the actual shape curve of the crossarm through the point cloud data. Calculate the maximum vertical distance (d) between the curve and the ideal straight line, and use the crossarm length (L) as the reference, the curvature (k = d / L). If (d = 0.1) meters, (L = 5) meters, then (k = 0.1 / 5 = 0.02). This quantified curvature is the key quantitative feature of the crossarm structural integrity in the target inspection point cloud element knowledge.

[0051] Quantitative characteristics related to equipment operating status

[0052] 1) Quantification of transformer oil temperature

[0053] When the transformer is operating normally, the oil temperature has a reasonable design range, such as (T_{min}=50^{circ}C) to (T_{max}=80^{circ}C). The actual oil temperature (T_a) of the transformer is obtained by laser point cloud scanning in combination with a temperature sensor (for example, point cloud data can assist in locating the data obtained by the temperature sensor). If (T_a=70^{circ}C), this oil temperature value is a quantitative feature in the target inspection point cloud element knowledge, which can be used to determine whether the operating status of the transformer is normal.

[0054] 2) Quantification of switch opening and closing status

[0055] The switch state is defined as a binary variable, (S=0) indicates that the switch is closed, and (S=1) indicates that the switch is open. The actual switch state is determined by combining laser point cloud data with image recognition technology (for example, point cloud data can assist in locating the switch and performing image recognition). This (S) value, as part of the target inspection point cloud element knowledge, directly reflects the operating state of the switch.

[0056] Quantitative features related to the surrounding environment

[0057] 1) Quantification of the safe distance between surrounding vegetation and lines

[0058] For overhead lines in distribution networks, the safety distance between the surrounding vegetation and the line is specified to be (D_s=3) meters. The actual distance (D_a) between the vegetation and the line is obtained through laser point cloud scanning. Distance deviation (DeltaD=|D_s-D_a|). For example, when (D_a=2.5) meters, (DeltaD=|3-2.5|=0.5) meters. This quantified distance deviation is an important indicator of the impact of the surrounding environment on line safety in the target inspection point cloud element knowledge.

[0059] 2) Quantification of distance between surrounding buildings and towers

[0060] For example, the safety distance between surrounding buildings and poles is (D_{bt}=10) meters. The actual distance calculated by laser point cloud data is (D_{abt}). Distance deviation (DeltaD_{bt}=|D_{bt}-D_{abt}|). For example, when (D_{abt}=8) meters, (DeltaD_{bt}=|10-8|=2) meters. This quantitative value can reflect the potential impact of surrounding buildings on the stability of poles and towers, and is part of the knowledge of target inspection point cloud elements.

[0061] 2. Quantitative Characteristics of Target Route Planning Suggestion Features

[0062] Flight path related quantitative features

[0063] 1) Quantification of flight distance

[0064] The flight distance in the target route planning suggestion feature is an important quantitative indicator. For example, the flight distance of the drone from the take-off point (P_1(x_1, y_1, z_1)) to the landing point (P_2(x_2, y_2, z_2)) is (L=sqrt{(x_2-x_1)^2+(y_2-y_1)^2+(z_2-z_1)^2}). For example, (x_1=0, y_1=0, z_1=0), (x_2=100, y_2=100, z_2=50), then: (L=sqrt{ (100-0)^2+ (100-0)^2+ (50-0)^2}=sqrt{10000+10000+2500}=sqrt{22500}=150) meters. The quantitative value of this flight distance directly affects the efficiency and energy consumption of inspection.

[0065] 2) Quantification of flight trajectory smoothness

[0066] Consider the flight trajectory as a curve consisting of a series of discrete points (P_i(x_i, y_i, z_i))((i=1, 2, cdots, n)). Calculate the vector between two adjacent points (vec{v}_i=overrightarrow{P_{i}P_{i+1}}=(x_{i+1}-x_i,y_{i+1}-y_i,z_{i+1}-z_i)).

[0067] Then calculate the angle between adjacent vectors: (theta_i = arccos(frac{vec{v}_icdotvec{v}_{i+1}}{|vec{v}_i|*|vec{v}_{i+1}|})). Flight trajectory smoothness (S = frac{1}{n-1}sum_{i=1}^{n-1}theta_i). A smaller (S) value indicates a smoother flight trajectory, for example (S = 0.1) radians indicate a relatively smooth trajectory, which helps reduce energy consumption during drone flight and improve flight stability.

[0068] Flight parameter related quantitative characteristics

[0069] 1) Quantification of flight speed

[0070] The target route planning suggestion features include the flight speed (v) of the drone in meters per second. For example, (v = 5) meters per second means that the drone flies at this speed. Different flight missions and environments may require different flight speeds. In inspection tasks, it is necessary to comprehensively consider the inspection accuracy and efficiency to determine the appropriate flight speed. If a more detailed inspection is required (such as a close-up inspection of key equipment), the flight speed may need to be reduced; in long-distance line inspections, the flight speed can be appropriately increased.

[0071] 2) Flight altitude quantification

[0072] The flight altitude (h) is also a key quantitative feature, measured in meters. For example, (h = 100) meters. The choice of flight altitude needs to consider multiple factors, such as avoiding obstacles and obtaining a better inspection angle. For the inspection of overhead lines in the distribution network, it may be necessary to determine the appropriate flight altitude based on the height of the line, the surrounding environment (such as the height of buildings and vegetation), etc.

[0073] Inspection coverage related quantitative features

[0074] 1) Quantification of device coverage ratio

[0075] Define the distribution network overhead line equipment set as (E = {e_1, e_2, cdots, e_m}), and the equipment set covered by the target route planning recommendation feature is (E_c = {e_{c1}, e_{c2}, cdots, e_{ck}}). Equipment coverage ratio (C = frac{k}{m}). For example, if there are a total of (m = 10) devices, and the target route planning recommendation feature can cover (k = 8) devices, then (C = frac{8}{10} = 0.8). This quantitative value reflects the comprehensiveness of route planning for equipment inspection.

[0076] 2) Quantification of key equipment coverage

[0077] For the key equipment set (E_{key}={e_{k1},e_{k2},cdots,e_{kn}})((nleqm)), define the key equipment coverage ratio (C_{key}=frac{n_{covered}}{n}), where (n_{covered}) is the number of key equipment covered by the target route planning suggestion feature. For example, there are (n=5) key equipment, and the target route planning suggestion feature can cover (n_{covered}=4) key equipment, then (C_{key}=frac{4}{5}=0.8). This quantitative feature focuses on the inspection coverage of key equipment, because the normal operation of key equipment is crucial to the entire distribution network overhead line system.

[0078] By applying the above technical solution, the laser point cloud inspection information set obtained in step 110 contains rich information. The key point detection information covers the detailed data of many components of the distribution network overhead line equipment, such as the coordinates and shape of the tower body, which helps to accurately judge the structural integrity of the equipment. For example, it can promptly detect slight deformation or damage to the tower and avoid potential line failure risks. The flight status monitoring information can monitor the attitude changes of the drone in real time. For example, through the change records of the pitch angle, roll angle and yaw angle, it can promptly detect abnormal conditions in the flight of the drone, whether it is external airflow interference or its own control system failure, to ensure the smooth progress of the inspection task. The inspection route record provides basic data for subsequent planning and decision-making, and clarifies the key parameters of the drone such as flight path and speed.

[0079] In step 120, the target inspection point cloud element knowledge is generated by the point cloud element recognition branch in the inspection scheduling processing network. This branch is debugged based on multiple inspection route records, and can fully consider the line crossing information between drones, whether it is a collaborative drone or a conflicting drone. The convolutional neural network (CNN) model is used for processing, such as effectively extracting features through specific convolution kernels and pooling operations. The target inspection point cloud element knowledge finally generated can make a more accurate assessment of the distribution network overhead line equipment, such as accurately assessing the degree of damage to the tower and the working status of the transformer, and improving the accuracy of the inspection.

[0080] In step 130, the route optimization decision branch integrates the target inspection point cloud element knowledge and inspection route records, and uses a genetic algorithm to generate target route planning suggestion features. This method can comprehensively consider multiple factors such as flight safety, inspection efficiency, and equipment coverage, and obtain optimized route planning suggestions through fitness function evaluation and population evolution operations, such as adjusting flight altitude, speed, and waypoints, which can not only improve inspection efficiency, but also ensure the comprehensiveness of inspection.

[0081] The route clustering processing branch of step 140 groups the target route planning suggestion features, and uses a clustering algorithm (such as the K-Means clustering algorithm) to group similar route planning features into one category, making subsequent processing more efficient. This helps to find commonalities and differences in different types of route planning, and provides a basis for determining the final drone inspection planning cluster.

[0082] Finally, step 150 determines the inspection planning cluster of the drone by identifying the inspection energy efficiency judgment weight of the clustering result and combining it with the spherical query safety verification technology. The inspection energy efficiency judgment weight determined by considering multiple factors can reasonably screen out efficient route planning clustering results, and then combine it with the spherical query safety verification technology to build a safe space (such as a spherical space with a radius of 10 meters) with the drone as the center to avoid obstacles and interference from other drones, accurately determine the inspection planning cluster of the drone, improve inspection efficiency and reduce safety accidents.

[0083] In some alternative embodiments, the X laser point cloud inspection information sets include X key point detection information and X flight status monitoring information, and the point cloud element recognition branch includes a key point quantification recognition module, a posture point cloud recognition module and a fully connected module; then the step 120 described in which the X laser point cloud inspection information sets are used as the incoming information of the point cloud element recognition branch in the inspection scheduling processing network, and X target inspection point cloud element knowledge is generated through the point cloud element recognition branch, includes: using the X key point detection information as the incoming information of the key point quantification recognition module, and generating X key point quantification knowledge codes through the key point quantification recognition module; using the X flight status monitoring information as the incoming information of the posture point cloud recognition module, and generating X posture point cloud knowledge codes through the point cloud element recognition branch; using the X key point quantification knowledge codes and the X posture point cloud knowledge codes as the incoming information of the fully connected module, and generating X target inspection point cloud element knowledge through the incoming information of the fully connected module.

[0084] In addition, in the following preferred embodiment 1, the key point quantization recognition module includes a vector space conversion unit and a knowledge collision unit; based on this, the X key point detection information is used as the input information of the key point quantization recognition module, and X key point quantization knowledge codes are generated through the key point quantization recognition module, including: based on the key point image semantics of the X key point detection information and the X key point detection information, X local point cloud description semantic graphs are generated; the X local point cloud description semantic graphs are used as the input information of the vector space conversion unit, and X local point cloud based space migration codes are generated through the vector space conversion unit, wherein the The vector space conversion unit is used to sum the corresponding feature members in each of the X local point cloud description semantic maps; the X local point cloud description semantic maps are used as the input information of the knowledge collision unit, and X local point cloud advanced space migration codes are generated by the knowledge collision unit, wherein the knowledge collision unit is used to product the corresponding feature members in each of the X local point cloud description semantic maps, and sum the results of the product of the corresponding feature members; the X local point clouds are integrated based on the space migration coding and the X local point cloud advanced space migration coding to obtain X key point quantization knowledge coding.

[0085] In addition, in the following preferred embodiment 2, the attitude point cloud recognition module includes a residual connection unit and a point cloud reconstruction unit; the X flight status monitoring information is used as the input information of the attitude point cloud recognition module, and X attitude point cloud knowledge codes are generated through the point cloud element recognition branch, including: using the X flight status monitoring information as the input information of the residual connection unit, and generating X*R attitude monitoring data groups through the residual connection unit, wherein the attitude monitoring data group is composed of at least one sensor monitoring data in the flight status monitoring information, and R is an integer greater than 1; fusing the X*R attitude monitoring data groups through an attitude fusion algorithm to obtain X attitude monitoring event sets; using the X attitude monitoring event sets as the input information of the point cloud reconstruction unit, and generating X attitude point cloud knowledge codes through the point cloud reconstruction unit.

[0086] In addition, in the following preferred embodiment 3, the fully connected module includes a data weighting unit and an excitation unit; the X key point quantization knowledge codes and the X posture point cloud knowledge codes are used as the input information of the fully connected module, and X target inspection point cloud element knowledge is generated through the input information of the fully connected module, including: integrating each key point quantization knowledge code in the X key point quantization knowledge codes with each posture point cloud knowledge code in the X posture point cloud knowledge codes to obtain X integrated knowledge codes; using the X integrated knowledge codes as the input information of the data weighting unit, and generating X multi-mode inspection knowledge codes through the data weighting unit; using the X multi-mode inspection knowledge codes as the input information of the excitation unit, and generating X target inspection point cloud element knowledge through the excitation unit.

[0087] It can be understood that the above replaceable embodiments and the corresponding three preferred embodiments are developed around the processing of laser point cloud inspection information sets by drones in the inspection of overhead lines in the distribution network. It mainly involves the processing process of the point cloud element recognition branch of X laser point cloud inspection information sets in the inspection scheduling processing network to generate X target inspection point cloud element knowledge. Among them, the laser point cloud inspection information set contains X key point detection information and X flight status monitoring information, and the point cloud element recognition branch includes a key point quantization recognition module, a posture point cloud recognition module and a fully connected module.

[0088] 1. Key point quantification identification module

[0089] 1.1. Generate local point cloud description semantic map based on key point detection information

[0090] In this module, firstly, based on the key point image semantics of X key point detection information and the X key point detection information itself, X local point cloud description semantic graphs are generated. For example, X=3, for the first key point detection information, it may be the detection information about a certain part of the tower body. The image semantics of this part (such as the representation of the shape features and texture features of the part in the image) is combined with the detection information (such as coordinates, relative positions with other parts, etc.) to generate a local point cloud description semantic graph. This semantic graph can describe in detail the various feature relationships of the local point cloud.

[0091] 1.2 Working process of vector space conversion unit

[0092] Next, the X local point cloud description semantic graphs are used as the input information of the vector space conversion unit. The vector space conversion unit is used to sum the corresponding feature members in each local point cloud description semantic graph in the X local point cloud description semantic graphs. For example, for the coordinate features in the local point cloud description semantic graph, for example, the coordinate feature vector in one local point cloud description semantic graph is ((x_1, y_1, z_1)) and the other is ((x_2, y_2, z_2)), the sum is ((x_1+x_2, y_1+y_2, z_1+z_2)). Through such operations, each local point cloud description semantic graph is processed, and finally X local point clouds are generated based on space migration coding. Taking X=3 as an example, 3 local point clouds based on space migration coding will be obtained.

[0093] 1.3 Working process of knowledge collision unit

[0094] At the same time, the X local point cloud description semantic graphs are used as the input information of the knowledge collision unit. The operation of the knowledge collision unit is to multiply the corresponding feature members in each local point cloud description semantic graph in the X local point cloud description semantic graphs, and then sum the results of the corresponding feature member multiplication. For example, for the coordinate feature vectors ((x_1, y_1, z_1)) and ((x_2, y_2, z_2)) mentioned above, the multiplication is ((x_1x_2, y_1y_2, z_1z_2)), and then all the multiplication results are summed. In this way, X local point cloud advanced spatial migration encodings are generated.

[0095] 1.4. Integration to obtain key point quantitative knowledge encoding

[0096] Finally, the X local point clouds are integrated based on the spatial migration coding and the X local point clouds are integrated based on the advanced spatial migration coding to obtain the X key point quantized knowledge coding. The integration method can be to combine the corresponding coding according to certain rules, such as adding the corresponding elements in the spatial migration coding and the advanced spatial migration coding or adding them according to certain weights.

[0097] 2. Posture point cloud recognition module

[0098] 2.1. Residual connection unit generates posture monitoring data set

[0099] The residual connection unit in the attitude point cloud recognition module takes X flight status monitoring information as input information. The flight status monitoring information contains information such as the pitch angle, roll angle, and yaw angle of the drone. For example, the flight status monitoring information is collected in a certain time series, and there is a set of data at each time point. The residual connection unit processes these data to generate X*R attitude monitoring data groups, where R is an integer greater than 1. For example, X=2, R=3, then 2*3=6 attitude monitoring data groups will be generated. The attitude monitoring data group here consists of at least one sensor monitoring data in the flight status monitoring information. The residual connection unit may adopt the residual connection idea in the residual network (ResNet), by adding the input information to the output information after some convolutional layers, activation functions, etc., so as to retain more original information and improve the learning ability of the network.

[0100] 2.2. Posture Fusion Algorithm Fusion Posture Monitoring Data Set

[0101] Then, through the attitude fusion algorithm, the X*R attitude monitoring data groups are fused to obtain X attitude monitoring event sets. The attitude fusion algorithm can use the extended Kalman filter (EKF) algorithm. For example, for the pitch angle data in the attitude monitoring data group, there may be certain errors and uncertainties in different attitude monitoring data groups. The EKF algorithm fuses the pitch angle data in multiple attitude monitoring data groups through prediction and update steps, using the state equation and measurement equation of the system. Taking the pitch angle data (theta_1), (theta_2), and (theta_3) in the three attitude monitoring data groups as an example, the EKF algorithm will calculate the fused pitch angle (theta) based on their measurement covariance matrix and other information, and finally obtain X attitude monitoring event sets.

[0102] 2.3. Point cloud reconstruction unit generates pose point cloud knowledge encoding

[0103] Finally, the X posture monitoring event sets are used as the input information of the point cloud reconstruction unit, and the point cloud reconstruction unit generates X posture point cloud knowledge encodings. The point cloud reconstruction unit may adopt a point cloud generation algorithm based on deep learning. For example, the posture monitoring event set is used as input and processed by a multi-layer neural network. Each layer of the neural network extracts and transforms the input data. For example, the first layer may be a convolution layer, which uses a 3×3 convolution kernel with a step size of 1 to perform a convolution operation on the input posture monitoring event set to extract features. After multiple layers of such operations, X posture point cloud knowledge encodings are finally generated.

[0104] 3. Fully connected module

[0105] 3.1 Integrated Knowledge Encoding

[0106] The fully connected module first integrates each key point quantization knowledge code in the X key point quantization knowledge codes with each posture point cloud knowledge code in the X posture point cloud knowledge codes to obtain X integrated knowledge codes. The integration method can be a splicing operation. For example, the key point quantization knowledge code is (K = (k_1, k_2, cdots, k_n)), the posture point cloud knowledge code is (A = (a_1, a_2, cdots, a_m)), and the integrated knowledge code can be (I = (k_1, k_2, cdots, k_n, a_1, a_2, cdots, a_m)).

[0107] 3.2. Data weighting unit generates multi-mode inspection knowledge encoding

[0108] Next, X integrated knowledge codes are used as the input information of the data weighting unit, and X multi-mode inspection knowledge codes are generated through the data weighting unit. The data weighting unit may assign different weights according to the importance of different features. For example, higher weights are assigned to some key features in the integrated knowledge code (such as the representation of the structural features of the key parts of the tower in the key point quantitative knowledge code), while lower weights are assigned to some relatively minor features. For example, the integrated knowledge code is (I = (i_1, i_2, cdots, i_p)), and the weight vector is (W = (w_1, w_2, cdots, w_p)), then the multi-mode inspection knowledge code is (M = (i_1w_1, i_2w_2, cdots, i_pw_p)).

[0109] 3.3. The incentive unit generates the target inspection point cloud element knowledge

[0110] Finally, the X multi-mode inspection knowledge encodings are used as the input information of the excitation unit, and the X target inspection point cloud element knowledge is generated through the excitation unit. The excitation unit can use an activation function, such as the ReLU (RectifiedLinearUnit) function. For example, for each element (m_i) in the multi-mode inspection knowledge encoding, after being processed by the ReLU function, (y_i = max(0, m_i)) is obtained, and the final (X) (y_i) constitute the X target inspection point cloud element knowledge.

[0111] With this design, first, in the key point quantification recognition module, through the unique operations of the vector space conversion unit and the knowledge collision unit, the features in the key point detection information can be more comprehensively and deeply mined, and accurate key point quantification knowledge encoding can be generated. For example, by processing the semantic graph describing the local point cloud through operations such as summation and product, the structural relationship of each component in the distribution network overhead line equipment can be accurately captured. In the attitude point cloud recognition module, the application of the residual connection unit and the attitude fusion algorithm enables the flight status monitoring information to be effectively integrated and generate accurate attitude point cloud knowledge encoding. For example, the residual connection unit uses the idea of ​​the residual network to retain more original information, and the attitude fusion algorithm uses the EKF algorithm to improve the accuracy of the attitude data. The collaborative work of the data weighting unit and the excitation unit in the fully connected module enables the integrated knowledge to be adjusted according to the importance of the features. The target inspection point cloud element knowledge finally generated can more accurately reflect the status of the distribution network overhead line equipment and the flight status of the drone, providing a reliable data basis for subsequent inspection decisions and improving the accuracy and reliability of the entire inspection system.

[0112] On the basis of the above content, the method also includes the following training process: obtaining a laser point cloud inspection information set and an initial inspection route record of the drone sample, Y cooperative drones of the drone sample, and a conflicting drone of the drone sample, wherein the laser point cloud inspection information set includes key point detection information and flight status monitoring information, the key point detection information is used to characterize the information obtained after the laser point cloud scanning and collection of the distribution network overhead line equipment of the drone sample within a specified inspection task cycle, the flight status monitoring information is used to characterize the monitoring results corresponding to the posture changes of the drone sample in the specified inspection task cycle, and the initial inspection route record is used to characterize the line intersection information of the drone sample and the cooperative drone; determining Z cooperative drones from the Y cooperative drones, and generating a first inspection route record of the drone sample based on the Z cooperative drones and the initial inspection route record of the drone sample. Record samples; simulate the route intersection information of the drone sample and the conflicting drone, and generate a second inspection route record sample for the drone sample; use the laser point cloud inspection information set as the input information of the point cloud element recognition branch, and generate an inspection point cloud element knowledge sample through the point cloud element recognition branch; use the inspection point cloud element knowledge sample, the initial inspection route record, the first inspection route record sample and the second inspection route record sample as the input information of the route optimization decision branch, and generate a first route planning suggestion feature sample, an initial route planning suggestion feature sample and a second route planning suggestion feature sample through the route optimization decision branch; generate a training evaluation function based on the first route planning suggestion feature sample, the initial route planning suggestion feature sample and the second route planning suggestion feature sample; update the neuron weights of the point cloud element recognition branch and the route optimization decision branch through the training evaluation function.

[0113] The training process in the above technical solution aims to optimize the point cloud feature recognition branch and the line optimization decision branch to improve the efficiency and accuracy of drones in the inspection of distribution network overhead lines. The entire training process involves the acquisition, processing and sample generation of multiple data, and finally the neuron weights of the relevant branches are updated through the training evaluation function.

[0114] 1. Acquisition of training data

[0115] First, the laser point cloud inspection information set and initial inspection route record of the drone sample are obtained. The laser point cloud inspection information set contains key point detection information and flight status monitoring information. The key point detection information is the information obtained after the drone sample performs laser point cloud scanning and collection of the distribution network overhead line equipment within the specified inspection task cycle. For example, for the laser point cloud scanning of the tower body, the coordinates, shape and other data of the tower may be obtained as key point detection information. The flight status monitoring information characterizes the monitoring results corresponding to the posture changes of the drone sample in the specified inspection task cycle, such as the changes in the pitch angle, roll angle and yaw angle of the drone. At the same time, the Y cooperative drones of the drone sample and the conflicting drones of the drone sample are obtained, and the initial inspection route record is used to characterize the line intersection information between the drone sample and the cooperative drone.

[0116] 2. Generation of inspection route record samples

[0117] 2.1. Generation of the first inspection route record sample

[0118] Determine Z cooperative drones from Y cooperative drones (Z is greater than or equal to 1 and less than or equal to Y). Based on the initial inspection route records of the Z cooperative drones and the drone sample, generate the first inspection route record sample of the drone sample. For example, if Y=5, Z=3, select three specific cooperative drones. For example, the initial inspection route record records the route intersections of the drone sample and the five cooperative drones in the form of coordinates, such as the drone sample intersecting with the route of cooperative drone 1 at the coordinates ((x_1, y_1, z_1)), and intersecting with the route of cooperative drone 2 at ((x_2, y_2, z_2)), etc. For the three selected cooperative drones, extract the corresponding intersection coordinate information, organize this information in a certain format, and generate the first inspection route record sample. This sample records in detail the route intersections of the drone sample and the specific cooperative drone, such as the frequency of intersections, the relative speed at the time of intersection, and other information.

[0119] 2.2. Generation of the second inspection route record sample

[0120] By simulating the intersection information of the routes of the drone sample and the conflicting drone, a second inspection route record sample of the drone sample is generated. For example, the conflicting drone may conflict due to accidentally entering the inspection area of ​​the drone sample. For example, the route of the conflicting drone is a straight line (L_1), and the route of the drone sample is a straight line (L_2). The coordinates of their intersection ((x_3, y_3, z_3)) and the angle of intersection (theta) (which can be obtained by vector calculation, such as (costheta = frac{vec{v_1}cdotvec{v_2}}{|vec{v_1}|*|vec{v_2}|}), where (vec{v_1}) and (vec{v_2}) are the direction vectors of the two routes respectively) and other information are simulated and calculated to form the second inspection route record sample.

[0121] 3. Generation of inspection point cloud element knowledge samples

[0122] The acquired laser point cloud inspection information set is used as the input information of the point cloud feature recognition branch. The point cloud feature recognition branch includes the key point quantization recognition module, attitude point cloud recognition module and fully connected module as described above. According to the previous processing flow, the key point detection information is passed through the vector space conversion unit of the key point quantization recognition module (such as summing the corresponding feature members in the semantic graph describing the local point cloud) and the knowledge collision unit (such as product and sum of the corresponding feature members) to generate the key point quantization knowledge encoding; the flight status monitoring information is passed through the residual connection unit of the attitude point cloud recognition module (such as using the residual network idea to process the flight status monitoring information to generate the attitude monitoring data group) and the point cloud reconstruction unit (such as using the point cloud generation algorithm based on deep learning to process the attitude monitoring event set) to generate the attitude point cloud knowledge encoding; and then through the integration of the fully connected module, data weighting and excitation operations, the inspection point cloud feature knowledge sample is finally generated.

[0123] 4. Generation of feature samples for route planning suggestions

[0124] The inspection point cloud element knowledge sample, the initial inspection route record, the first inspection route record sample and the second inspection route record sample are used as the input information of the line optimization decision branch. The line optimization decision branch may be processed by genetic algorithms and other methods. For example, these input information are encoded, such as the degree of damage to the tower in the inspection point cloud element knowledge sample (represented by a value between 0 and 1, 0 means no damage, 1 means severe damage), the working state of the transformer (represented by a value between -1 and 1, -1 means fault, 1 means normal operation), and the flight altitude (such as 100 meters) and speed (such as 5 meters / second) in the inspection route record are encoded into binary strings. Then, the quality of each individual (the encoded route plan) is evaluated according to the set fitness function (such as (F = a*S+b*E+c*C), where (F) is the fitness value, (S) is the flight safety index (the value range is 0-1, 1 means the safest), (E) is the inspection efficiency index (for example, the number of equipment inspected per unit time), (C) is the coverage of the distribution network overhead line equipment (the value range is 0-1, 1 means full coverage), (a), (b), (c) are weight coefficients). Through selection, crossover and mutation operations, the population is continuously evolved, and finally the first route planning suggestion feature sample (corresponding to the route planning related to the first inspection route record sample), the initial route planning suggestion feature sample (corresponding to the route planning related to the initial inspection route record) and the second route planning suggestion feature sample (corresponding to the route planning related to the second inspection route record sample) are generated.

[0125] 5. Generation of training evaluation function and updating of neuron weights

[0126] 5.1. Generation of training evaluation function

[0127] A training evaluation function is generated based on the first route planning suggestion feature sample, the initial route planning suggestion feature sample and the second route planning suggestion feature sample. This training evaluation function can comprehensively consider multiple factors, for example, calculating the differences in the accuracy, efficiency and other indicators of the route planning of each sample. For example, the accuracy index (A) is calculated by comparing the deviation between the planned route and the actual optimal route (such as the average value of the deviation distance), and the efficiency index (E) is calculated based on the number of devices inspected per unit time or the flight distance. The training evaluation function can be (T = d*(A_1-A_0) + e*(E_1-E_0)), where (T) is the training evaluation value, (d) and (e) are weight coefficients, (A_1) and (A_0) are the accuracy indicators of the newly generated sample and the initial sample, respectively, and (E_1) and (E_0) are the efficiency indicators of the newly generated sample and the initial sample, respectively.

[0128] 5.2. Neuron Weight Update

[0129] The neuron weights of the point cloud feature recognition branch and the route optimization decision branch are updated through the training evaluation function. For example, in the neural network, the gradient of each neuron is calculated using the back propagation algorithm according to the value of the training evaluation function. For the neurons in the key point quantization recognition module, the posture point cloud recognition module and the fully connected module in the point cloud feature recognition branch, and the related neurons in the route optimization decision branch, the weights of the neurons are updated according to the calculated gradients and a certain learning rate (such as (alpha = 0.01)):

[0130] (w_{new}=w_{old}-alpha*frac{partialT}{partialw}), where (w_{new}) is the updated weight and (w_{old}) is the original weight.

[0131] With this design, through a detailed training process, a variety of inspection route record samples and inspection point cloud element knowledge samples are used to continuously optimize the point cloud element recognition branch and the line optimization decision branch. When generating inspection route record samples, the situations of cooperative drones and conflicting drones are accurately considered, such as the calculation of specific line intersection information, which provides a comprehensive reference for subsequent route planning. In the process of generating inspection point cloud element knowledge samples, the fine processing of each module can effectively extract equipment and flight status information. The line optimization decision branch generates route planning suggestion feature samples based on multiple samples, and evaluates them through a comprehensive training evaluation function, which can accurately adjust the neuron weights, improving the accuracy and efficiency of the entire drone inspection system for distribution network overhead line equipment inspections, as well as its adaptability to various complex situations (such as the presence of cooperative and conflicting drones).

[0132] In an exemplary technical solution, the inspection point cloud element knowledge sample, the initial inspection route record, the first inspection route record sample and the second inspection route record sample are used as the incoming information of the route optimization decision branch, and the first route planning suggestion feature sample, the initial route planning suggestion feature sample and the second route planning suggestion feature sample are generated through the route optimization decision branch, including: using the inspection point cloud element knowledge sample and the initial inspection route record as the incoming information of the route optimization decision branch, and generating the initial route planning suggestion feature sample through the route optimization decision branch; using the inspection point cloud element knowledge sample and the first inspection route record sample as the incoming information of the route optimization decision branch, and generating the first route planning suggestion feature sample through the route optimization decision branch; using the inspection point cloud element knowledge sample and the second inspection route record sample as the incoming information of the route optimization decision branch, and generating the second route planning suggestion feature sample through the route optimization decision branch.

[0133] In the present invention, the technical solution focuses on how the line optimization decision branch generates specific route planning suggestion feature samples based on different input information, including initial route planning suggestion feature samples, first route planning suggestion feature samples and second route planning suggestion feature samples. This process is the key link in optimizing route planning in the entire UAV inspection system. By rationally processing different input information combinations, it provides an accurate basis for the route planning of UAVs in the distribution network overhead line inspection.

[0134] 1. Generate initial route planning suggestion feature samples

[0135] 1.1 Input and processing basis of line optimization decision branch

[0136] First, the inspection point cloud element knowledge sample and the initial inspection route record are used as the input information of the line optimization decision branch. The inspection point cloud element knowledge sample contains the comprehensive knowledge about the distribution network overhead line equipment and the flight status of the drone obtained by the point cloud element recognition branch. For example, the inspection point cloud element knowledge sample may contain the quantitative value of the damage degree of the tower (for example, the value range is 0-1, 0 means no damage, 1 means serious damage), and the quantitative value of the attitude stability of the drone during flight (for example, the value range is -1 to 1, -1 means very unstable, 1 means very stable). The initial inspection route record records the line intersection information of the drone sample and the cooperative drone in detail, such as the coordinates of the intersection point (for example, ((x_0, y_0, z_0))), the relative height difference at the intersection (for example, (h_0) meters), etc.

[0137] 1.2 Internal processing mechanism of line optimization decision branch

[0138] The route optimization decision branch may be processed using optimization algorithms such as genetic algorithms. During the processing, the input inspection point cloud element knowledge sample and the initial inspection route record are first encoded. For example, the tower damage degree 0.3, the drone attitude stability 0.5 in the inspection point cloud element knowledge sample, and the intersection coordinates ((10, 20, 30)) and the relative height difference 5 meters in the initial inspection route record are encoded into a binary string. For example, the encoded inspection point cloud element knowledge sample is (P = [010101]) (for example only), and the initial inspection route record is (R = [110010]).

[0139] Then, the quality of each individual (encoded route plan) is evaluated according to the set fitness function. The fitness function may take into account multiple factors, such as (F = a*S + b*E + c*C). Among them, (S) is the flight safety index, for example, it can be calculated based on the situation of the intersection and the stability of the drone's attitude, such as (S = 0.8) (the value range is 0-1, 1 means the safest); (E) is the inspection efficiency index, such as the distance from the current position of the drone to the next inspection point and the expected flight speed, such as (E = 0.6) (for example, the number of equipment inspected per unit time and other factors); (C) is the coverage of the distribution network overhead line equipment, for example, according to the information such as the degree of damage to the tower in the inspection point cloud element knowledge sample, the coverage of the equipment on the current route is judged, (C = 0.7) (the value range is 0-1, 1 means full coverage); (a = 0.3), (b = 0.4), (c = 0.3) are weight coefficients.

[0140] The population is continuously evolved through selection, crossover and mutation operations. For example, in the selection operation, individuals with higher fitness are selected according to the fitness function value to enter the next generation of population. For example, if the population size is (N=100), the individuals ranked in the top (50) in fitness are selected. In the crossover operation, a single-point crossover is used, and a crossover point is randomly selected to exchange the parts of the two individuals after the crossover point. For example, for individual (A=[010101]) and individual (B=[110010]), for example, the crossover point is the third position, and after crossover, (A'=[010010]) and (B'=[110101]) are obtained. The mutation operation changes a certain position in the individual with a certain probability (for example, (p=0.01)). After multiple generations of evolution (for example, (G=100) generations), the initial route planning recommendation feature sample is finally obtained. This sample may include information such as new flight altitude recommendations (for example, from the original 100 meters to 90 meters), new flight speed recommendations (from 5 meters / second to 6 meters / second), and a new inspection point sequence.

[0141] 2. Generate the first route planning suggestion feature sample

[0142] 2.1 Input information and initial processing

[0143] Next, the inspection point cloud element knowledge sample and the first inspection route record sample are used as the input information of the route optimization decision branch. The first inspection route record sample is generated based on the initial inspection route record of the Z cooperative drones determined from the Y cooperative drones and the drone sample, which contains more detailed route intersection information between the drone sample and the specific cooperative drone. For example, the first inspection route record sample may contain information such as the intersection frequency with the specific cooperative drone (for example, once every 10 minutes) and the relative speed when crossing (for example, 2 meters / second).

[0144] 2.2. Processing of route optimization decision branches

[0145] Similarly, the inspection point cloud feature knowledge sample and the first inspection route record sample are first encoded. For example, the inspection point cloud feature knowledge sample is encoded as (P_1 = [101010]), and the first inspection route record sample is encoded as (R_1 = [011011]). Then the pros and cons of each individual are evaluated according to the fitness function. The fitness function here is similar to that when generating the initial route planning suggestion feature sample, but the weight may be adjusted or new considerations may be added based on the specific information in the first inspection route record sample. For example, since the first inspection route record sample contains crossover frequency and relative speed information, the fitness function may be adjusted to (F_1=a_1*S_1+b_1*E_1+c_1*C_1+d_1*F_q+e_1*V_r), where (F_q) is the quantized value of the crossover frequency (for example, calculated based on the deviation between the actual crossover frequency and the ideal crossover frequency), (V_r) is the quantized value of the relative speed (for example, calculated based on the deviation between the relative speed and the safe relative speed), and (a_1=0.2), (b_1=0.3), (c_1=0.2), (d_1=0.2), and (e_1=0.1) are weight coefficients.

[0146] Evolve the population according to the selection, crossover and mutation operations of the genetic algorithm. For example, select the top (60) individuals in fitness ranking (the population size is still (N=100)), use two-point crossover (randomly select two intersection points and exchange the part between the two points), the mutation probability is (p=0.02), and after (G=150) generations of evolution, the first route planning suggestion feature sample is obtained. This sample may contain route adjustments for specific cooperative drones, such as adjusting the flight path to avoid too frequent intersections (adjusting the original straight path that may cause frequent intersections to a slightly curved path), or adjusting the flight speed according to the relative speed (if the relative speed is too fast, reduce its own flight speed to ensure safety) and other information.

[0147] 3. Generate the second route planning suggestion feature sample

[0148] 3.1 Input information and its characteristics

[0149] Finally, the inspection point cloud element knowledge sample and the second inspection route record sample are used as the input information of the route optimization decision branch. The second inspection route record sample is obtained by simulating the route intersection information of the drone sample and the conflicting drone, which contains key information such as the intersection angle (for example, (theta = 30^{circ})) and the distance at the time of intersection (for example, (d = 5) meters).

[0150] 3.2. Processing logic of route optimization decision branch

[0151] Encode the inspection point cloud element knowledge sample and the second inspection route record sample. Suppose the inspection point cloud element knowledge sample is encoded as (P_2 = [010111]) and the second inspection route record sample is encoded as (R_2 = [100101]). The fitness function will focus on conflict-related factors, such as (F_2 = a_2*S_2+b_2*E_2+c_2*C_2+f_2*theta_d+g_2*d_d), where (theta_d) is the quantized value of the intersection angle (for example, calculated based on the deviation between the intersection angle and the safe intersection angle, the safe intersection angle is, for example, (45^{circ}), then:

[0152] (theta_d=frac{|theta-45^{circ}|}{45^{circ}})), (d_d) is the quantized value of the crossing distance (for example, calculated based on the deviation between the crossing distance and the safety distance, the safety distance is, for example, 10 meters, then (d_d=frac{|d-10|}{10})), (a_2=0.3), (b_2=0.2), (c_2=0.2), (f_2=0.2), (g_2=0.1) are weight coefficients.

[0153] The population is evolved through selection, crossover and mutation operations of the genetic algorithm. For example, the top (40) individuals in fitness ranking (population size (N = 100)) are selected, uniform crossover (each individual is exchanged with a certain probability), the mutation probability is (p = 0.03), and after (G = 200) generations of evolution, the second route planning suggestion feature sample is obtained. This sample may include emergency route adjustment for obstacle avoidance, such as a significant change in flight direction (such as changing the (90^{circ}) direction of flight) when the intersection angle is small and the intersection distance is close, or an emergency reduction in flight altitude to avoid collision with conflicting drones.

[0154] In this way, by generating different route planning suggestion feature samples based on different input information combinations, different inspection situations can be responded to more accurately. When generating the initial route planning suggestion feature sample, the equipment status and basic collaborative drone intersection information are comprehensively considered to optimize the route planning under general circumstances, improve the inspection efficiency and equipment coverage. When generating the first route planning suggestion feature sample, based on the detailed intersection information with the specific collaborative drone, the route related to the collaborative drone is further optimized, the intersection risk is reduced, and the flight safety is improved. When generating the second route planning suggestion feature sample, for the situation of conflicting drones, factors such as intersection angle and distance are considered, which effectively avoids potential collision risks and ensures the safe implementation of drone inspection tasks. Overall, this situation-based processing method improves the autonomy, safety and efficiency of drones in the inspection of distribution network overhead lines.

[0155] In an optional design idea A, the relevant processing of the original sample is involved: the initial inspection route record includes the initial UAV inspection route record and the initial distribution network overhead structure spatial distribution, the initial UAV inspection route record includes the initial directional UAV, and the initial directional UAV and the UAV sample correspond to the same inspection task event; the initial distribution network overhead structure spatial distribution includes the key point structure spatial distribution, and the key point structure spatial distribution and the UAV sample correspond to the same distribution network overhead structure area; the line optimization decision branch includes a UAV line intersection mining module, a structure space distribution intersection mining module, an inspection route record combination module and a route planning output module; the inspection point cloud element knowledge sample and the initial inspection route record are used as the input information of the line optimization decision branch, and the initial route planning suggestion feature sample is generated through the line optimization decision branch, including: the initial UAV inspection route record and the inspection point cloud element knowledge sample are used as the input information of the UAV line intersection mining module, and the initial UAV line intersection feature is generated through the UAV line intersection mining module; the initial distribution network overhead structure spatial distribution and the inspection point cloud element knowledge sample are used as the input information of the UAV line intersection mining module, and the initial UAV line intersection feature is generated through the UAV line intersection mining module. The inspection cloud element knowledge sample is used as the input information of the structural space distribution cross mining module, and the initial structural space distribution cross mining module is used to generate the initial structural space distribution cross feature; the initial drone line cross feature and the initial structural space distribution cross feature are used as the input information of the inspection route record combination module, and the inspection route record combination module is used to generate the associated structural space distribution cross heat map of the initial line cross combination information and the associated drone line cross heat map of the initial structural space distribution cross combination information, wherein the associated structural space distribution cross heat map of the initial line cross combination information is obtained based on the initial structural space distribution cross feature and the structural space collaborative distribution relationship network, and the associated drone line cross heat map of the initial structural space distribution cross combination information is obtained based on the initial drone line cross feature and the drone collaborative line relationship network; the associated structural space distribution cross heat map of the initial line cross combination information and the associated drone line cross heat map of the initial structural space distribution cross combination information are used as the input information of the route planning output module, and the route planning output module is used to generate the initial route planning suggestion feature sample.

[0156] In an optional design idea B, related processing of positive samples is involved: the first inspection route record sample includes a first drone inspection route sample and a first distribution network overhead structure spatial distribution sample, the first drone inspection route sample includes a first directional cruise route sample, and the first directional cruise route sample and the drone sample correspond to the same inspection task event; the first distribution network overhead structure spatial distribution sample includes a first key point structure spatial distribution sample, and the first key point structure spatial distribution sample and the drone sample correspond to the same distribution network overhead structure area; the line optimization decision branch includes a drone line intersection mining module, a structure space distribution intersection mining module, an inspection route record combination module and a route planning output module; the inspection point cloud element knowledge sample and the first inspection route record sample are used as the input information of the line optimization decision branch, and the first route planning suggestion feature sample is generated through the line optimization decision branch, including: the first drone inspection route sample and the inspection point cloud element knowledge sample are used as the input information of the drone line intersection mining module, and the first drone line intersection feature sample is generated through the drone line intersection mining module; the first distribution network overhead structure spatial distribution The sample and the inspection point cloud element knowledge sample are used as the input information of the structural space distribution cross mining module, and the first structural space distribution cross feature sample is generated through the structural space distribution cross mining module; the first drone line cross feature sample and the first structural space distribution cross feature sample are used as the input information of the inspection route record combination module, and the first associated distribution cross heat map sample of the first line cross combination information sample and the first associated line cross heat map sample of the first structural distribution involved feature sample are generated through the inspection route record combination module, wherein the first associated distribution cross heat map sample of the first line cross combination information sample is obtained based on the first structural space distribution cross feature sample and the structural space collaborative distribution relationship network, and the first associated line cross heat map sample of the first structural distribution involved feature sample is obtained based on the first drone line cross feature sample and the drone collaborative route relationship network; the first associated distribution cross heat map sample of the first line cross combination information sample and the first associated line cross heat map sample of the first structural distribution involved feature sample are used as the input information of the route planning output module, and the first route planning suggestion feature sample is generated through the route planning output module.

[0157] In an optional design idea C, related processing of negative samples is involved: the second inspection route record sample includes a second UAV inspection route sample and a second distribution network overhead structure spatial distribution sample, the second UAV inspection route sample includes a second directional cruise route sample, and the second directional cruise route sample and the UAV sample correspond to the same inspection task event; the second distribution network overhead structure spatial distribution sample includes a second key point structure spatial distribution sample, and the second key point structure spatial distribution sample and the UAV sample correspond to the same distribution network overhead structure area; the line optimization decision branch includes a UAV line intersection mining module, a structure space distribution intersection mining module, an inspection route record combination module and a route planning output module; the inspection point cloud element knowledge sample and the second inspection route record sample are used as the input information of the line optimization decision branch, and the second route planning suggestion feature sample is generated through the line optimization decision branch, including: the second UAV inspection route sample and the inspection point cloud element knowledge sample are used as the input information of the UAV line intersection mining module, and the second UAV line intersection feature sample is generated through the UAV line intersection mining module; the second distribution network overhead structure spatial distribution The sample and the inspection point cloud element knowledge sample are used as the input information of the structural space distribution cross mining module, and the second structural space distribution cross feature sample is generated through the structural space distribution cross mining module; the second drone line cross feature sample and the second structural space distribution cross feature sample are used as the input information of the inspection route record combination module, and the second associated distribution cross heat map sample of the second line cross combination information sample and the second associated line cross heat map sample of the second structural distribution involved feature sample are generated through the inspection route record combination module, wherein the second associated distribution cross heat map sample of the second line cross combination information sample is obtained based on the second structural space distribution cross feature sample and the structural space collaborative distribution relationship network, and the second associated line cross heat map sample of the second structural distribution involved feature sample is obtained based on the second drone line cross feature sample and the drone collaborative route relationship network; the second associated distribution cross heat map sample of the second line cross combination information sample and the second associated line cross heat map sample of the second structural distribution involved feature sample are used as the input information of the route planning output module, and the second route planning suggestion feature sample is generated through the route planning output module.

[0158] In the present invention, the above three design ideas revolve around the operation flow of the route optimization decision branch when processing different types of samples (original samples, positive samples, negative samples) to generate different route planning suggestion feature samples (initial route planning suggestion feature samples, first route planning suggestion feature samples, second route planning suggestion feature samples). The route optimization decision branch includes a drone route cross mining module, a structural space distribution cross mining module, an inspection route record combination module, and a route planning output module. Different types of samples include different inspection route records and distribution network overhead structure space distribution related content. The generation of route planning suggestion feature samples is achieved through the collaborative processing of each module.

[0159] Design idea A-Processing the original sample to generate the initial route planning suggestion feature sample

[0160] 1. Analysis of the content of the initial inspection route records

[0161] 1.1. Initial drone inspection route record

[0162] The initial drone inspection route record in the initial inspection route record contains the initial directional drone, and the initial directional drone and the drone sample correspond to the same inspection task event. This means that in the same inspection task, there is a certain correlation between the route of the initial directional drone and the route of the drone sample, which is of great significance for subsequent operations such as line intersection excavation.

[0163] 1.2. Spatial distribution of the initial distribution network overhead structure

[0164] The initial distribution network overhead structure spatial distribution includes the key point structure spatial distribution, and the key point structure spatial distribution and the drone sample correspond to the same distribution network overhead structure area. For example, in a specific distribution network overhead line area, the key point structure spatial distribution may include the location coordinates of the tower, the spatial distribution of the cross arm and other information, which is a spatial description of the distribution network overhead structure in the area, providing basic data for subsequent cross-mining of the structure spatial distribution.

[0165] 2. Processing flow of each module

[0166] 2.1. UAV route intersection mining module generates initial UAV route intersection features

[0167] The initial drone inspection route record and the inspection point cloud element knowledge sample are used as the input information of the drone line intersection mining module. The inspection point cloud element knowledge sample contains comprehensive knowledge of the distribution network overhead line equipment and the flight status of the drone. For example, the inspection point cloud element knowledge sample contains the current position coordinates of the drone ((x_1, y_1, z_1)), the flight direction vector (vec{v}=(v_x, v_y, v_z)) and the position coordinates of the distribution network overhead line equipment (such as poles and towers) ((x_t, y_t, z_t)) and other information. The initial drone inspection route record may be represented in the form of a series of waypoint coordinates ((x_{p1}, y_{p1}, z_{p1})), ((x_{p2}, y_{p2}, z_{p2})) and so on.

[0168] The drone route intersection mining module may determine whether there is an intersection by calculating the distance formula (d = frac{|vec{AB}*vec{AC}|}{|vec{AB}|}) between two routes (drone sample route and initial directional drone route) (where (vec{AB}) and (vec{AC}) are vectors constructed based on route coordinates). If the calculated distance (d) is less than a certain threshold (for example, (d<5) meters, where 5 meters is a safety distance threshold set based on actual needs and experience), it is considered that there is a route intersection. By calculating and analyzing the entire route, the initial drone route intersection feature is generated. This feature may be a vector containing information such as the coordinates of the intersection, the possibility of intersection (expressed in probability form, such as 0.8 means there is an 80% probability of intersection), etc.

[0169] 2.2. Structural space distribution cross mining module generates initial structural space distribution cross features

[0170] The initial distribution network overhead structure spatial distribution and the inspection point cloud element knowledge sample are used as the input information of the structure spatial distribution cross mining module. The key point structure spatial distribution information in the initial distribution network overhead structure spatial distribution (such as the coordinates and size of the tower) is combined with the drone position information in the inspection point cloud element knowledge sample. For example, a spatial geometry algorithm is used to calculate the spatial relationship between the drone and the tower. If a coordinate system is established with the tower as the center, the relative position coordinates of the drone in this coordinate system ((x_r, y_r, z_r)) are calculated.

[0171] According to the relative position relationship between the drone and the key points of each distribution network overhead structure, the initial structure spatial distribution cross feature is generated. This feature may include the closest distance between the drone and the key structure (such as the distance from the drone to the nearest tower is (d_{min}=10) meters), the azimuth of the drone relative to the key structure (such as the azimuth relative to a tower is (theta=45^{circ})) and other information.

[0172] 2.3. Inspection route record combination module generates heat map

[0173] The initial UAV line intersection characteristics and the initial structural space distribution intersection characteristics are used as the input information of the inspection route record combination module. For example, the structural space collaborative distribution relationship network is a graph structure with the key points of the distribution network overhead structure as nodes and the spatial relationship between them (such as distance, azimuth, etc.) as the edge weight. Based on the initial structural space distribution intersection characteristics and the structural space collaborative distribution relationship network, the degree of association between each node (key point) and the line intersection is calculated through a certain algorithm (such as an algorithm based on a graph convolutional neural network) to generate a correlation structural space distribution intersection heat map of the initial line intersection combination information. This heat map is represented in the form of a matrix. Each element in the matrix represents the correlation strength between a key point and the line intersection. The value range can be 0-1. For example, the element value corresponding to a certain tower is 0.6, indicating that the correlation strength between the tower and the line intersection is high.

[0174] At the same time, based on the initial UAV route intersection characteristics and the UAV collaborative route relationship network (which may be a graph structure with UAV routes as nodes and the intersection relationship between routes as edge weights), a similar algorithm is used to calculate the degree of correlation between each UAV route node and the intersection of the structural spatial distribution, and generate a UAV route intersection heat map of the initial structural spatial distribution intersection combination information. For example, the element value corresponding to a UAV route is 0.7, indicating that this route has a strong correlation with the intersection of the structural spatial distribution.

[0175] 2.4. Route planning output module generates initial route planning suggestion feature samples

[0176] The cross heat map of the associated structure space distribution of the initial route intersection combination information and the cross heat map of the associated drone route of the initial structure space distribution cross combination information are used as the input information of the route planning output module. The route planning output module may use a path planning algorithm (such as the A* algorithm) to plan the route based on the information in the heat map. For example, the A* algorithm calculates an optimal path based on the starting point (the current position of the drone), the target point (the next inspection point), and the obstacle information in the heat map (such as areas with high cross-correlation intensity with the route are considered obstacles). This optimal path may include new waypoint coordinates, flight altitude adjustment suggestions (such as adjusting from the current 100 meters to 90 meters), flight speed adjustment suggestions (such as adjusting from 5 meters / second to 6 meters / second), etc., thereby generating an initial route planning suggestion feature sample.

[0177] Design idea B-positive sample processing to generate the first route planning suggestion feature sample

[0178] 1. Analysis of the sample content of the first inspection route record

[0179] 1.1. Example of the first drone inspection route

[0180] The first drone inspection route sample in the first inspection route record sample contains the first directional cruise route sample, and the first directional cruise route sample and the drone sample correspond to the same inspection task event. This is similar to the initial drone inspection route record, but it is for the positive sample case, and its route information is important for analyzing the route crossing and other situations in specific situations.

[0181] 1.2. Example of spatial distribution of the first distribution network overhead structure

[0182] The first distribution network overhead structure spatial distribution sample includes the first key point structure spatial distribution sample, which corresponds to the same distribution network overhead structure area as the drone sample. This part is similar to the initial distribution network overhead structure spatial distribution, which is a description of the distribution network overhead structure spatial information, but for the positive sample, there may be some differences in the data, such as different tower layouts or equipment status.

[0183] 2. Processing flow of each module

[0184] 2.1. The UAV route intersection mining module generates the first UAV route intersection feature sample

[0185] The first UAV inspection route sample and the inspection point cloud element knowledge sample are used as input information for the UAV route intersection mining module. For example, the inspection point cloud element knowledge sample contains information such as the attitude angle of the UAV (pitch angle (alpha = 10^{circ}), roll angle (beta = 5^{circ})), and the first UAV inspection route sample is given by a specific route trajectory equation (such as the quadratic curve equation (y = ax^2 + bx + c) to represent the route trajectory).

[0186] The drone route intersection mining module calculates the intersection with other related routes (such as collaborative drone routes) by analyzing the route trajectory equation and the drone's attitude angle and other information. For example, the trajectory equations of the two routes are solved together to obtain the possible intersection coordinates ((x_c, y_c)). At the same time, consider the impact of the drone's attitude angle on the intersection, such as the attitude angle may affect the drone's flight radius. Through these calculations and analyses, the first drone route intersection feature sample is generated. This sample may contain information such as the coordinates of the intersection point, the attitude angle information at the time of intersection, and the predicted time of intersection (such as expected intersection after 10 minutes).

[0187] 2.2. The structural space distribution cross mining module generates the first structural space distribution cross feature sample

[0188] The first distribution network overhead structure spatial distribution sample and the inspection point cloud element knowledge sample are used as the input information of the structure spatial distribution cross mining module. The first key point structure spatial distribution sample in the first distribution network overhead structure spatial distribution sample may contain more detailed equipment status information, such as the inclination of the tower (for example, (theta_t = 3^{circ})), the deformation of the cross arm (for example, (Deltal = 0.1) meters), etc.

[0189] Combined with the UAV flight status information in the inspection point cloud element knowledge sample (such as a flight altitude of 100 meters), the intersection between the UAV and the overhead structure of the distribution network is calculated through a spatial analysis algorithm (such as a three-dimensional spatial coordinate transformation algorithm). For example, the collision risk of the UAV with the tilted tower under different flight postures is calculated. Based on these calculation results, a first structural spatial distribution intersection feature sample is generated. This sample may include information such as the collision risk probability of the UAV with different equipment (such as a collision risk probability of 0.1 with a certain tower), and the safety distance margin from the key structure (such as a safety distance margin of 2 meters from a certain cross arm).

[0190] 2.3. Inspection route record combination module generates heat map

[0191] The first UAV route cross feature sample and the first structural space distribution cross feature sample are used as the input information of the inspection route record combination module. The structural space collaborative distribution relationship network and the UAV collaborative route relationship network may have different weights or connection relationships in the case of positive samples, reflecting the unique situation of the positive samples.

[0192] Through the algorithm based on the graph neural network, according to the first structure space distribution cross feature sample and the structure space collaborative distribution relationship network, the first correlation distribution cross heat map sample of the first line cross combination information sample is calculated and generated. The element value in this heat map represents the degree of correlation between different key points and line intersections, which may be different from the original sample situation. For example, the element value corresponding to a certain device is 0.4, which is lower than the value in the original sample, indicating that the degree of correlation between the device and the line intersection is low in the positive sample case.

[0193] At the same time, based on the first drone route intersection feature sample and the drone collaborative route relationship network, a first associated route intersection heat map sample of the first structural distribution involving feature samples is generated. The element value in this heat map indicates the degree of association between different routes and the intersection of the structural spatial distribution. For example, the element value corresponding to a certain route is 0.5, which reflects the association of this route with the intersection of the structural spatial distribution in the case of a positive sample.

[0194] 2.4. The route planning output module generates the first route planning suggestion feature sample

[0195] The first associated distribution cross heat map sample of the first route cross combination information sample and the first associated route cross heat map sample of the first structural distribution involved feature sample are used as input information of the route planning output module. The route planning output module adopts an improved path planning algorithm (such as an improved version of the Dijkstra algorithm) to consider the information in the heat map and some constraints unique to the positive sample (such as key inspection requirements for certain equipment) to perform route planning. For example, under the premise of meeting the key inspection requirements, an optimized route is calculated based on the obstacle information and path cost in the heat map. This route may include information such as a new inspection order (inspect a key device first), an adjusted flight speed (such as adjusting the speed according to the distance from the equipment), etc., thereby generating a first route planning recommendation feature sample.

[0196] Design idea C-Negative sample processing to generate the second route planning suggestion feature sample

[0197] 1. Analysis of the sample content of the second inspection route record

[0198] 1.1. Second drone inspection route example

[0199] The second drone inspection route sample in the second inspection route record sample contains the second directional cruise route sample, which corresponds to the same inspection task event as the drone sample. In the case of negative samples, this route may have potential conflicts with other factors (such as conflicting drones), and its route information is an important basis for analyzing such conflicts.

[0200] 1.2. Example of spatial distribution of the second distribution network overhead structure

[0201] The second distribution network overhead structure spatial distribution sample includes the second key point structure spatial distribution sample, which corresponds to the same distribution network overhead structure area as the drone sample. Similar to the positive sample and the original sample, in the case of the negative sample, the spatial relationship of the distribution network overhead structure may have different considerations when analyzing the route planning due to factors such as conflicts.

[0202] 2. Processing flow of each module

[0203] 2.1. The drone route intersection mining module generates the second drone route intersection feature sample

[0204] The second UAV inspection route sample and the inspection point cloud element knowledge sample are used as the input information of the UAV route intersection mining module. For example, the inspection point cloud element knowledge sample contains the sensor status information of the UAV (such as a sensor failure, the fault code is F001), and the second UAV inspection route sample is given in the form of discrete waypoints and flight direction vectors, such as waypoints ((x_{n1}, y_{n1}, z_{n1})), ((x_{n2}, y_{n2}, z_{n2})) and flight direction vector (vec{v_n}=(v_{nx}, v_{ny}, v_{nz})).

[0205] The drone route intersection mining module analyzes the intersection of the second drone inspection route sample and other related routes (such as conflicting drone routes) under the consideration of sensor failure. For example, due to sensor failure, there may be a certain error in the positioning of the drone, such as a positioning error range of (pm5) meters. This error range needs to be considered when calculating the intersection with the conflicting drone route. Through complex calculations (such as Monte Carlo simulation method, simulating multiple possible situations within the positioning error range), the second drone route intersection feature sample is generated. This sample may contain information such as the probability of intersection under the consideration of positioning error (such as 0.3, indicating a 30% probability of intersection), the possible area of ​​intersection (represented by a coordinate range, such as ((x_{r1}, y_{r1}, z_{r1})) to ((x_{r2}, y_{r2}, z_{r2}))) and so on.

[0206] 2.2. The structural space distribution cross mining module generates the second structural space distribution cross feature samples

[0207] The second distribution network overhead structure spatial distribution sample and the inspection point cloud element knowledge sample are used as the input information of the structure spatial distribution cross-mining module. The second key point structure spatial distribution sample in the second distribution network overhead structure spatial distribution sample may contain some special cases, such as the dynamic position change of the tower due to external interference (such as strong wind causing the tower to shake) (for example, the tower shakes within a certain range, the shaking center coordinates are ((x_{s}, y_{s}, z_{s})), and the shaking radius is (r=1) meters).

[0208] Combined with the UAV flight status information in the inspection point cloud element knowledge sample (such as unstable flight speed, speed fluctuation range of (pm1) meters / second), the intersection between the UAV and the overhead structure of the distribution network is calculated through a dynamic spatial analysis algorithm (such as a spatial coordinate transformation algorithm considering time series). For example, calculate the collision risk between the UAV and the shaking pole tower at different times. Based on these calculation results, generate a second structural spatial distribution intersection feature sample, which may include the collision risk probability of the UAV with different equipment under dynamic conditions (such as the collision risk probability with a shaking pole tower is 0.2), the dynamic safety distance margin with the key structure (such as the dynamic safety distance margin from a cross arm at a certain moment is 1.5 meters) and other information.

[0209] 2.3. Inspection route record combination module generates heat map

[0210] The second drone route cross feature sample and the second structure space distribution cross feature sample are used as the input information of the inspection route record combination module. In the case of negative samples, the structure space collaborative distribution relationship network and the drone collaborative route relationship network will be adjusted according to the special circumstances of the negative samples, for example, higher weights will be given to routes and equipment with conflict risks.

[0211] Through the algorithm based on dynamic graph neural network (capable of processing dynamically changing node and edge relationships), the second correlation distribution cross heat map sample of the second line intersection combination information sample is calculated based on the second structure space distribution cross feature sample and the structure space collaborative distribution relationship network. The element value in this heat map indicates the degree of correlation between different key points and line intersection in the case of negative samples. For example, the element value corresponding to a tower affected by strong wind is 0.7, which is higher than the value in normal conditions, indicating that its correlation with line intersection is higher in the case of negative samples.

[0212] At the same time, based on the second drone route intersection feature sample and the drone collaborative route relationship network, a second associated route intersection heat map sample of the second structural distribution involving feature samples is generated. The element value in this heat map indicates the degree of correlation between different routes and the intersection of the structural spatial distribution in the case of negative samples. For example, the element value corresponding to a route with conflict risk is 0.8, which reflects that this route has a strong correlation with the intersection of the structural spatial distribution in the case of negative samples.

[0213] 2.4. Route planning output module generates the second route planning suggestion feature sample

[0214] The second associated distribution cross heat map sample of the second route intersection combination information sample and the second associated route distribution cross heat map sample of the second structural distribution involvement feature sample are used as input information of the route planning output module. The route planning output module adopts an emergency route planning algorithm (such as a priority-based path planning algorithm) to give priority to avoiding conflicts and ensuring safety. For example, based on the information in the heat map, areas with high conflict risks are marked as high-priority avoidance areas, and a route that can quickly avoid these areas is calculated. This route may include information such as emergency turning instructions (such as turning left (90^{circ})), emergency reduction of flight altitude (such as emergency reduction from 100 meters to 80 meters), etc., thereby generating a second route planning recommendation feature sample.

[0215] With this design, the drone inspection routes can be planned comprehensively and accurately through different processing methods for original samples, positive samples and negative samples. For the original samples, the relationship between the intersection of drone routes and the intersection of structural space distribution is comprehensively considered in the initial route planning. The generated initial route planning suggestion feature samples can optimize the route under normal inspection conditions and improve the inspection efficiency and safety. When processing positive samples, based on their specific routes and structural space distribution, the generated first route planning suggestion feature samples can adapt to special inspection needs, such as key inspections of specific equipment, while optimizing the relationship with other related factors. When processing negative samples, the generated second route planning suggestion features can take into account various abnormal situations (such as sensor failures, external interference, etc.), thereby improving the safety of cruising.

[0216] In some other embodiments, a specific method for determining a training evaluation function is involved, that is, generating a training evaluation function based on the first route planning suggestion feature sample, the initial route planning suggestion feature sample and the second route planning suggestion feature sample, including: generating a first training error index based on a commonality score between the first route planning suggestion feature sample and the initial route planning suggestion feature sample; generating a second training error index based on a commonality score between the second route planning suggestion feature sample and the initial route planning suggestion feature sample; generating a training evaluation function based on the first training error index and the second training error index.

[0217] It can be understood that this embodiment focuses on how to generate a training evaluation function based on the first route planning suggestion feature sample, the initial route planning suggestion feature sample and the second route planning suggestion feature sample. This process mainly calculates the commonality scores between different samples, and then obtains the training error index, and finally constructs a training evaluation function, which is of key significance for optimizing the entire drone inspection route planning system.

[0218] First, calculate the commonality score of the first route planning suggestion feature sample and the initial route planning suggestion feature sample. The first route planning suggestion feature sample and the initial route planning suggestion feature sample contain a variety of information about the UAV inspection route planning. For example, the first route planning suggestion feature sample may include a new flight altitude suggestion (for example, (h_1 = 90) meters), a new flight speed suggestion (for example, (v_1 = 6) meters / second), a new inspection point sequence (for example, (P_1 = (p_{11}, p_{12}, p_{13}))), etc.; the initial route planning suggestion feature sample may include a flight altitude (h_0 = 100) meters, a flight speed (v_0 = 5) meters / second, an inspection point sequence (P_0 = (p_{01}, p_{02}, p_{03})), etc.

[0219] There are many ways to calculate the commonality score. One possible way is to use weighted summation based on features. For example, for flight altitude, define the altitude difference weight (w_h = 0.3), then the contribution of altitude difference to the commonality score is (w_h*|h_1-h_0| = 0.3*|90-100| = 3). For flight speed, define the speed difference weight (w_v = 0.3), and the contribution of speed difference to the commonality score is (w_v*|v_1-v_0| = 0.3*|6-5| = 0.3). For the inspection point sequence, define a inspection point sequence difference function (d(P_1, P_0)), for example, the value calculated by (d(P_1, P_0)) is (0.4), the inspection point sequence difference weight (w_p = 0.4), then the contribution of inspection point sequence difference to the commonality score is (w_p*d(P_1, P_0) = 0.4*0.4 = 0.16). These contributions are added together to obtain the commonality score (S_{10}=3+0.3+0.16=3.46) (the calculation here is only an example, and the actual calculation may be more complicated).

[0220] After obtaining the commonality score (S_{10}), the first training error index is generated based on it. A common method is to normalize the commonality score and take the inverse. For example, linear normalization is used to map (S_{10}) to the (0-1) interval. For example, the result after mapping is (S'_{10}=0.6) (the specific normalization function is determined according to actual needs), then the first training error index (E_1=frac{1}{1-S'_{10}}=frac{1}{1-0.6}=2.5). This first training error index reflects the degree of difference between the first route planning suggestion feature sample and the initial route planning suggestion feature sample. The larger the error index, the greater the difference.

[0221] Similarly, the commonality score between the second route planning suggestion feature sample and the initial route planning suggestion feature sample is calculated. For example, the second route planning suggestion feature sample includes flight altitude (h_2=80) meters, flight speed (v_2=4) meters / second, inspection point sequence (P_2=(p_{21}, p_{22}, p_{23})), etc.

[0222] According to the calculation method similar to the previous one, for the flight altitude, the altitude difference weight (w_h=0.3), the contribution of the altitude difference to the commonality score is (w_h*|h_2-h_0|=0.3*|80-100|=6). For the flight speed, the speed difference weight (w_v=0.3), the contribution of the speed difference to the commonality score is (w_v*|v_2-v_0|=0.3*|4-5|=0.3). For the inspection point sequence, for example, the value calculated by (d(P_2, P_0)) is (0.5), and the inspection point sequence difference weight (w_p=0.4), then the inspection point sequence difference contributes to the commonality score (w_p*d(P_2, P_0)=0.4*0.5=0.2). Adding these contributions together gives the commonality score (S_{20}=6+0.3+0.2=6.5).

[0223] Then, based on this commonality score, a second training error index is generated. The normalization is also performed and the inverse is taken. For example, if the normalized result is (S'_{20}=0.8), then the second training error index (E_2=frac{1}{1-S'_{20}}=frac{1}{1-0.8}=5) is obtained. This second training error index reflects the degree of difference between the second route planning suggestion feature sample and the initial route planning suggestion feature sample.

[0224] A training evaluation function is generated based on the first training error index (E_1) and the second training error index (E_2). One construction method is a linear combination, for example, a training evaluation function (F=a*E_1+b*E_2), where (a) and (b) are weight coefficients. For example, (a=0.4), (b=0.6), then (F=0.4*2.5+0.6*5=1+3=4). The value of this training evaluation function reflects the overall training error situation, and the relevant algorithms or models can be adjusted and optimized according to the value of this function. For example, if the value of (F) is greater than a certain threshold (for example, the threshold is (3)), it means that the differences between the feature samples of the current route planning suggestions are large, and it may be necessary to adjust certain parameters or algorithms in the route optimization decision branch, such as adjusting the crossover probability and mutation probability in the genetic algorithm, or adjusting the weight coefficient when calculating the commonality score.

[0225] In this way, by calculating the commonality scores between the feature samples of different route planning suggestions to generate training error indicators, and then constructing training evaluation functions, this method can quantitatively evaluate the differences between different samples. Specific numerical examples show how to accurately calculate commonality scores, training error indicators, and training evaluation functions, making the evaluation of route planning more accurate. For example, when calculating commonality scores, the comprehensive consideration of factors such as flight altitude, speed, and inspection point sequence can fully reflect the similarities between samples. The training evaluation function can guide the adjustment of related algorithms or models based on the overall error situation, which helps to improve the accuracy and stability of drone inspection route planning and reduce problems such as low inspection efficiency or safety risks caused by unreasonable route planning.

[0226] Further, Figure 2 FIG. 2 is a schematic diagram of a UAV autonomous inspection processing system 200 provided by the present invention. Figure 2 The drone autonomous inspection processing system 200 shown includes a processor 210, which can call and run a computer program from a memory to implement the method of the present invention.

[0227] Alternatively, if Figure 2 As shown, the UAV autonomous inspection processing system 200 may further include a memory 230. The processor 210 may call and run a computer program from the memory 230 to implement the method of the present invention.

[0228] The memory 230 may be a separate device independent of the processor 210 , or may be integrated into the processor 210 .

[0229] Alternatively, if Figure 2 As shown, the UAV autonomous inspection processing system 200 may also include a transceiver 220, and the processor 210 may control the transceiver 220 to interact with other devices, specifically, to send information or data to other devices, or to receive information or data sent by other devices.

[0230] Optionally, the UAV autonomous inspection processing system 200 can implement the corresponding processes corresponding to the storage engine or components in the storage engine (such as a processing module) or a device deployed with a storage engine in each method of the present invention. For the sake of brevity, they will not be repeated here.

[0231] It should be understood that the processor of the present invention may be an integrated circuit chip having signal processing capabilities.

[0232] It is understood that the memory in the present invention can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memory. It should be noted that the memory of the systems and methods described herein is intended to include but is not limited to suitable types of memory.

[0233] Based on the above, a readable storage medium is provided, on which a program or instruction is stored, and when the program or instruction is executed by a processor, the steps of the above method are implemented.

[0234] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, can be embodied in the form of a computer software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for a terminal (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0235] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present invention, ordinary technicians in this field can also make many forms without departing from the purpose of the present invention and the scope of protection of the present invention, all of which are within the protection of the present invention.

Claims

1. A method for autonomous inspection of overhead line equipment using a UAV, characterized in that: The method is applied to a UAV autonomous inspection processing system, and the method comprises: Obtain X laser point cloud inspection information sets and X inspection route records of X target UAVs, wherein each of the X laser point cloud inspection information sets includes key point detection information and flight status monitoring information, the key point detection information is used to characterize information obtained after laser point cloud scanning and collection of distribution network overhead line equipment for the target UAV within a specified inspection task cycle, and the flight status monitoring information is used to characterize monitoring results corresponding to posture changes of the target UAV within the specified inspection task cycle, and X is greater than or equal to 1; The X laser point cloud inspection information sets are used as the input information of the point cloud element recognition branch in the inspection scheduling processing network, and X target inspection point cloud element knowledge is generated through the point cloud element recognition branch, wherein the inspection scheduling processing network is debugged based on the initial inspection route record, the first inspection route record sample and the second inspection route record sample of the drone sample, the initial inspection route record is used to characterize the route intersection information of the drone sample and the Y coordinated drones corresponding to the drone sample, the first inspection route record sample is used to characterize the route intersection information of the drone sample and the Z coordinated drones determined from the Y coordinated drones, and the second inspection route record sample is used to characterize the route intersection information corresponding to the drone sample and the conflicting drones, and Z is greater than or equal to 1 and less than or equal to Y; The X target inspection point cloud element knowledge and the X inspection route records are used as input information of a route optimization decision branch in the inspection scheduling processing network, and X target route planning suggestion features are generated through the route optimization decision branch; Based on the route clustering processing branch, the X target route planning suggestion features are grouped to generate Q grouping results, where Q is greater than or equal to 1; Identify the inspection energy efficiency judgment weights of the Q sub-cluster results, and determine a drone inspection planning cluster based on the inspection energy efficiency judgment weights of the Q sub-cluster results, wherein the drone inspection planning cluster includes at least one of the target drones.

2. The autonomous inspection and processing method of UAV based on distribution network overhead line equipment according to claim 1, characterized in that: The method further comprises: Obtain a laser point cloud inspection information set and an initial inspection route record of a drone sample, Y cooperative drones of the drone sample, and a conflicting drone of the drone sample, wherein the laser point cloud inspection information set includes key point detection information and flight status monitoring information, the key point detection information is used to characterize information obtained after laser point cloud scanning and collection of distribution network overhead line equipment for the drone sample within a specified inspection task cycle, the flight status monitoring information is used to characterize monitoring results corresponding to posture changes of the drone sample within a specified inspection task cycle, and the initial inspection route record is used to characterize line intersection information between the drone sample and the cooperative drone; Determine Z cooperative drones from the Y cooperative drones, and generate a first inspection route record sample of the drone sample based on the Z cooperative drones and the initial inspection route records of the drone sample; Simulate the route intersection information of the drone sample and the conflicting drone to generate a second inspection route record sample of the drone sample; Using the laser point cloud inspection information set as the input information of the point cloud element recognition branch, and generating inspection point cloud element knowledge samples through the point cloud element recognition branch; The inspection point cloud element knowledge sample, the initial inspection route record, the first inspection route record sample and the second inspection route record sample are used as input information of the route optimization decision branch, and the first route planning suggestion feature sample, the initial route planning suggestion feature sample and the second route planning suggestion feature sample are generated through the route optimization decision branch; Generate a training evaluation function based on the first route planning suggestion feature sample, the initial route planning suggestion feature sample and the second route planning suggestion feature sample; The neuron weights of the point cloud element recognition branch and the route optimization decision branch are updated through the training evaluation function.

3. The autonomous inspection and processing method of UAV based on distribution network overhead line equipment according to claim 2, characterized in that: The method uses the inspection point cloud element knowledge sample, the initial inspection route record, the first inspection route record sample and the second inspection route record sample as input information of a route optimization decision branch, and generates a first route planning suggestion feature sample, an initial route planning suggestion feature sample and a second route planning suggestion feature sample through the route optimization decision branch, including: The inspection point cloud element knowledge sample and the initial inspection route record are used as input information of the route optimization decision branch, and an initial route planning suggestion feature sample is generated through the route optimization decision branch; The inspection point cloud element knowledge sample and the first inspection route record sample are used as input information of the route optimization decision branch, and a first route planning suggestion feature sample is generated through the route optimization decision branch; The inspection point cloud element knowledge sample and the second inspection route record sample are used as input information of the route optimization decision branch, and the second route planning suggestion feature sample is generated through the route optimization decision branch.

4. The autonomous inspection and processing method of UAV based on distribution network overhead line equipment according to claim 3 is characterized in that: The initial inspection route record includes an initial drone inspection route record and an initial distribution network overhead structure spatial distribution, wherein the initial drone inspection route record includes an initial directional drone, and the initial directional drone and the drone sample correspond to the same inspection task event; The initial distribution network overhead structure spatial distribution includes key point structure spatial distribution, and the key point structure spatial distribution and the drone sample correspond to the same distribution network overhead structure area; the line optimization decision branch includes a drone line intersection mining module, a structure space distribution intersection mining module, an inspection route record combination module and a route planning output module; The method of using the inspection point cloud element knowledge sample and the initial inspection route record as input information of a route optimization decision branch, and generating an initial route planning suggestion feature sample through the route optimization decision branch, includes: The initial drone inspection route record and the inspection point cloud element knowledge sample are used as input information of the drone route intersection mining module, and the initial drone route intersection feature is generated by the drone route intersection mining module; The initial distribution network overhead structure spatial distribution and the inspection point cloud element knowledge sample are used as input information of the structure spatial distribution cross mining module, and the initial structure spatial distribution cross features are generated by the structure spatial distribution cross mining module; The initial UAV route intersection feature and the initial structural spatial distribution intersection feature are used as input information of the inspection route record combination module, and the associated structural spatial distribution intersection heat map of the initial route intersection combination information and the associated UAV route intersection heat map of the initial structural spatial distribution intersection combination information are generated by the inspection route record combination module, wherein the associated structural spatial distribution intersection heat map of the initial route intersection combination information is obtained based on the initial structural spatial distribution intersection feature and the structural spatial collaborative distribution relationship network, and the associated UAV route intersection heat map of the initial structural spatial distribution intersection combination information is obtained based on the initial UAV route intersection feature and the UAV collaborative route relationship network; The associated structural spatial distribution cross heat map of the initial route intersection combination information and the associated drone route cross heat map of the initial structural spatial distribution cross combination information are used as input information of the route planning output module, and the initial route planning suggestion feature sample is generated through the route planning output module.

5. The autonomous inspection and processing method of UAV based on distribution network overhead line equipment according to claim 3 is characterized in that: The first inspection route record sample includes a first drone inspection route sample and a first distribution network overhead structure spatial distribution sample, the first drone inspection route sample includes a first directional cruise route sample, and the first directional cruise route sample and the drone sample correspond to the same inspection task event; The first distribution network overhead structure spatial distribution sample includes a first key point structure spatial distribution sample, and the first key point structure spatial distribution sample and the drone sample correspond to the same distribution network overhead structure area; the line optimization decision branch includes a drone line intersection mining module, a structure space distribution intersection mining module, an inspection route record combination module and a route planning output module; The method of using the inspection point cloud element knowledge sample and the first inspection route record sample as input information of a route optimization decision branch, and generating a first route planning suggestion feature sample through the route optimization decision branch, includes: Using the first drone inspection route sample and the inspection point cloud element knowledge sample as input information of the drone route intersection mining module, and generating a first drone route intersection feature sample through the drone route intersection mining module; The first distribution network overhead structure spatial distribution sample and the inspection point cloud element knowledge sample are used as input information of the structure spatial distribution cross mining module, and the first structure spatial distribution cross feature sample is generated by the structure spatial distribution cross mining module; The first UAV route intersection feature sample and the first structural space distribution intersection feature sample are used as input information of the inspection route record combination module, and the first associated distribution cross heat map sample of the first route intersection combination information sample and the first associated route cross heat map sample of the first structural distribution involved feature sample are generated by the inspection route record combination module, wherein the first associated distribution cross heat map sample of the first route intersection combination information sample is obtained based on the first structural space distribution cross feature sample and the structural space collaborative distribution relationship network, and the first associated route cross heat map sample of the first structural distribution involved feature sample is obtained based on the first UAV route intersection feature sample and the UAV collaborative route relationship network; The first associated distribution cross heat map sample of the first route intersection combination information sample and the first associated route cross heat map sample of the first structural distribution involved feature sample are used as input information of the route planning output module, and the first route planning suggestion feature sample is generated through the route planning output module.

6. The autonomous inspection and processing method of UAV based on distribution network overhead line equipment according to claim 3, characterized in that: The second inspection route record sample includes a second UAV inspection route sample and a second distribution network overhead structure spatial distribution sample, the second UAV inspection route sample includes a second directional cruise route sample, and the second directional cruise route sample and the UAV sample correspond to the same inspection task event; The second distribution network overhead structure spatial distribution sample includes a second key point structure spatial distribution sample, and the second key point structure spatial distribution sample and the drone sample correspond to the same distribution network overhead structure area; the line optimization decision branch includes a drone line intersection mining module, a structure space distribution intersection mining module, an inspection route record combination module and a route planning output module; The method of using the inspection point cloud element knowledge sample and the second inspection route record sample as input information of a route optimization decision branch, and generating a second route planning suggestion feature sample through the route optimization decision branch, includes: Using the second drone inspection route sample and the inspection point cloud element knowledge sample as input information of the drone route intersection mining module, and generating a second drone route intersection feature sample through the drone route intersection mining module; The second distribution network overhead structure spatial distribution sample and the inspection point cloud element knowledge sample are used as input information of the structure spatial distribution cross mining module, and a second structure spatial distribution cross feature sample is generated by the structure spatial distribution cross mining module; The second UAV route intersection feature sample and the second structural space distribution intersection feature sample are used as input information of the inspection route record combination module, and the second associated distribution intersection heat map sample of the second route intersection combination information sample and the second associated route intersection heat map sample of the second structural distribution involved feature sample are generated by the inspection route record combination module, wherein the second associated distribution cross heat map sample of the second route intersection combination information sample is obtained based on the second structural space distribution intersection feature sample and the structural space collaborative distribution relationship network, and the second associated route intersection heat map sample of the second structural distribution involved feature sample is obtained based on the second UAV route intersection feature sample and the UAV collaborative route relationship network; The second associated distribution cross heat map sample of the second route intersection combination information sample and the second associated route cross heat map sample of the second structural distribution involved feature sample are used as input information of the route planning output module, and the second route planning suggestion feature sample is generated through the route planning output module.

7. The autonomous inspection and processing method of UAV based on distribution network overhead line equipment according to claim 2, characterized in that: The generating of a training evaluation function based on the first route planning suggestion feature sample, the initial route planning suggestion feature sample and the second route planning suggestion feature sample comprises: generating a first training error indicator based on a commonality score between the first route planning suggestion feature sample and the initial route planning suggestion feature sample; generating a second training error indicator based on a commonality score between the second route planning suggestion feature sample and the initial route planning suggestion feature sample; A training evaluation function is generated based on the first training error indicator and the second training error indicator.

8. The autonomous inspection and processing method of UAV based on distribution network overhead line equipment according to claim 1, characterized in that: The X laser point cloud inspection information sets include X key point detection information and X flight status monitoring information, and the point cloud element recognition branch includes a key point quantization recognition module, a posture point cloud recognition module and a full connection module; The method of using the X laser point cloud inspection information sets as the incoming information of the point cloud element recognition branch in the inspection scheduling processing network, and generating X target inspection point cloud element knowledge through the point cloud element recognition branch, includes: Using the X key point detection information as input information of the key point quantization identification module, and generating X key point quantization knowledge codes through the key point quantization identification module; Using the X flight status monitoring information as input information of the attitude point cloud recognition module, and generating X attitude point cloud knowledge codes through the point cloud element recognition branch; The X key point quantization knowledge codes and the X posture point cloud knowledge codes are used as input information of the fully connected module, and X target inspection point cloud element knowledge is generated through the input information of the fully connected module.

9. The autonomous inspection and processing method of UAV based on distribution network overhead line equipment according to claim 8, characterized in that: The key point quantization identification module includes a vector space conversion unit and a knowledge collision unit; The step of using the X key point detection information as input information of the key point quantization identification module and generating X key point quantization knowledge codes through the key point quantization identification module includes: Based on the key point image semantics of the X key point detection information and the X key point detection information, generate X local point cloud description semantic graphs; The X local point cloud description semantic graphs are used as input information of the vector space conversion unit, and X local point clouds are generated based on space migration encoding by the vector space conversion unit, wherein the vector space conversion unit is used to sum the corresponding feature members in each local point cloud description semantic graph in the X local point cloud description semantic graphs; The X local point cloud description semantic graphs are used as input information of the knowledge collision unit, and X local point cloud advanced space migration codes are generated by the knowledge collision unit, wherein the knowledge collision unit is used to calculate the product of corresponding feature members in each local point cloud description semantic graph in the X local point cloud description semantic graphs, and sum the results of calculating the product of corresponding feature members; The X local point clouds are integrated based on the space migration coding and the X local point clouds are integrated based on the space migration coding to obtain X key point quantization knowledge coding.

10. An autonomous inspection and processing system for unmanned aerial vehicles, characterized in that: The method comprises at least one processor and a memory; the memory stores computer-executable instructions; the at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes the method according to any one of claims 1 to 9.

Images