Unmanned aerial vehicle inspection management system and method suitable for intelligent distribution network

Abstract

The invention discloses an unmanned aerial vehicle inspection management system and method suitable for an intelligent distribution network, relates to the technical field of unmanned aerial vehicle control, and solves the technical problem that in the prior art, inspection is completed by using an algorithm until all grids are traversed, and the influence of equipment state change on partitions is not considered, so that the distribution of inspection resources is unreasonable. The method comprises the steps of performing basic partition on a substation; judging whether equipment is being maintained in the transformer substation or not; acquiring an equipment operation condition corresponding to the area information; constructing a partition updating mechanism according to the operation condition of the equipment to obtain a new level area; setting an inspection period according to the grade area; an inspection observation period is set for the maintained equipment; predicting the damage probability of the maintained equipment in the inspection observation period through a formula; judging whether the damage probability is smaller than a preset stable threshold value or not; setting a stable period through the damage probability, and carrying out basic partitioning on an equipment maintenance area after the stable period is ended; the routing inspection resource configuration is optimized, and the routing inspection efficiency is improved.

Description

Technical Field

[0001] This invention belongs to the field of drone inspection and relates to drone control technology. Specifically, it is a drone inspection management system and method applicable to smart power distribution networks. Background Technology

[0002] Drone inspections, equipped with high-definition cameras, infrared thermal imagers, and other sensors, can collect various data in real time, including high-definition video, temperature, and humidity. Through intelligent algorithm analysis, drones can accurately identify potential problem areas, improving the accuracy and reliability of inspections. Drones can cover large inspection areas in a short time, greatly increasing the coverage and efficiency of inspections. By incorporating advanced sensors and data processing systems, drones can collect and analyze data in real time, quickly identifying potential problems and reducing the time and labor required for manual inspections. Drone inspection systems can also transmit data in real time, providing timely and accurate information support for decision-makers.

[0003] The prior art (patent application CN118567370A) discloses a gridded inspection method and system for power distribution networks using drones, comprising the following steps: Step 1, determining the inspection area and dividing the area into several grids; Step 2, acquiring the drone's take-off and landing positions and its cruising distance, and determining the grids traversed by the drone in a single inspection based on the drone's take-off and landing positions and cruising distance, until all grids are traversed, completing the inspection plan; Step 3, the drone inspects the inspection area according to the inspection plan; Step 4, verifying the images inspected by the drone. If the image verification is correct, the inspection is completed; if an image verification error occurs, a re-inspection is initiated for the grids with incorrect image verification, until all images are verified correctly, completing the inspection. Existing technologies use algorithms to complete inspections until all grids are traversed, without considering the impact of equipment status changes on partitions. This leads to unreasonable allocation of inspection resources, increases the time spent on ineffective flights, and reduces the efficiency of UAV inspections.

[0004] This invention provides a drone inspection management system and method suitable for smart power distribution networks to solve the above-mentioned technical problems. Summary of the Invention

[0005] This invention aims to solve at least one of the technical problems existing in the prior art; to this end, this invention proposes a drone inspection management system and method suitable for smart distribution networks, which solves the technical problem in the prior art that the inspection is completed by using an algorithm until all grids are traversed, without considering the impact of equipment status changes on the partition, thus resulting in unreasonable allocation of inspection resources.

[0006] To achieve the above objectives, a first aspect of the present invention provides a drone inspection and management system suitable for smart power distribution networks, comprising: a data acquisition module and an inspection cycle setting module; Data acquisition module: Divides the substation into basic zones to obtain several different levels of areas; collects area information of different levels of areas through several sensors of the UAV; Inspection cycle setting module: Determines whether there is equipment under maintenance in the substation; if yes, marks the area corresponding to the equipment under maintenance as the equipment maintenance area and stops the inspection of the equipment maintenance area; if no, analyzes the area information through image analysis technology to obtain the equipment operation status corresponding to the area information; constructs a partition update mechanism based on the equipment operation status to obtain new level areas; sets the inspection cycle based on the level areas. Set an inspection and observation period for the equipment that has been repaired; predict the probability of damage to the repaired equipment during the inspection and observation period using a formula; determine whether the probability of damage is less than the preset stability threshold; if yes, set a stability period based on the probability of damage and stop inspections during the stability period; after the stability period ends, divide the equipment repair area into basic zones and re-acquire the inspection cycle; if not, issue a reminder signal to notify the staff.

[0007] Preferably, a drone inspection management system suitable for smart distribution networks further includes: a database for recording whether there is equipment under maintenance in the substation; staff preset the importance, thermal threshold, stability threshold, basic cycle, inspection observation period and operating coefficient of the equipment in the substation in the database; wherein the importance value ranges from [0, 10].

[0008] Preferably, the substation is divided into several different levels of zones based on basic zoning, including: Retrieve the terrain complexity and equipment importance of the substation from the database; the terrain complexity includes the slope of the terrain and the distribution of obstacle types. Areas with a slope less than 20° are designated as Zone 1A, areas with a slope between 20° and 50° are designated as Zone 2A, and areas with a slope greater than 50° are designated as Zone 3A. The system retrieves regional information and combines it with the YOLOv5 object detection algorithm to generate an obstacle distribution heatmap. It then obtains heat values ​​based on the heatmap, setting preset heat thresholds T1 and T2. Regions with heat values ​​less than T2 are designated as Level 1 B zones; regions with heat values ​​between [T2, T1] are designated as Level 2 B zones; and regions with heat values ​​greater than T2 are designated as Level 3 B zones, where T1 > T2 > 0. The importance of the devices is obtained from the database and classified. Devices with an importance of [0, 3] are marked as Level 1 C zone, devices with an importance of [3, 6] are marked as Level 2 C zone, and devices with an importance of (6, 10] are marked as Level 3 C zone. The region series is obtained by summing the series of overlapping regions of three categories (A, B, and C) or any two categories. The region series includes regions of levels one to nine. Regions of levels one to three are marked as category one. Regions of levels four and five are marked as category one. Regions of levels six to nine are marked as category three.

[0009] This invention introduces intelligent technologies such as the YOLOv5 target detection algorithm to achieve automatic identification and classification of substation areas, improving work efficiency and accuracy. By comprehensively considering the complexity of the terrain and the importance of the equipment, the substation is divided into areas of different levels, realizing refined management of the substation. This helps managers better understand the actual situation of the substation and formulate more reasonable operation and maintenance strategies.

[0010] Preferably, the step of generating an obstacle distribution heatmap by combining regional information with the YOLOv5 target detection algorithm includes: Retrieve area information and define the types of obstacles included in the substation within the area information; retrieve the YOLOv5 model and use aerial images containing obstacle types as input to train the YOLOv5 model; among which, obstacle types include: high-voltage lines, construction machinery, and trees that obstruct equipment; The YOLOv5 model is used to convert the coordinates of the detected obstacle center points into a geographic coordinate system, and the location data of the UAV is used for georeferencing. Based on the coordinates of the obstacle center points, a density distribution map is generated using a Gaussian kernel function. The kernel density estimate is obtained from the density distribution map and mapped to a color gradient to generate a heat map of obstacle distribution. The color gradient includes red for high density and blue for low density.

[0011] This invention uses the YOLOv5 model to accurately quantify the density of obstacles, providing data-driven decision support for drone inspection.

[0012] Preferably, the step of obtaining the device operating status corresponding to the region information by analyzing the region information through image analysis technology includes: Retrieve region information; the region information includes: visible light image, infrared thermal image and ultraviolet image of the same region; preprocess the region information; the preprocessing includes: registering the visible light, infrared thermal and ultraviolet images to the same coordinate system, enhancing low-light images and removing noise using median filtering; The ResNet-50 classification model is invoked, and the preprocessed visible light image is input to obtain the corrosion level of the equipment. The corrosion level includes: none, light, and heavy. The image corresponding to the equipment without corrosion is assigned to XS1, the image corresponding to the equipment with light corrosion is assigned to XS2, and the image corresponding to the equipment with heavy corrosion is assigned to XS3. The number of equipment with no corrosion, light corrosion, and heavy corrosion is counted and recorded as x1, x2, and x3, respectively. The crack situation of the device in the visible light image is obtained by using the Mask R-CNN algorithm; the image corresponding to the device with crack is assigned to LW, and the number of devices with crack in the image is counted l. Obtain the surface temperature distribution of the corresponding equipment in the regional information, and obtain the historical baseline temperature of the equipment from the database. Use a clustering algorithm to mark the equipment with a temperature exceeding the historical baseline temperature as overheated equipment. Assign a segment GR to the image containing overheated equipment, and count the number r of overheated equipment. Ultraviolet imaging is retrieved, and the insulator degradation value LH of the equipment is obtained by detecting the corona discharge spot using ultraviolet imaging technology, as well as the number h of degraded equipment in the corresponding image.

[0013] This invention uses visible light images to detect damage to the appearance of equipment; generates infrared thermal images to detect abnormal temperatures in equipment; captures corona discharge phenomena using ultraviolet imaging; and constructs a high-precision three-dimensional model using LiDAR point clouds to assist in localization and occlusion analysis.

[0014] Preferably, the step of constructing a partition update mechanism based on device operating conditions to obtain new graded regions includes: Retrieve equipment operating status using formulas The image score is calculated; where a1, a2, a3, and a4 are the running coefficients one, two, three, and four; the running coefficients are obtained from the running coefficient table; Get the maximum value among all image ratings and minimum value The image rating will be placed in [ , The image's corresponding rating region remains within its original rating region; the image rating is placed within [...]. , The region corresponding to the image is marked as a new level region with one increment from the original level.

[0015] It should be noted that when the original region is a level three region, if the image score is in […] , The corresponding grading region of the image remains unchanged.

[0016] Preferably, the step of setting the inspection cycle according to the grade area includes: The average inspection time for each Level i area was obtained through pre-flight testing. Where i = 1, 2, 3; Through formula The inspection cycle of region i is calculated; where, As a preset base period, This represents the total number of devices in the inspection area. This represents the number of devices that malfunctioned in the inspection area.

[0017] It should be noted that the preset basic cycle is set based on inspection costs and mandatory industry standards.

[0018] Preferably, the method of predicting the probability of damage to the equipment that has undergone maintenance during the inspection and observation period using a formula includes: Determine if the equipment being repaired is the same model as the equipment before repair; if yes, record the probability of damage during the inspection and observation period as 0; otherwise, obtain the accuracy of the corresponding equipment processing data. Retrieve the preset inspection observation period and divide it into N short cycles; construct an accuracy change curve with each short cycle as the independent variable and accuracy as the dependent variable; where N>3. Determine if the slope of the curve is less than 0; if yes, issue warning signal one and notify staff to handle the situation; if no, determine if the slope continues to increase; if yes, then use the formula... The probability of damage is calculated; if not, a second warning signal is issued; if not, a second warning signal is issued; where A(t) is obtained by fitting the accuracy change curve through linear regression.

[0019] Preferably, the step of setting a stabilization period based on the probability of damage includes: retrieve damage probability To obtain the lifespan S of the corresponding equipment, use the formula Tw=S×[|1-ln( The stable period is calculated using the method described above.

[0020] To achieve the above objectives, a second aspect of the present invention provides a method for unmanned aerial vehicle (UAV) inspection and management suitable for smart distribution networks, comprising: The substation is divided into several different levels of zones; regional information of these different levels of zones is collected using several sensors of a drone. Determine if any equipment in the substation is under maintenance; if yes, mark the area corresponding to the equipment under maintenance as the equipment maintenance area and stop the inspection of the equipment maintenance area; if no, analyze the area information using image analysis technology to obtain the equipment operation status corresponding to the area information; construct a zoning update mechanism based on the equipment operation status to obtain new level areas; set the inspection cycle based on the level areas; Set an inspection and observation period for the equipment that has been repaired; predict the probability of damage to the repaired equipment within the preset inspection and observation period using a formula; determine whether the probability of damage is less than the preset stability threshold; if yes, set a stability period based on the probability of damage and stop inspections during the stability period; after the stability period ends, divide the equipment repair area into basic zones and re-acquire the inspection cycle; if no, issue a reminder signal to notify the staff.

[0021] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention divides substations into different levels of zones through basic zoning, enabling targeted inspections and avoiding the resource waste caused by indiscriminate inspections. The drones, equipped with multiple sensors, can quickly collect area information, improving the speed and accuracy of data acquisition. When equipment under maintenance is detected, the area is automatically marked and inspection is stopped, avoiding unnecessary inspection tasks and saving drones and manpower. The zoning and inspection cycle are dynamically adjusted based on equipment operating conditions, allowing inspection resources to be more rationally allocated to the areas that need them. Drone inspections reduce the need for personnel to enter high-voltage, high-risk areas of the substation, lowering the risk of injury. Real-time monitoring of equipment status allows for timely detection of potential problems, helping to prevent accidents caused by equipment failures.

[0022] 2. This invention analyzes regional information using image analysis technology, enabling a more accurate understanding of equipment operating status and providing a basis for predicting equipment failures. It uses formulas to predict the probability of equipment damage during the post-repair inspection and observation period, providing a scientific basis for setting a stabilization period and adjusting inspection strategies. Setting an inspection and observation period for repaired equipment and determining whether it has entered a stabilization period based on the damage probability helps ensure stable operation of the equipment after repair. Stopping inspections during the stabilization period reduces unnecessary interference and is beneficial for the long-term stable operation of the equipment. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a schematic diagram illustrating the module relationships included in this invention; Figure 2 This is a schematic diagram illustrating the specific steps of dividing the inspection area according to the present invention; Figure 3 A schematic diagram illustrating the specific steps for setting the inspection cycle of this invention; Figure 4This is a flowchart illustrating the inspection management process of the present invention. Detailed Implementation

[0025] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] Please see Figure 1 The first aspect of the present invention provides a drone inspection and management system suitable for smart power distribution networks, including: a data acquisition module, an inspection cycle setting module, and a database; Data acquisition module: Divides the substation into basic zones to obtain several different levels of areas; collects area information of different levels of areas through several sensors of the UAV; Inspection cycle setting module: Determines whether there is equipment under maintenance in the substation; if yes, marks the area corresponding to the equipment under maintenance as the equipment maintenance area and stops the inspection of the equipment maintenance area; if no, analyzes the area information through image analysis technology to obtain the equipment operation status corresponding to the area information; constructs a partition update mechanism based on the equipment operation status to obtain new level areas; sets the inspection cycle based on the level areas. Set an inspection and observation period for the equipment that has been repaired; predict the probability of damage to the equipment that has been repaired during the inspection and observation period using a formula; determine whether the probability of damage is less than the preset stability threshold; if yes, set a stability period based on the probability of damage and stop inspection during the stability period; after the stability period ends, divide the equipment repair area into basic zones and re-acquire the inspection cycle; if no, issue a reminder signal to notify the staff. Database: Used by staff to add equipment with sudden abnormal status to the database; stores real-time inspection information; presets the importance of equipment in the substation; real-time inspection information includes: real-time weather conditions, real-time equipment fault warnings, and faulty equipment that appears during the inspection; the importance value ranges from [0, 10].

[0027] Please see Figure 2 The specific steps for dividing the inspection area include obtaining the terrain complexity and equipment importance in the substation from the database; among which, terrain complexity includes: the slope of the terrain and the distribution of obstacle types; Areas with a slope less than 20° are designated as Zone 1A, areas with a slope between 20° and 50° are designated as Zone 2A, and areas with a slope greater than 50° are designated as Zone 3A. The system retrieves regional information and combines it with the YOLOv5 object detection algorithm to generate an obstacle distribution heatmap. It then obtains heat values ​​based on the heatmap, setting preset heat thresholds T1 and T2. Regions with heat values ​​less than T2 are designated as Level 1 B zones; regions with heat values ​​between [T2, T1] are designated as Level 2 B zones; and regions with heat values ​​greater than T2 are designated as Level 3 B zones, where T1 > T2 > 0. The importance of the devices is obtained from the database and classified. Devices with an importance of [0, 3] are marked as Level 1 C zone, devices with an importance of [3, 6] are marked as Level 2 C zone, and devices with an importance of (6, 10] are marked as Level 3 C zone. The region series is obtained by summing the series of overlapping regions of three categories (A, B, and C) or any two categories. The region series includes regions of levels one to nine. Regions of levels one to three are marked as category one. Regions of levels four and five are marked as category one. Regions of levels six to nine are marked as category three.

[0028] Retrieve region information; the region information includes: visible light image, infrared thermal image and ultraviolet image of the same region; preprocess the region information; the preprocessing includes: registering the visible light, infrared thermal and ultraviolet images to the same coordinate system, enhancing low-light images and removing noise using median filtering; The ResNet-50 classification model is invoked, and the preprocessed visible light image is input to obtain the corrosion level of the equipment. The corrosion level includes: none, light, and heavy. The image corresponding to the equipment without corrosion is assigned to XS1, the image corresponding to the equipment with light corrosion is assigned to XS2, and the image corresponding to the equipment with heavy corrosion is assigned to XS3. The number of equipment with no corrosion, light corrosion, and heavy corrosion is counted and recorded as x1, x2, and x3, respectively. The crack situation of the device in the visible light image is obtained by using the Mask R-CNN algorithm; the image corresponding to the device with crack is assigned to LW, and the number of devices with crack in the image is counted l. Obtain the surface temperature distribution of the corresponding equipment in the regional information, and obtain the historical baseline temperature of the equipment from the database. Use a clustering algorithm to mark the equipment with a temperature exceeding the historical baseline temperature as overheated equipment. Assign a segment GR to the image containing overheated equipment, and count the number r of overheated equipment. Retrieve ultraviolet imaging and use ultraviolet imaging technology to detect corona discharge spots to obtain the insulator degradation value LH of the equipment, as well as the number h of degraded equipment in the corresponding image; Retrieve equipment operating status using formulas The image score is calculated; where a1, a2, a3, and a4 are the running coefficients one, two, three, and four; the running coefficients are obtained from the running coefficient table; Get the maximum value among all image ratings and minimum value The image rating will be placed in [ , The image's corresponding rating region remains within its original rating region; the image rating is placed within [...]. , The region corresponding to the image is marked as a new level region with one increment from the original level.

[0029] It should be noted that when the original region is a level three region, if the image score is in […] , The corresponding grading region of the image remains unchanged.

[0030] For example, it is necessary to conduct drone inspections of substation D. The task is to obtain the terrain complexity and the importance of all equipment in substation D. There are 6 Class A areas with a slope of less than 20°, 12 Class A areas with a slope between 20° and 50°, and 2 Class A areas with a slope greater than 50°. The system retrieves regional information and combines it with the YOLOv5 object detection algorithm to generate an obstacle distribution heatmap. Heat values ​​are then obtained from the heatmap, with preset heat thresholds T1 at the 0.3 quantile and T2 at the 0.7 quantile. Specifically, there are 21 Level 1 B zones with heat values ​​less than T2; 9 Level 2 B zones with heat values ​​between [T2, T1]; and 0 Level 3 B zones with heat values ​​greater than T2. The importance of the devices is obtained from the database and classified. There are 9 devices in the first-level C zone with an importance of [0, 3], 15 devices in the second-level C zone with an importance of [3, 6], and 6 devices in the third-level C zone with an importance of (6, 10). There are 4 overlapping areas among regions A1, B1, and C1; 4 overlapping areas among regions A2, B1, and C1; 1 overlapping area among regions A3, B1, and C1; 2 overlapping areas among regions A1, B2, and C2; 7 overlapping areas among regions A2, B2, and C1; and 2 overlapping areas among regions A2, B1, and C2. There is one overlapping region between regions A3 and C1; there is one overlapping region between regions B1 and C2; there are four overlapping regions between regions B1 and C3; and there are five non-overlapping regions, all of which are regions B1. Among them, there are 5 regions at level 1, 0 regions at level 2, 5 regions at level 3, 9 regions at level 4, 12 regions at level 5, and 0 regions at level 6 or above. Ten regions belonging to the first to third levels of regional hierarchy were classified as Category 1 regions; 21 regions belonging to the fourth and fifth levels of regional hierarchy were classified as Category 2 regions; there were no Category 3 regions. Four regions were randomly selected from the above 21 regions, and their corresponding region information was retrieved. This region information included: visible light image, infrared thermal image, and ultraviolet image of the same region. The region information was then preprocessed, including: registering the visible light, infrared thermal, and ultraviolet images to the same coordinate system, enhancing low-light images, and using median filtering to remove noise. A total of four visible light images were acquired and denoted as R1, R2, R3, and R4. R1 and R2 represent region type I, and R3 and R4 represent region type II. The ResNet-50 classification model was invoked, and the corrosion level of the equipment was obtained by inputting the preprocessed visible light image. The image corresponding to the equipment without corrosion was assigned a score of 0, the image corresponding to the equipment with slight corrosion was assigned a score of 30, and the image corresponding to the equipment with severe corrosion was assigned a score of 50. In Figure R1, the number of equipment with no corrosion, slight corrosion, and severe corrosion was counted as 10, 1, and 1, respectively; in Figure R2, the number of equipment with no corrosion, slight corrosion, and severe corrosion was counted as 6, 0, and 0, respectively; in Figure R3, the number of equipment with no corrosion, slight corrosion, and severe corrosion was counted as 5, 1, and 0, respectively; and in Figure R4, the number of equipment with no corrosion, slight corrosion, and severe corrosion was counted as 4, 1, and 1, respectively. The crack situation of the device in the visible light image was obtained by using the Mask R-CNN algorithm; the image corresponding to the device with crack was divided into 30 parts. The number of devices with crack in Figure R1 is 2; the number of devices with crack in Figures R2, R3 and R4 is 0. The surface temperature distribution of the corresponding equipment in the regional information is obtained, and the historical baseline temperature of the equipment is obtained from the database. The clustering algorithm is used to mark the equipment with temperature exceeding the historical baseline temperature as overheated equipment. The image with overheated equipment is assigned a division of 30. The number of overheated equipment in Figure R1 and Figure R4 is 0; the number of overheated equipment in Figure R2 is 2; and the number of overheated equipment in Figure R3 is 1. Through formula The kernel density estimate for the center point of the obstacle (x, y) is calculated. The coordinates of the detected obstacle center points are converted to a geographic coordinate system using the YOLOv5 model, and georeferencing is performed using the UAV's location data; using the formula... The kernel density estimate of the obstacle center point coordinates (x, y) is calculated; where n represents the number of obstacles, h represents the bandwidth, and ( , () represents the coordinates of the i-th obstacle; the kernel density estimate is mapped to a color gradient to generate a heat map of obstacle distribution; ultraviolet imaging is retrieved, and the corona discharge spot is detected by ultraviolet imaging technology to obtain images R1, R2, R3 and R4 where there are no degraded devices; Retrieve equipment operating status using formulas The image score is calculated; where a1, a2, a3, and a4 are the running coefficients one, two, three, and four; the running coefficients are obtained from the running coefficient table (as shown in Table 1); and the formula H = The total obstacle score is calculated. Table 1. Operating Coefficients

[0031] Through the formula H= The total obstacle score is calculated; since 130 > 100, then a1 = 0.2, a2 ​​= 0.08, a3 = 0.07, a4 = 0.18; Through formula The calculated R1 score for the image is 20.8; using the formula... The R² score for the image is 4.2; the R³ score is 8.1; and the R⁴ score is 16. The maximum value among the image scores is... It is 20.8, the minimum value. It is 4.2; The grade region corresponding to the image with a score of [4.2, 12.5) is maintained in its original grade region; the grade region corresponding to the image with a score of [12.5, 20.8] is marked with one more level than its original grade and is then marked as a new grade region; if the original grade region is a level 3 region, the grade region corresponding to the image with a score of [12.5, 20.8] remains unchanged. Image R2 remains a Class I region, while images R3 and R4 remain Class II regions; image R1 is upgraded to a Class II region.

[0032] Please see Figure 3 The specific steps for setting the inspection cycle are as follows: the average inspection time for each level i area is obtained through pre-flight testing. Where i = 1, 2, 3; Through formula The inspection cycle of region i is calculated; where, As a preset base period, This represents the total number of devices in the inspection area. This represents the number of malfunctioning devices in the inspection area. Determine if the equipment being repaired is the same model as the equipment before repair; if yes, record the probability of damage during the inspection and observation period as 0; otherwise, obtain the accuracy of the corresponding equipment processing data. Retrieve the preset inspection observation period and divide it into N short cycles; construct an accuracy change curve with each short cycle as the independent variable and accuracy as the dependent variable; where N>3. Determine if the slope of the curve is less than 0; if yes, issue warning signal one and notify staff to handle the situation; if no, determine if the slope continues to increase; if yes, then use the formula... The probability of damage is calculated; otherwise, a second warning signal is issued; where A(t) is obtained by fitting the accuracy change curve through linear regression. Determine if the probability of failure is less than a preset stability threshold; if yes, retrieve the probability of failure. To obtain the lifespan S of the corresponding equipment, use the formula Tw=S×[|1-ln( If the stabilization period is not calculated, a reminder signal will be issued to notify the staff.

[0033] It should be noted that, When it is 0, then .

[0034] For example, Figures R1, R2, and R4 represent secondary regions, while Figure R3 represents a primary region. The average inspection time for each primary region is obtained through pre-flight testing. 1 day 2 days; In accordance with the minimum frequency requirements of the "Power Equipment Inspection Regulations" (such as DL / T741-2019), Set to 7 days; total number of devices inspected in the first-level area The number is 60, representing the number of devices that malfunctioned during inspections in the first-level area. The total number of devices inspected in the secondary area is 2. The number is 10, representing the number of devices that malfunctioned during inspections in the secondary area. =1; Through formula =26.956 calculates to an inspection cycle of 27 days for a Class I area; Through formula =31.8 Calculation yields an inspection cycle of 32 days for the Class II area; The equipment that has been repaired is a different model from the equipment before repair. The preset inspection and observation period of 16 days is used and divided into 4 short cycles. Each short cycle is used as the independent variable and the accuracy is used as the dependent variable to construct an accuracy change curve. If the slope of the accuracy change curve is consistently greater than 0 and continuously increasing, then by fitting the accuracy change curve using linear regression, we obtain A(t) = 0.8 + 0.6t. = =21.25%; the preset stability threshold is 30%; Since 21.25% < 30%, the corresponding equipment lifespan is 600 days, calculated using the formula Tw = S × [|1 - |ln( The calculation yields a stable period of 444 days. During the stable period, inspections are stopped. After the stable period ends, the equipment maintenance area is divided into basic zones, and the inspection cycle is re-acquired.

[0035] Please see Figure 4 A second aspect of the present invention provides a method for unmanned aerial vehicle (UAV) inspection and management suitable for smart power distribution networks, comprising: The substation is divided into several different levels of zones; regional information of these different levels of zones is collected using several sensors of a drone. Determine if any equipment in the substation is under maintenance; if yes, mark the area corresponding to the equipment under maintenance as the equipment maintenance area and stop the inspection of the equipment maintenance area; if no, analyze the area information using image analysis technology to obtain the equipment operation status corresponding to the area information; construct a zoning update mechanism based on the equipment operation status to obtain new level areas; set the inspection cycle based on the level areas; Set an inspection and observation period for the equipment that has been repaired; predict the probability of damage to the repaired equipment within the preset inspection and observation period using a formula; determine whether the probability of damage is less than the preset stability threshold; if yes, set a stability period based on the probability of damage and stop inspections during the stability period; after the stability period ends, divide the equipment repair area into basic zones and re-acquire the inspection cycle; if no, issue a reminder signal to notify the staff.

[0036] Some of the data in the above formula are calculated by removing dimensions and taking their numerical values. The formula is the closest to the real situation obtained by software simulation of a large amount of collected data. The preset parameters and preset thresholds in the formula are set by those skilled in the art according to the actual situation or obtained through simulation of a large amount of data.

[0037] The working principle of this invention is as follows: The substation is divided into basic zones to obtain several zones of different levels; zone information of different levels is collected using several sensors of a drone; it is determined whether any equipment in the substation is under maintenance; if so, the zone corresponding to the equipment under maintenance is marked as the equipment maintenance zone, and inspection of the equipment maintenance zone is stopped; if not, the zone information is analyzed using image analysis technology to obtain the equipment operating status corresponding to the zone information; a zone update mechanism is constructed based on the equipment operating status to obtain new level zones; an inspection cycle is set according to the level zones; an inspection observation period is set for the equipment that has completed maintenance; the probability of damage to the equipment that has completed maintenance within the preset inspection observation period is predicted using a formula; it is determined whether the probability of damage is less than a preset stability threshold; a stability period is set based on the probability of damage, and inspection is stopped within the stability period; after the stability period ends, the equipment maintenance zone is divided into basic zones, and the inspection cycle is obtained again.

[0038] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A drone inspection and management system suitable for smart power distribution networks, characterized in that, include: Data acquisition module and inspection cycle setting module; Data acquisition module: Divides the substation into basic zones to obtain several different levels of areas; collects area information of different levels of areas through several sensors of the UAV; Inspection cycle setting module: Determines whether there is equipment under maintenance in the substation; if yes, marks the area corresponding to the equipment under maintenance as the equipment maintenance area and stops the inspection of the equipment maintenance area; if no, analyzes the area information through image analysis technology to obtain the equipment operation status corresponding to the area information; constructs a partition update mechanism based on the equipment operation status to obtain new level areas; sets the inspection cycle based on the level areas. Set an inspection and observation period for the equipment that has been repaired; predict the probability of damage to the equipment that has been repaired during the inspection and observation period using a formula; determine whether the probability of damage is less than the preset stability threshold; if so, set a stability period based on the probability of damage and stop inspection during the stability period; after the stability period ends, divide the equipment repair area into basic zones and obtain the inspection cycle again. If not, a reminder signal will be sent to notify the staff.

2. The UAV inspection and management system for smart power distribution networks according to claim 1, characterized in that, include: Database: Used by staff to add devices that suddenly malfunction into the database; Store real-time inspection information; The importance of equipment in the substation is preset; real-time inspection information includes: real-time weather conditions, real-time equipment fault warnings, and faulty equipment that appears during the inspection; the importance value ranges from [0, 10].

3. The UAV inspection and management system for smart power distribution networks according to claim 2, characterized in that, The substation is divided into several different levels of zones based on its basic zoning, including: The terrain complexity and equipment importance of substations are obtained from the database; the substations are classified into different levels based on the terrain complexity and equipment conditions; the terrain complexity includes the slope of the terrain and the distribution of obstacle types. The areas divided according to the slope of the terrain are marked as different levels of Zone A; the area information is retrieved and combined with the YOLOv5 target detection algorithm to generate an obstacle distribution heat map, and the areas divided according to the distribution of obstacle types are marked as different levels of Zone B; the areas divided according to the importance of the equipment are marked as different levels of Zone C. The region series is obtained by summing the series of overlapping regions of three categories (A, B, and C) or any two categories. The region series includes regions of levels one to nine. Regions of levels one to three are marked as category one. Regions of levels four and five are marked as category one. Regions of levels six to nine are marked as category three.

4. The UAV inspection and management system for smart power distribution networks according to claim 3, characterized in that, The step of generating an obstacle distribution heatmap by combining regional information with the YOLOv5 target detection algorithm includes: Retrieve area information and define the types of obstacles included in the substation within the area information; retrieve the YOLOv5 model and use aerial images containing obstacle types as input to train the YOLOv5 model; among which, obstacle types include: high-voltage lines, construction machinery, and trees that obstruct equipment; The YOLOv5 model is used to convert the coordinates of the detected obstacle center points into a geographic coordinate system, and the location data of the UAV is used for georeferencing. Based on the coordinates of the obstacle center points, a density distribution map is generated using a Gaussian kernel function. The kernel density estimate is obtained from the density distribution map and mapped to a color gradient to generate a heat map of obstacle distribution.

5. A drone inspection and management system for smart power distribution networks according to claim 1, characterized in that, The method of obtaining the device operation status corresponding to the region information by analyzing region information through image analysis technology includes: Retrieve region information; the region information includes: visible light image, infrared thermal image and ultraviolet image of the same region; preprocess the region information; the preprocessing includes: registering the visible light, infrared thermal and ultraviolet images to the same coordinate system, enhancing low-light images and removing noise using median filtering; The ResNet-50 classification model is invoked, and the preprocessed visible light image is input to obtain the corrosion level of the equipment. The corrosion level includes: none, light, and heavy. The image corresponding to the equipment without corrosion is assigned to XS1, the image corresponding to the equipment with light corrosion is assigned to XS2, and the image corresponding to the equipment with heavy corrosion is assigned to XS3. The number of equipment with no corrosion, light corrosion, and heavy corrosion is counted and recorded as x1, x2, and x3, respectively. The crack situation of the device in the visible light image is obtained by using the Mask R-CNN algorithm; the image corresponding to the device with crack is assigned to LW, and the number of devices with crack in the image is counted l. Obtain the surface temperature distribution of the corresponding equipment in the regional information, and obtain the historical baseline temperature of the equipment from the database. Use a clustering algorithm to mark the equipment with a temperature exceeding the historical baseline temperature as overheated equipment. Assign a segment GR to the image containing overheated equipment, and count the number r of overheated equipment. Ultraviolet imaging is retrieved, and the insulator degradation value LH of the equipment is obtained by detecting the corona discharge spot using ultraviolet imaging technology, as well as the number h of degraded equipment in the corresponding image.

6. A drone inspection and management system suitable for smart power distribution networks according to claim 2, characterized in that, The step of establishing a partition update mechanism based on equipment operating conditions to obtain new graded regions includes: Retrieve equipment operating status and obtain all image scores based on the mapping relationship between equipment operating status and preset equipment operating coefficients; Get the maximum value among all image ratings and minimum value The image rating will be placed in [ , The image's corresponding rating region remains within its original rating region; the image rating is placed within [...]. , The region corresponding to the image is marked as a new level region with one increment from the original level.

7. A drone inspection and management system suitable for smart power distribution networks according to claim 2, characterized in that, The method of setting the inspection cycle according to the grade area includes: The average inspection time for each Level i area was obtained through pre-flight testing. Where i = 1, 2, 3; Through formula The inspection cycle of region i is calculated; where, As a preset base period, This represents the total number of devices in the inspection area. This represents the number of devices that malfunctioned in the inspection area.

8. A drone inspection and management system suitable for smart power distribution networks according to claim 2, characterized in that, The method of predicting the probability of equipment damage during the inspection and observation period after maintenance through formulas includes: Determine if the equipment being repaired is the same model as the equipment before repair; if yes, record the probability of damage during the inspection and observation period as 0; otherwise, obtain the accuracy of the corresponding equipment processing data. Retrieve the preset inspection observation period and divide it into N short cycles; construct an accuracy change curve with each short cycle as the independent variable and accuracy as the dependent variable; where N>3. Determine if the slope of the curve is less than 0; if yes, issue warning signal one and notify staff to handle the situation; if no, determine if the slope continues to increase; if yes, then use the formula... The probability of damage is calculated; otherwise, a second warning signal is issued; where A(t) is obtained by fitting the accuracy change curve through linear regression.

9. A drone inspection and management system for smart power distribution networks according to claim 1, characterized in that, The method of setting a stabilization period based on the probability of damage includes: retrieve damage probability To obtain the lifespan S of the corresponding equipment, use the formula Tw=S×[|1-ln( The stable period is calculated using the method described above.

10. A method for managing drone inspections in smart power distribution networks, applied to the drone inspection management system for smart power distribution networks as described in any one of claims 1-9, characterized in that, include: The substation is divided into basic zones to obtain several different levels of areas; The drone collects regional information at different levels using several sensors. Determine if any equipment in the substation is under maintenance; if yes, mark the area corresponding to the equipment under maintenance as the equipment maintenance area and stop the inspection of the equipment maintenance area; if no, analyze the area information using image analysis technology to obtain the equipment operation status corresponding to the area information; construct a zoning update mechanism based on the equipment operation status to obtain new level areas; set the inspection cycle based on the level areas; Set an inspection and observation period for the equipment that has been repaired; predict the probability of damage to the equipment that has been repaired within the preset inspection and observation period using a formula; determine whether the probability of damage is less than the preset stability threshold; if yes, set a stability period based on the probability of damage and stop inspections during the stability period; after the stability period ends, divide the equipment repair area into basic zones and re-acquire the inspection cycle. If not, a reminder signal will be sent to notify the staff.

Images