A UAV power inspection method and system
Through stochastic optimization algorithm and deep learning technology, the drone inspection points are adjusted, the route planning is optimized, and the neural network is used to integrate the inspection data, the problem of aviation route planning and data acquisition in drone power inspection is solved, and efficient Quality overhead line patrols and risk judgments.
Patent Information
- Application Number
- CN202410081332.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-01-19
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2044-01-19
AI Technical Summary
The existing drone power inspection technology is difficult to effectively plan the route, ignores the best task points of the inspection point, affects the normal driving and communication of the drone, and is difficult to combine the inspection process with the power inspection business to generate high Quality inspection report.
Using stochastic optimization algorithm combined with deep learning, adjust the inspection points based on the communication signal strength and image recognition results of the drone, optimize the route planning, ensure efficient inspection of the drone on overhead lines, and integrate multiple patrols using neural network models. Generate standardized reports for inspection task data.
It has realized efficient planning and high-quality data acquisition of drone inspection routes, ensured efficient inspection and risk judgment of overhead lines, met the high-quality development requirements of the power system for inspection, and reduced inspection costs.
Smart Images

Figure CN118331289B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of UAV power inspection, and specifically to a UAV power inspection method and system. Background Art
[0002] Overhead lines are the main equipment for daily power transmission. The quality inspection of overhead lines is related to the power demand of thousands of users. Since traditional manual inspection is time-consuming and labor-intensive, and because the overhead lines themselves have high voltage and partial discharge, which endangers life safety, the use of drones for power inspection has become the most common inspection method.
[0003] The key to the drone power inspection technology is to use drones to complete the route planning of overhead line inspections and use the overhead line data collected by drones to complete high-quality inspections and risk assessments of overhead lines. Existing inspection route planning mainly focuses on the improvement of route algorithms, such as traditional heuristic algorithms to various improved deep learning algorithms, etc., to achieve the optimal solution of multi-dimensional objective functions and constraints, while ignoring the optimization of the inspection point itself. Different inspection point settings will affect the normal driving and communication of drones due to obstacles, which will ultimately affect the overall planning of drone routes.
[0004] In addition, there are a large number of power equipment in overhead lines, and therefore there are various types of faults and risks. How to combine the drone inspection process with the power inspection business, string together the inspection work, inspection data collection work, and inspection results, and form a professional and standardized inspection report to meet the power system's high-quality development requirements for overhead line inspections is a technical problem that needs to be solved. Summary of the invention
[0005] In order to plan UAV inspection routes with high efficiency and quality, the present application provides a UAV power inspection method and system.
[0006] In a first aspect, the present invention provides a UAV power inspection method, comprising:
[0007] Based on the regular flight altitude of the drone, several inspection points were preliminarily confirmed along the overhead lines within the inspection area;
[0008] According to the communication signal strength and the returned image of the test flight drone at the location of the inspection point and the obstacle situation in the inspection point space buffer area established by the inspection point, it is judged whether each inspection point needs to be adjusted, and the displacement adjustment of the corresponding inspection point is completed according to the judgment result, and the adjusted inspection points and the inspection points that do not need to be adjusted are merged to generate the final confirmed inspection point set;
[0009] Establish the first constraint condition and the optimization target of specific inspection task allocation, and use the random optimization algorithm to obtain the optimal solution of the optimization target of specific inspection task allocation, that is, the inspection points to which each assignable drone that performs a specific inspection task is assigned; the first constraint condition includes: the task of each inspection point in the final confirmed inspection point set is only performed once, and a deployable drone that performs a specific inspection task is assigned to at least one inspection point in the final confirmed inspection point set; the optimization target of specific inspection task allocation includes: the total range of multiple deployable drones that perform specific inspection tasks to complete all specific inspection tasks is minimized;
[0010] Establish the second constraint condition and the path optimization goal, and use the random optimization algorithm to obtain the optimal solution of the path optimization goal, that is, the inspection path of each assignable UAV that performs a specific inspection task; the second constraint condition includes: the maximum flight distance constraint of a single deployable UAV that performs a specific inspection task; the path optimization goal includes: the total range of a single deployable UAV that performs a specific inspection task to execute all assigned inspection points and including the charging and round-trip range of the inspection process is minimized; obtain the inspection path of each assignable UAV that performs different specific inspection tasks.
[0011] By adopting the above scheme, the communication signal strength of the test flight drone, the returned image and the obstacle conditions in the spatial buffer area established by the inspection point are used to complete the adjustment and judgment of the preliminary inspection points, and the optimal setting of the inspection points is achieved to ensure the normal operation and communication of the drone inspection process, providing an important data acquisition basis for realizing high-quality inspections of overhead lines; considering the charging status of the drone, the maximum flight distance constraint of the drone is set and the random optimization algorithm is used to solve the drone inspection path with the shortest total range, so as to achieve the most efficient use and deployment of drones for overhead line inspections and save inspection costs.
[0012] Preferably, the step of determining whether each inspection point needs to be adjusted based on the communication signal strength and the returned image of the test flight drone at the location of the inspection point and the obstacle situation in the established inspection point space buffer area and completing the displacement adjustment of the corresponding inspection point based on the determination result includes:
[0013] Establish a corresponding inspection point space buffer area for each inspection point; obtain the communication signal strength and return image of the test flight drone at each inspection point;
[0014] Determine whether each inspection point meets the conditions that the communication signal strength of the test flight drone at the location of the inspection point is greater than the preset communication signal strength, the image transmitted by the test flight drone at the location of the inspection point identifies that there is an overhead line, and there are no obstacles in the inspection point space buffer area corresponding to the inspection point; if so, determine that the inspection point is an inspection point that does not need to be adjusted; otherwise, determine that the inspection point is an inspection point to be adjusted;
[0015] For each inspection point to be adjusted, vertical and / or horizontal displacement adjustments are continuously performed until the communication signal strength of the test flight UAV at the location of the adjusted inspection point is greater than the preset communication signal strength, there are no obstacles in the inspection point space buffer area corresponding to the adjusted inspection point, and overhead lines are identified in the image transmitted back from the adjusted inspection point.
[0016] By adopting the above scheme, the spatial buffer area of the inspection point that accommodates the safe flight space of the UAV is obtained based on the spatial analysis technology to judge whether there are obstacles in the spatial buffer area. The communication signal strength and return image of the test flight UAV at the inspection point are used to judge the communication quality and shooting quality of the UAV. According to the judgment result, the inspection point is adjusted preferentially to achieve high-quality inspection of overhead lines.
[0017] Preferably, the continuous adjustment of the vertical and / or horizontal displacement for each inspection point to be adjusted also includes: if the inspection point corresponding to a certain position of the overhead line has a height or distance requirement, the inspection point to be adjusted corresponding to a certain position of the current overhead line completes horizontal displacement under a fixed height setting or completes vertical and / or horizontal displacement under fixed distance conditions.
[0018] By adopting the above scheme, taking into account the requirements for inspection height and inspection distance for key components in overhead lines, when the inspection points are adjusted, it is ensured that the adjusted inspection points meet the corresponding requirements, thereby improving the accuracy of risk assessment results for key components in overhead lines.
[0019] Preferably, the conventional flight altitude of the UAV is used as a reference, and several inspection points are confirmed along the overhead lines in the inspection area, including: ordinary inspection points and key inspection points;
[0020] The common inspection points include inspection points corresponding to a single angle around the overhead line;
[0021] The key inspection points include inspection points corresponding to different angles around key components of the overhead line or historical damage locations of the overhead line.
[0022] By adopting the above scheme, taking into account that key components or historically damaged locations in overhead lines have inspection requirements from multiple angles, key inspection points are set up to conduct all-round inspections of key locations and historically damaged locations from multiple angles, thereby improving the accuracy of risk assessment results of key components in overhead lines.
[0023] Preferably, it also includes:
[0024] Marking the image data or audio data of the same position of the overhead line collected by the deployable drones that perform different specific inspection tasks with corresponding inspection tasks; the different specific inspection tasks include: defect detection tasks, temperature detection tasks, partial discharge detection tasks, and flaw detection tasks;
[0025] Merge image data or audio data belonging to the same position of the overhead line, and input the image data or audio data of the same position of the overhead line into the corresponding trained neural network model according to the corresponding detection task mark to obtain the detection result of the same position of the overhead line; the trained neural network model includes: a first neural network model that takes image data with defect detection task mark as input and outputs whether the overhead line has defects, a second neural network model that takes image data with temperature detection task mark as input and outputs whether the overhead line has temperature abnormality, a third neural network model that takes audio data with partial discharge detection task mark as input and outputs whether the overhead line has partial discharge, and a fourth neural network model that takes image data with flaw detection task mark as input and outputs whether the overhead line has scars;
[0026] Generate a test report based on the test results at different locations of the overhead line.
[0027] By adopting the above solution, considering that the normal flight of the UAV cannot carry the mounted modules of different inspection tasks (such as: infrared camera module, voiceprint partial discharge detection module, etc.), and the power consumption of the UAV flight is related to the weight of the UAV itself and the mounted objects, the more mounts, the greater the flight power consumption, which leads to an increase in the range of overhead inspection; therefore, the UAV inspection is combined with the overhead line inspection business, and the deployable UAVs that perform different specific inspection tasks are classified and set up. The deployable UAVs that perform different specific inspection tasks are used to complete the inspection path planning and inspection data acquisition of various inspection tasks, and the neural network algorithm is combined to realize the comprehensive fault risk judgment of the overhead line and generate standardized inspection reports, so as to meet the high-quality development requirements of the power system for overhead line inspections.
[0028] Preferably, the output of the first neural network model also includes a defect risk level, the output of the second neural network model also includes a temperature anomaly risk level, the output of the third neural network model also includes a partial discharge risk level, and the output of the fourth neural network model also includes a scar risk level;
[0029] The detection results for the specific location of the overhead line include: any one of the following: the existence of defects and a high risk level of the defect, temperature abnormality and a high risk level of the temperature abnormality, and the existence of partial discharge and a high risk level of the partial discharge. The specific location of the overhead line is set as an emergency processing location and an emergency processing prompt for the specific location of the overhead line is generated. The emergency processing prompt for the specific location of the overhead line and the detection results of the specific location of the overhead line are sent to the terminal of the fault repair personnel.
[0030] By adopting the above scheme, the neural network is used to obtain the fault risk level of the overhead line, emergency processing prompts are set according to the fault risk level of the overhead line, and problem feedback is given to high-risk locations of the overhead line in a timely manner.
[0031] Preferably, the number of image data or audio data belonging to the same position of the overhead line and having different marks for the same type of detection task is greater than or equal to a preset number;
[0032] Multiple image data or audio data with the same type of detection task labels belonging to the same position of the overhead line are input into the corresponding trained neural network model, and the detection results with the largest number of identical outputs are taken as the final detection results.
[0033] By adopting the above scheme, multiple inspections are performed on each inspection task at the same position of the overhead line to ensure the accuracy of the fault risk judgment of the overhead line.
[0034] In a second aspect, the present application provides a UAV power inspection system, comprising:
[0035] A module for preliminarily confirming several inspection points is used to preliminarily confirm several inspection points along the overhead lines in the inspection area based on the conventional flight altitude of the drone;
[0036] The final inspection point set acquisition module is used to determine whether each inspection point needs to be adjusted based on the communication signal strength and the returned image of the test flight drone at the location of the inspection point and the obstacle situation in the inspection point space buffer area established by the inspection point, and complete the displacement adjustment of the corresponding inspection point according to the judgment result, and merge the adjusted inspection points with the inspection points that do not need to be adjusted to generate the final confirmed inspection point set;
[0037] The UAV inspection point allocation module is used to establish the first constraint condition and the optimization target of the specific inspection task allocation, and use the random optimization algorithm to obtain the optimal solution of the optimization target of the specific inspection task allocation, that is, the inspection points to which each assignable UAV that performs a specific inspection task is assigned; the first constraint condition includes: the task of each inspection point in the final confirmed inspection point set is only executed once, and a deployable UAV that performs a specific inspection task is assigned to at least one inspection point in the final confirmed inspection point set; the optimization target of the specific inspection task allocation includes: the total flight distance of multiple deployable UAVs that perform specific inspection tasks to complete all specific inspection tasks is minimized;
[0038] The UAV inspection path acquisition module is used to establish the second constraint condition and the path optimization target, and use the random optimization algorithm to obtain the optimal solution of the path optimization target, that is, the inspection path of each assignable UAV that performs a specific inspection task; the second constraint condition includes: the maximum flight distance constraint of a single deployable UAV that performs a specific inspection task; the path optimization target includes: the total range of a single deployable UAV that performs a specific inspection task to execute all assigned inspection points and including the charging and round-trip range of the inspection process is minimized; obtain the inspection path of each assignable UAV that performs different specific inspection tasks.
[0039] By adopting the above scheme, the final inspection point set acquisition module is used to obtain the optimal inspection points to ensure high-quality inspection of overhead lines; the drone inspection point allocation module and the drone inspection path acquisition module are used to obtain the drone inspection path with the shortest total flight range to ensure efficient overhead line inspection.
[0040] In a third aspect, the present application provides a computer-readable storage medium, wherein the computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the method as described above.
[0041] In a fourth aspect, the present application provides a computer device, comprising a memory, a processor, and a program stored and executable on the memory, wherein the program implements the steps of the above method when executed by the processor.
[0042] In summary, this application has the following beneficial effects:
[0043] 1. The inspection points are selected based on the communication signal strength of the test flight drone at the inspection point, the image recognition results sent back by the test flight drone at the inspection point, and the obstacle conditions in the spatial buffer area established by the inspection point to ensure the normal operation and communication of the drone inspection process; the maximum flight distance constraint of the drone is set considering the charging condition of the drone, and the random optimization algorithm is selected to obtain the optimal drone inspection path to achieve efficient and high-quality drone inspection route planning; 2. The drone inspection is combined with the overhead line inspection business, and various inspection tasks are completed by using deployable drones that perform different specific inspection tasks, and multiple detection neural network algorithms are combined to comprehensively obtain the inspection results of different locations of the overhead lines; standardized inspection reports are generated based on the comprehensive fault risk judgment results of the overhead lines to meet the high-quality development requirements of the power system for overhead line inspections. BRIEF DESCRIPTION OF THE DRAWINGS
[0044] Figure 1 It is a flow chart of the optimal inspection path of overhead lines in the UAV power inspection method described in the specific embodiment;
[0045] Figure 2 It is a flow chart of obtaining the inspection results of overhead lines in the UAV power inspection method described in the specific embodiment;
[0046] Figure 3 It is an image of the same position of an overhead line collected by a deployable drone that performs different specific inspection tasks in the process of obtaining the inspection results of the overhead line in the drone power inspection method described in the specific embodiment; Figure 3 (a) with Figure 3 (b) Illustrations showing different angles of insulators in overhead lines captured by a drone performing defect inspection tasks; Figure 3 (c) shows an infrared image of an insulator in an overhead line captured by a drone performing a temperature inspection mission;
[0047] Figure 4 It is a diagram of a partial report of overhead line inspection results in the UAV power inspection method described in a specific embodiment; Figure 4 (a) is a diagram describing the defects of the cross arm end of the insulator on the small side of the middle phase; Figure 4 (b) is a diagram describing the defects of the cross arm ends of the small-sized insulators on the left and right mid-phases; Figure 4 (c) is a diagram describing the defects of the insulator conductor end at the small side of the middle phase;
[0048] Figure 5 It is a structural schematic diagram of the UAV power inspection system described in a specific embodiment. DETAILED DESCRIPTION
[0049] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.
[0050] like Figure 1 As shown, in order to obtain high-efficiency and high-quality UAV inspection route planning, the embodiment of the present application discloses a UAV power inspection method, which specifically includes:
[0051] S10. Preliminarily determine the inspection points of overhead lines.
[0052] Specifically, several inspection points are evenly spaced along the overhead lines in the inspection area. The X-axis and Y-axis data corresponding to each inspection point are set based on the X-axis and Y-axis data of several positions in the overhead lines, offset by preset deviation data, and the Z-axis data corresponding to each inspection point uses the conventional flight altitude of the drone.
[0053] Taking into account that overhead lines include transmission lines and transmission towers, and transmission towers include a variety of power equipment, when preliminarily determining a number of inspection points, multi-angle inspection requirements are set for key components and historically damaged components in the overhead lines; therefore, the inspection points of overhead lines include: ordinary inspection points and key inspection points: the ordinary inspection points include inspection points corresponding to a single angle around the overhead lines; the key inspection points include inspection points corresponding to different angles around key components of overhead lines or locations of historical damage to overhead lines.
[0054] S11. Adjust the initially determined inspection points to obtain the final inspection points of the overhead line.
[0055] Specifically, a spatial analysis technology is used to establish a corresponding inspection point spatial buffer area for each inspection point; wherein the size of the spatial buffer area is set based on the size of the drone and the preset deviation data of the offset.
[0056] The communication signal strength of the test drone flying to each inspection point and the image of the overhead line transmitted back by the test drone flying to each inspection point are obtained through the drone background operating system; the images of the overhead line transmitted back by the test drone flying to each inspection point are identified through deep learning algorithms and image recognition technology to determine whether there are overhead lines in the images of the overhead lines transmitted back by the test drone flying to each inspection point;
[0057] Determine whether each inspection point meets the conditions that the communication signal strength of the test flight drone at the location of the inspection point is greater than the preset communication signal strength, the image transmitted by the test flight drone at the location of the inspection point identifies that there is an overhead line, and there are no obstacles in the inspection point space buffer area corresponding to the inspection point; if so, determine that the inspection point is an inspection point that does not need to be adjusted; otherwise, determine that the inspection point is an inspection point to be adjusted;
[0058] Continuously adjust the vertical and / or horizontal displacement of each inspection point to be adjusted until the communication signal strength of the test flight drone at the location of the adjusted inspection point is greater than the preset communication signal strength, there are no obstacles in the spatial buffer area of the inspection point corresponding to the adjusted inspection point, and the overhead line is identified in the image sent back by the adjusted inspection point, and the adjustment is stopped; wherein, considering that the key components in the overhead line have inspection height and inspection distance requirements, to ensure that the adjusted inspection point meets the corresponding requirements, continuously adjust the vertical and / or horizontal displacement of each inspection point to be adjusted also includes: if the inspection point corresponding to a certain position of the overhead line has height or distance requirements, the inspection point to be adjusted corresponding to a certain position of the current overhead line completes horizontal displacement under a fixed height setting or completes vertical and / or horizontal displacement under fixed distance conditions.
[0059] The final confirmed inspection point set is generated by merging the adjusted inspection points with the inspection points that do not need to be adjusted.
[0060] S12: Obtain the assigned inspection point of each UAV that can be assigned to perform a specific inspection task.
[0061] Specifically, considering the diversity of overhead line inspection services and the fact that each inspection service covers almost all inspection points, the number of different specific inspection tasks is confirmed as the inspection point L in the final confirmed inspection point set = {L1, L2...L l}, determine the number of drones M that can be assigned to perform specific inspection tasks = {M1, M2...M m} and the corresponding location data, and the number of drone nests N in the inspection area that can be used for unlimited charging of drones = {N1, N2...N n} and the corresponding position data.
[0062] A first constraint condition is established, wherein the first constraint condition comprises:
[0063] The task of each inspection point in the final confirmed inspection point set is executed only once, that is:
[0064]
[0065] In the formula, i represents the i-th inspection task node, j represents the j-th inspection task node, and the inspection task nodes include the take-off nodes and inspection points of the drone, and the number is m+l; k represents the k-th drone that can be assigned to perform a specific inspection task; It represents the decision variable of the kth UAV from the i-th inspection task node to the j-th inspection task node. When it is 1, it means that the kth UAV is assigned to perform the inspection task from the i-th inspection task node to the j-th inspection task node. When it is 0, it means that the kth UAV is not assigned to perform the inspection task from the i-th inspection task node to the j-th inspection task node.
[0066] A deployable drone that performs a specific inspection task is assigned to at least one inspection point in the final confirmed inspection point set, namely:
[0067]
[0068] Establish a specific inspection task allocation optimization target, the function of which is:
[0069] Without considering the round trip distance of the drone for charging, only considering the total range of multiple drones that can be deployed to perform specific inspection tasks to complete all inspection tasks (that is, flying to each inspection point to complete the inspection task) is the minimum, that is:
[0070]
[0071] In the formula, represents the distance of the kth UAV from the i-th inspection task node to the j-th inspection task node, Indicates whether the UAV and the target node correspond to the decision variable. When it is 1, it corresponds, that is, it is not the round-trip distance for charging the UAV. When it is 0, it does not correspond, that is, it is the round-trip distance for charging the UAV.
[0072] The random optimization algorithm is used to obtain the optimal solution of the task allocation optimization objective, that is, the inspection points assigned to each drone that can perform a specific inspection task.
[0073] Specifically, in this embodiment, an improved Monte Carlo algorithm is selected to obtain the optimal solution for the task allocation optimization objective, and the steps include: randomly assigning inspection points to each deployable drone that performs a specific inspection task; evaluating the current allocation plan to obtain the overall total range distance according to a pre-set overall total range evaluation parameter; each drone randomly selects the inspection point to be exchanged and assigned by any drone within the search distance range according to the randomly generated search distance, and obtains a new allocation strategy; evaluating the new allocation plan to obtain the overall total range distance according to the pre-set overall total range evaluation parameter; determining whether to accept the new solution as the current solution according to certain probability rules, if the new solution is better than the current solution, directly accept it, otherwise, accept it with a certain probability; repeating the above steps until the limit on the number of iterations is reached or the running time exceeds the limit time, and outputting the final solution, which is the optimal task allocation plan.
[0074] S13. Obtain the optimal inspection path for each assignable drone that performs a specific inspection task.
[0075] Specifically, establish the second constraint:
[0076] The maximum flight distance constraint of a single deployable drone that performs a specific inspection mission is:
[0077]
[0078] Where D C represents the path length, D i-1,i represents the length from the i-1th inspection point to the i-th inspection point; a represents the number of inspection points and charging stations that can be assigned to a single drone that performs a specific inspection task; D max Indicates the flight distance allowed by the drone on a single full charge; x i-1,i It represents the decision quantity about whether the i-1th inspection point is a charging station or not. It takes 1 to indicate that the i-1th inspection point is not a charging station, and takes 0 to indicate that the i-1th inspection point is a charging station.
[0079] Establish the path optimization goal for each assigned drone to perform a specific inspection mission:
[0080] The total range of a single deployable drone that performs a specific inspection task is the smallest, including the total round trip range of all assigned inspection points and the total round trip range of the inspection process, that is:
[0081]
[0082] The random optimization algorithm is used to obtain the optimal solution of the path optimization target, that is, the inspection path of each assignable drone that performs a specific inspection task. As above, in this embodiment, the improved Monte Carlo algorithm is selected to obtain the optimal solution of the path optimization target.
[0083] S14. Obtain the inspection paths of all the drones that can be assigned to perform different specific inspection tasks.
[0084] Specifically, due to the diversity of overhead line inspection services, such as different specific inspection tasks including: defect detection tasks, temperature detection tasks, partial discharge detection tasks and flaw detection tasks; statistics are obtained for the inspection paths of each assignable drone that performs different specific inspection tasks.
[0085] In addition, in order to obtain a comprehensive and standardized overhead line fault detection report and meet the high-quality development requirements of the power system for overhead line inspections, such as Figure 2 As shown, a drone power inspection method disclosed in an embodiment of the present application also includes: S20, marking the image data or audio data of the same position of the overhead line collected by the drone that can be deployed to perform different specific inspection tasks with corresponding detection tasks.
[0086] Specifically, Figure 3 As shown in FIG. 1 , some of the image data of a specific position of an overhead line collected by a deployable drone that performs different specific inspection tasks are labeled; the drones that perform different specific inspection tasks are drones that include different mounting modules, such as: drones mounted with wide-angle and telephoto camera modules perform defect detection tasks, and the image data collected at a specific position of the overhead line is as shown in FIG. Figure 3 (a) with Figure 3 (b), denoted as P 缺检 ; The drone equipped with thermal imaging camera module or ultraviolet imaging detection module performs temperature detection task, and the image data collected at a specific position of the overhead line is as follows: Figure 3 (c), denoted as P 温检 ; The drone equipped with the voiceprint partial discharge detection module performs the partial discharge detection task, and the audio data collected at a specific location of the overhead line is recorded as S 局放检 ; The drone equipped with the flaw detection module performs the flaw detection task, and the image data collected at a specific position of the overhead line is recorded as P 伤检 .
[0087] S21. Merge the image data or audio data belonging to the same position of the overhead line and input the corresponding trained neural network model according to the corresponding detection task label to obtain the detection result of the same position of the overhead line.
[0088] Specifically, the trained neural network models include:
[0089] The first neural network model takes image data marked with defect detection tasks as input and outputs whether the overhead line has defects and the defect risk level; wherein defects include: optical cable breakage, missing pins at specific ends of insulators on each side, etc., and defect risk levels include: low-level risk, medium-level risk and high-level risk; the first neural network model is generated by training historical image data of overhead lines with various defects and defect type and defect risk level annotations.
[0090] The second neural network model takes image data marked with temperature detection tasks as input and outputs whether the temperature of the overhead line is abnormal and the abnormal risk level; wherein the temperature risk levels include: low-level risk, medium-level risk and high-level risk; the second neural network model is generated by training historical image data of overhead lines with abnormal temperatures and temperature anomaly annotations and temperature anomaly risk level annotations.
[0091] The third neural network model takes audio data marked with partial discharge detection tasks as input and outputs whether partial discharge exists in the overhead line and the partial discharge risk level; wherein the partial discharge risk levels include: low-level risk, medium-level risk and high-level risk; the third neural network model is generated by training historical audio signal data of overhead lines with various partial discharges and marked with partial discharge and partial discharge risk level.
[0092] The fourth neural network model takes image data marked with flaw detection tasks as input, and outputs whether the overhead line has flaws and the flaw risk level; wherein, flaws include: surface flaws of equipment, internal cracks and aging, etc., and flaw risk levels include: low-level risk, medium-level risk and high-level risk; the fourth neural network model is generated by training historical image data of overhead lines with various flaws and with flaw type annotations but no flaw risk level annotations.
[0093] S22. Generate a test report based on the test results at different locations of the overhead line.
[0094] like Figure 4 As shown, the above method can be used to obtain different positions of overhead lines (such as Figure 4 (a) Figure 4 (b) with Figure 4 (c)) generates a test report based on the test results, which comprehensively displays the inspection results of the overhead lines.
[0095] Considering that some overhead lines with high risk levels in the detection results need to be repaired in time, the method further includes:
[0096] The detection results for the specific location of the overhead line include: any one of the following: the existence of defects and a high risk level of the defect, temperature abnormality and a high risk level of the temperature abnormality, and the existence of partial discharge and a high risk level of the partial discharge. The specific location of the overhead line is set as an emergency processing location and an emergency processing prompt for the specific location of the overhead line is generated. The emergency processing prompt for the specific location of the overhead line and the detection results of the specific location of the overhead line are sent to the terminal of the fault repair personnel.
[0097] In order to obtain more accurate detection results, the method further includes:
[0098] The number of image data or audio data with different labels of the same type of detection task belonging to the same position of the overhead line is greater than or equal to a preset number; multiple image data or audio data with the same type of detection task labels belonging to the same position of the overhead line are input into the corresponding trained neural network model, and the detection result with the largest number of identical outputs is taken as the final detection result.
[0099] Considering the need for emergency overhead line inspection at night, in order to ensure the inspection is carried out, the method further includes:
[0100] Before executing steps S30 and S40 , a preliminary screening is performed on each assignable drone for performing a specific inspection task to determine whether the drone is equipped with a fill light device, and drones that are not equipped with a fill light device and can perform a specific inspection task are marked as unassignable.
[0101] like Figure 5 As shown, the embodiment of the present application discloses a UAV power inspection method, which specifically includes:
[0102] A module 101 for preliminarily confirming a number of inspection points is used to preliminarily confirm a number of inspection points along the overhead lines in the inspection area based on the conventional flight altitude of the UAV;
[0103] The final inspection point set acquisition module 102 is used to complete the judgment of whether each inspection point needs to be adjusted according to the communication signal strength and the returned image of the test flight drone at the location of the inspection point and the obstacle situation in the inspection point space buffer area established by the inspection point, and complete the displacement adjustment of the corresponding inspection point according to the judgment result, and merge the adjusted inspection points with the inspection points that do not need to be adjusted to generate the final confirmed inspection point set;
[0104] The drone inspection point allocation module 103 is used to establish the first constraint condition and the optimization target of the specific inspection task allocation, and use the random optimization algorithm to obtain the optimal solution of the optimization target of the specific inspection task allocation, that is, the inspection points to which each assignable drone that performs a specific inspection task is assigned; the first constraint condition includes: the task of each inspection point in the final confirmed inspection point set is only executed once, and a deployable drone that performs a specific inspection task is assigned to at least one inspection point in the final confirmed inspection point set; the optimization target of the specific inspection task allocation includes: the total range of multiple deployable drones that perform specific inspection tasks to complete all specific inspection tasks is minimized;
[0105] The UAV inspection path acquisition module 104 is used for the UAV inspection path acquisition module, which is used to establish a second constraint condition and a path optimization target, and use a random optimization algorithm to obtain the optimal solution of the path optimization target, that is, the inspection path of each assignable UAV that performs a specific inspection task; the second constraint condition includes: the maximum flight distance constraint of a single deployable UAV that performs a specific inspection task; the path optimization target includes: the total range of a single deployable UAV that performs a specific inspection task to execute all assigned inspection points and including the charging and round-trip range of the inspection process is minimized; obtain the inspection path of each assignable UAV that performs different specific inspection tasks.
[0106] In addition, the system disclosed in the embodiment of the present application also includes:
[0107] The overhead line detection report acquisition module 105 is used to mark the image data or audio data of the same position of the overhead line collected by the deployable drones that perform different specific inspection tasks with corresponding detection tasks; merge the image data or audio data belonging to the same position of the overhead line, and input the image data or audio data of the same position of the overhead line into the corresponding trained neural network model according to the corresponding detection task mark to obtain the detection result of the same position of the overhead line; the trained neural network model includes: a first neural network model that takes the image data with defect detection task mark as input and outputs whether the overhead line has defects, a second neural network model that takes the image data with temperature detection task mark as input and outputs whether the overhead line has temperature abnormality, a third neural network model that takes the audio data with partial discharge detection task mark as input and outputs whether the overhead line has partial discharge, and a fourth neural network model that takes the image data with flaw detection task mark as input and outputs whether the overhead line has scars; and generates a detection report according to the detection results of different positions of the overhead line.
[0108] The embodiment of the present application also discloses a computer-readable storage medium.
[0109] Specifically, the computer-readable storage medium stores a computer program that can be loaded by a processor and executed such as the above-mentioned UAV power inspection method. The computer-readable storage medium includes, for example: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and other media that can store program codes.
[0110] The embodiment of the present application also discloses a computer device.
[0111] Specifically, the computer device includes a memory and a processor, and the memory stores a computer program that can be loaded by the processor and execute the above-mentioned UAV power inspection method.
[0112] The above are all preferred embodiments of the present application, and are not intended to limit the protection scope of the present application. Any feature disclosed in this specification (including the abstract and drawings), unless otherwise stated, can be replaced by other equivalent or alternative features with similar purposes. That is, unless otherwise stated, each feature is only an example of a series of equivalent or similar features.
Claims
1. A UAV power inspection method, It is characterized in that include: Based on the regular flight altitude of the drone, several inspection points were preliminarily confirmed along the overhead lines within the inspection area; According to the communication signal strength and the returned image of the test flight drone at the location of the inspection point and the obstacle situation in the inspection point space buffer area established by the inspection point, it is judged whether each inspection point needs to be adjusted, and the displacement adjustment of the corresponding inspection point is completed according to the judgment result, and the adjusted inspection points and the inspection points that do not need to be adjusted are merged to generate the final confirmed inspection point set; Establish the first constraint condition and the optimization target of specific inspection task allocation, and use the random optimization algorithm to obtain the optimal solution of the optimization target of specific inspection task allocation, that is, the inspection points to which each assignable drone that performs a specific inspection task is assigned; The first constraint condition includes: the task of each inspection point in the final confirmed inspection point set is only performed once, and a deployable drone that performs a specific inspection task is assigned to at least one inspection point in the final confirmed inspection point set; the specific inspection task allocation optimization goal includes: the total flight distance of multiple deployable drones that perform specific inspection tasks to complete all specific inspection tasks is minimized; Establish the second constraint condition and the path optimization target, and use the random optimization algorithm to obtain the optimal solution of the path optimization target, that is, the inspection path of each assignable drone that performs a specific inspection task; the second constraint condition includes: the maximum flight distance constraint of a single deployable drone that performs a specific inspection task; the path optimization target includes: the total range of a single deployable drone that performs a specific inspection task to execute all assigned inspection points and including the charging round trip during the inspection process is minimized; Obtain the inspection paths of all the drones that can be assigned to perform different specific inspection tasks; The method of determining whether each inspection point needs to be adjusted based on the communication signal strength and the returned image of the test flight drone at the location of the inspection point and the obstacle situation in the inspection point space buffer area established by the inspection point and completing the displacement adjustment of the corresponding inspection point based on the determination result includes: Establish a corresponding inspection point space buffer area for each inspection point; obtain the communication signal strength and return image of the test flight drone at each inspection point; Determine whether each inspection point meets the conditions that the communication signal strength of the test flight drone at the location of the inspection point is greater than the preset communication signal strength, the image transmitted by the test flight drone at the location of the inspection point identifies that there is an overhead line, and there are no obstacles in the inspection point space buffer area corresponding to the inspection point; if so, determine that the inspection point is an inspection point that does not need to be adjusted; otherwise, determine that the inspection point is an inspection point to be adjusted; Continuously adjusting the vertical and / or horizontal displacement of each inspection point to be adjusted until the communication signal strength of the test flight drone at the location of the adjusted inspection point is greater than the preset communication signal strength, there are no obstacles in the inspection point space buffer area corresponding to the adjusted inspection point, and an overhead line is identified in the image sent back by the adjusted inspection point; the continuous adjustment of the vertical and / or horizontal displacement of each inspection point to be adjusted also includes: If the inspection point corresponding to a certain position of the overhead line has a height or distance requirement, the inspection point to be adjusted corresponding to a certain position of the current overhead line completes horizontal displacement under a fixed height setting or completes vertical and / or horizontal displacement under a fixed distance condition; Also includes: Marking the image data or audio data of the same position of the overhead line collected by the deployable drones that perform different specific inspection tasks with corresponding inspection tasks; the different specific inspection tasks include: defect detection tasks, temperature detection tasks, partial discharge detection tasks, and flaw detection tasks; Merge image data or audio data belonging to the same position of the overhead line, and input the image data or audio data of the same position of the overhead line into the corresponding trained neural network model according to the corresponding detection task mark to obtain the detection result of the same position of the overhead line; the trained neural network model includes: a first neural network model that takes image data with defect detection task mark as input and outputs whether the overhead line has defects, a second neural network model that takes image data with temperature detection task mark as input and outputs whether the temperature of the overhead line is abnormal, a third neural network model that takes audio data with partial discharge detection task mark as input and outputs whether the overhead line has partial discharge, and a fourth neural network model that takes image data with flaw detection task mark as input and outputs whether the overhead line has scars; Generate a test report based on the test results at different locations of the overhead line.
2. The UAV power inspection method according to claim 1, It is characterized in that The conventional flight altitude of the UAV is used as a reference to confirm several inspection points along the overhead lines in the inspection area, including ordinary inspection points and key inspection points; The common inspection points include inspection points corresponding to a single angle around the overhead line; The key inspection points include inspection points corresponding to different angles around key components of the overhead line or historical damage locations of the overhead line.
3. The UAV power inspection method according to claim 1, It is characterized in that The output of the first neural network model also includes a defect risk level, the output of the second neural network model also includes a temperature anomaly risk level, the output of the third neural network model also includes a partial discharge risk level, and the output of the fourth neural network model also includes a scar risk level; The detection results for the specific location of the overhead line include: any one of the following: the existence of defects and a high risk level of the defect, temperature abnormality and a high risk level of the temperature abnormality, and the existence of partial discharge and a high risk level of the partial discharge. The specific location of the overhead line is set as an emergency processing location and an emergency processing prompt for the specific location of the overhead line is generated. The emergency processing prompt for the specific location of the overhead line and the detection results of the specific location of the overhead line are sent to the terminal of the fault repair personnel.
4. The UAV power inspection method according to claim 1, It is characterized in that The number of image data or audio data belonging to the same position of the overhead line and having different marks for the same type of detection task is greater than or equal to a preset number; Multiple image data or audio data with the same type of detection task label belonging to the same position of the overhead line are input into the corresponding trained neural network model, and the detection results with the largest number of identical outputs are taken as the final detection results.
5. A UAV power inspection system, It is characterized in that include: A module for preliminarily confirming several inspection points is used to preliminarily confirm several inspection points along the overhead lines in the inspection area based on the conventional flight altitude of the drone; The final inspection point set acquisition module is used to determine whether each inspection point needs to be adjusted based on the communication signal strength and the returned image of the test flight drone at the location of the inspection point and the obstacle situation in the inspection point space buffer area established by the inspection point, and complete the displacement adjustment of the corresponding inspection point according to the judgment result, and merge the adjusted inspection points with the inspection points that do not need to be adjusted to generate the final confirmed inspection point set; The method of completing the judgment of whether each inspection point needs to be adjusted according to the communication signal strength and the returned image of the test flight drone at the location of the inspection point and the obstacle situation in the inspection point space buffer area established by the inspection point and completing the displacement adjustment of the corresponding inspection point according to the judgment result includes: establishing a corresponding inspection point space buffer area for each inspection point; obtaining the communication signal strength and the returned image of the test flight drone at each inspection point; judging whether each inspection point satisfies the conditions that the communication signal strength of the test flight drone at the location of the inspection point is greater than the preset communication signal strength, the returned image of the test flight drone at the location of the inspection point identifies that there is an overhead line, and there is no obstacle in the inspection point space buffer area corresponding to the inspection point; if so, the inspection point is judged to be an inspection point that does not need to be adjusted; otherwise, the inspection point is judged to be an inspection point to be adjusted; Continuously adjusting the vertical and / or horizontal displacement of each inspection point to be adjusted until the communication signal strength of the test flight drone at the location of the adjusted inspection point is greater than the preset communication signal strength, there are no obstacles in the inspection point space buffer area corresponding to the adjusted inspection point, and the overhead line is recognized in the image sent back by the adjusted inspection point; the continuous adjustment of the vertical and / or horizontal displacement of each inspection point to be adjusted also includes: if the inspection point corresponding to a certain position of the overhead line has a height or distance requirement, the inspection point to be adjusted corresponding to a certain position of the current overhead line completes the horizontal displacement under a fixed height setting or completes the vertical and / or horizontal position under a fixed distance condition; The UAV inspection point allocation module is used to establish the first constraint condition and the optimization target of the specific inspection task allocation, and use the random optimization algorithm to obtain the optimal solution of the optimization target of the specific inspection task allocation, that is, the inspection points to which each assignable UAV that performs a specific inspection task is assigned; the first constraint condition includes: the task of each inspection point in the final confirmed inspection point set is only executed once, and a deployable UAV that performs a specific inspection task is assigned to at least one inspection point in the final confirmed inspection point set; the optimization target of the specific inspection task allocation includes: the total flight distance of multiple deployable UAVs that perform specific inspection tasks to complete all specific inspection tasks is minimized; The UAV inspection path acquisition module is used to establish the second constraint condition and the path optimization target, and use the random optimization algorithm to obtain the optimal solution of the path optimization target, that is, the inspection path of each assignable UAV that performs a specific inspection task; the second constraint condition includes: the maximum flight distance constraint of a single deployable UAV that performs a specific inspection task; the path optimization target includes: the total range of all assigned inspection points and the round trip charging process of a single deployable UAV that performs a specific inspection task is minimized; obtain the inspection path of each assignable UAV that performs different specific inspection tasks; An overhead line detection report acquisition module is used to mark the image data or audio data of the same position of the overhead line collected by an adjustable drone that performs different specific inspection tasks with corresponding detection tasks; merge the image data or audio data belonging to the same position of the overhead line, and input the image data or audio data of the same position of the overhead line into the corresponding trained neural network model according to the corresponding detection task mark to obtain the detection result of the same position of the overhead line; the trained neural network model includes: a first neural network model that takes image data with defect detection task marks as input and outputs whether the overhead line has defects, a second neural network model that takes image data with temperature detection task marks as input and outputs whether the overhead line has temperature abnormalities, a third neural network model that takes audio data with partial discharge detection task marks as input and outputs whether the overhead line has partial discharge, and a fourth neural network model that takes image data with flaw detection task marks as input and outputs whether the overhead line has scars; and a detection report is generated according to the detection results of different positions of the overhead line.
6. A computer-readable storage medium, It is characterized in that The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the method according to any one of claims 1 to 4.
7. A computer device, It is characterized in that The computer device comprises a memory, a processor and a program stored and executable on the memory, and the program implements the steps of the method according to any one of claims 1 to 4 when executed by the processor.
Citation Information
Patent Citations
Power line unmanned aerial vehicle intelligent inspection practical operation simulation training method and system
CN110880258A
Power line unmanned aerial vehicle intelligent inspection multi-dimensional virtual scene construction method and system
CN111009164A
Unmanned aerial vehicle inspection path planning method based on random optimization
CN111536979A
Electric unmanned aerial vehicle patrol inspection path planning method
CN112013907A
Indication method for flight space information of unmanned aerial vehicle and equipment
CN112309175A
