Unmanned aerial vehicle inspection planning method and system for power fault feature perception

Abstract

The invention relates to an unmanned aerial vehicle inspection planning method and system for power fault feature perception, and relates to the field of route planning, and the method comprises the steps: obtaining data when a target power tower breaks down, dynamically adjusting a preset safety flight fence, marking key points, building an inspection planning constraint, and randomly planning an initial inspection route. According to the technical scheme of the invention, the technical problem that the unmanned aerial vehicle cannot effectively perceive the fault and dynamically adjust the inspection route and the safe flight fence during the power inspection of the unmanned aerial vehicle, resulting in insufficient inspection safety and effectiveness, is solved. The dynamic adaptation of the unmanned aerial vehicle inspection route to the fault is realized, and the inspection safety and effectiveness are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of route planning, in particular to a method and system for unmanned aerial vehicle inspection planning based on power failure feature perception. BACKGROUND

[0002] With the rapid development of power systems and the continuous expansion of power grid scale, the safe and stable operation of power towers, as important support structures for transmission lines, is crucial to ensuring the reliability of the entire power system. However, due to various factors such as natural environment, equipment aging, human operation, etc., power towers are inevitably subject to various failures during operation, such as tower deformation, component damage, etc. If these failures are not discovered and addressed in a timely manner, they may lead to more serious power accidents, even causing widespread power outages, and having a significant impact on society and people's lives.

[0003] Traditional power tower inspection methods mainly rely on manual periodic inspection, which not only has high labor intensity and low efficiency, but also is greatly limited by natural conditions such as weather and terrain, making it difficult to achieve comprehensive and timely inspection. With the rapid development of unmanned aerial vehicle technology, unmanned aerial vehicle inspection has gradually been applied to power tower inspection work due to its advantages of high efficiency, flexibility, and low cost. Unmanned aerial vehicle inspection can quickly cover large areas and perform high-altitude, multi-angle shooting and detection of towers, effectively improving inspection efficiency and accuracy.

[0004] However, existing unmanned aerial vehicle inspection technology still mostly relies on pre-set routes or manual control, lacking the ability to perceive and dynamically adjust to power tower failures. After a power tower failure occurs, such as tower deformation, its safety distance will change. The existing pre-set route inspection method cannot adapt to this change, resulting in insufficient safety distance between the unmanned aerial vehicle and the failed tower during inspection, which may cause safety accidents. At the same time, due to the lack of effective perception of failures, the unmanned aerial vehicle cannot accurately identify the failed components and locations, resulting in decreased inspection quality and inability to timely discover and address potential safety hazards. SUMMARY

[0005] The present application provides a method and system for unmanned aerial vehicle inspection planning based on power failure feature perception to solve the technical problem of insufficient safety and effectiveness of existing unmanned aerial vehicle power inspection due to the inability to effectively perceive failures and dynamically adjust inspection routes and safe flight fences.

[0006] The technical solution of the present application to solve the above technical problems is as follows: In a first aspect, the present application provides a method for unmanned aerial vehicle (UAV) inspection planning based on power failure characteristics, which comprises: obtaining failure data of a target power tower when a power failure occurs, adjusting a preset safe flight fence, obtaining an adjusted safe flight fence sequence, and marking a plurality of key points; constructing an inspection planning constraint according to the adjusted safe flight fence sequence, and randomly planning a first inspection route for the target power tower, wherein the first inspection route comprises a plurality of first stopping points; calculating and analyzing a first inspection fitness of the first inspection route according to the plurality of key points and the plurality of first stopping points; optimizing the inspection route to obtain an optimal inspection route, and controlling the UAV to perform inspection of the target power tower.

[0007] In a second aspect, the present application provides a system for UAV inspection planning based on power failure characteristics, which comprises: a data acquisition module for obtaining failure data of a target power tower when a power failure occurs, adjusting a preset safe flight fence, obtaining an adjusted safe flight fence sequence, and marking a plurality of key points; a route planning module for constructing an inspection planning constraint according to the adjusted safe flight fence sequence, and randomly planning a first inspection route for the target power tower, wherein the first inspection route comprises a plurality of first stopping points; a fitness calculation module for calculating and analyzing a first inspection fitness of the first inspection route according to the plurality of key points and the plurality of first stopping points; and an inspection control module for optimizing the inspection route to obtain an optimal inspection route, and controlling the UAV to perform inspection of the target power tower.

[0008] The present application has the following advantages: by obtaining data of a target power tower when a failure occurs, dynamically adjusting a preset safe flight fence and marking key points, constructing an inspection planning constraint, randomly planning an initial inspection route, calculating an inspection fitness based on key points and stopping points, and then optimizing an optimal inspection route to control the UAV to perform inspection, the UAV inspection route dynamically adapts to the failure characteristics of the power tower, and the safety and effectiveness of the inspection are improved. BRIEF DESCRIPTION OF DRAWINGS

[0009] Figure 1 A flowchart of a method for UAV inspection planning based on power failure characteristics provided by the present application.

[0010] Figure 2 A structural diagram of a system for UAV inspection planning based on power failure characteristics provided by the present application.

[0011] Reference signs: data acquisition module 11, route planning module 12, fitness calculation module 13, and inspection control module 14. DETAILED DESCRIPTION

[0012] The technical solutions in the embodiments of the present application will be clearly and completely described with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of the present application.

[0013] In the description of the present application, the terms "first", "second" are used only for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first", "second" can explicitly or implicitly include one or more of the features. In the description of the present application, the meaning of "multiple" is two or more, unless otherwise specifically limited.

[0014] In the description of the present application, the term "for example" is used to indicate "as an example, illustration or explanation". Any embodiment described as "for example" in the present application is not necessarily interpreted as more preferred or more advantageous than other embodiments. The following description is given in order to enable any person skilled in the art to implement and use the present application. In the following description, details are listed for the purpose of explanation. It should be understood that those skilled in the art can recognize that the present application can be implemented without using these specific details. In other examples, well-known structures and processes will not be described in detail to avoid unnecessary details making the description of the present application obscure. Therefore, the present application is not intended to be limited to the shown embodiments, but is consistent with the broadest scope in accordance with the principles and characteristics disclosed.

[0015] Embodiment one: As Figure 1 shown, the embodiment of the present application provides a power failure feature perception unmanned aerial vehicle inspection planning method, applied to a power failure feature perception unmanned aerial vehicle inspection planning system, the method comprises: S10: obtaining the failure data when the target power tower occurs power failure, integrating and adjusting the preset safe flight fence to obtain the adjusted safe flight fence sequence, and marking a plurality of key points.

[0016] Illustratively, the power tower is an important component in the power transmission line, mainly used for supporting and fixing the power transmission conductor to ensure that the power can be safely and stably transmitted. The power tower is usually made of steel or concrete and other materials, which has enough strength and stability to withstand the action of conductor, wind, ice and snow and other natural forces. At the same time, the power tower is also arranged and set up by reasonable layout and height to ensure that the conductor and the ground maintain enough safety distance to prevent electrical accidents.

[0017] In the scenario of UAV power inspection, the safe flight fence refers to a virtual or actual flight boundary set in advance to ensure that the UAV does not collide with the power tower and its attached facilities such as wires and insulators during the inspection process. This boundary is usually set based on the actual location, height, shape of the power tower and the surrounding environment. The main role of the safe flight fence is to ensure the safety of the UAV during the inspection process, prevent it from flying into dangerous areas and colliding with power facilities, causing equipment damage or UAV crashes, etc. By setting a safe flight fence, the UAV can be guided to follow a predetermined route for inspection, avoiding unnecessary flight and hovering, thereby improving inspection efficiency.

[0018] Specifically, in this scheme, during the power tower inspection process, when the target power tower fails, the relevant fault data is first obtained, including fault type, fault location, fault severity and other key feature information. Then, based on these fault data, the pre-set safe flight fence around the power tower is integrated and adjusted. The pre-set safe flight fence is essentially a safe flight range constructed around the target tower, which can be divided by constructing a coordinate system around the target tower. The UAV flying outside the pre-set safe flight fence can ensure that it maintains a safe distance from the power tower and its attached facilities, thereby ensuring flight safety. The integrated adjustment is to dynamically change the boundary, shape and other parameters of the pre-set safe flight fence according to the fault data to adapt to the inspection requirements under fault conditions, and ultimately obtain the adjusted safe flight fence sequence.

[0019] Among them, the adjusted safe flight fence sequence refers to a series of flight fence related content with specific order and rules for ensuring flight safety after modification, optimization, etc. The sequence indicates that the setting, parameters, etc. of these flight fences are arranged or combined in a certain order. After adjustment, it means that the original safe flight fence sequence has been changed according to new requirements, environmental changes, etc.

[0020] At the same time, the adjusted safe flight fence is further analyzed and a plurality of key points are marked. These key points are usually the coordinates of the largest deformation of the tower after failure, such as the vertex coordinates of the tower due to failure. The key points often correspond to important positions of the tower failure, providing an important basis for subsequent inspection route planning and fault analysis. For example, if a power tower is tilted due to strong winds, the system will adjust the safe flight fence accordingly to avoid the tilted area and mark the vertex coordinates of the most severe tilt as a key point so that the UAV can focus on inspection.

[0021] S20: According to the adjusted safety flight fence sequence, a first inspection route for inspecting the target power tower is randomly planned, wherein the first inspection route comprises a plurality of first stopping points.

[0022] Further, after obtaining the adjusted safety flight fence sequence of the target power tower, the fence sequences are union processed, that is, the adjusted safety flight fence regions are merged to form a total safety flight fence. Then, based on the total safety flight fence, an inspection planning constraint is constructed to stipulate that the unmanned aerial vehicle must fly outside the total safety flight fence, thereby ensuring the flight safety of the unmanned aerial vehicle during the entire inspection process.

[0023] Subsequently, a first inspection route for inspecting the target power tower is randomly planned. The route is not fixed and is generated based on a certain random algorithm, aiming to explore different inspection path possibilities. In the planning process, the structural characteristics and fault features of the target power tower are fully considered, and a plurality of first stopping points are reasonably set. These first stopping points are key nodes on the inspection route, and the unmanned aerial vehicle will hover when flying to these stopping points in order to use the image acquisition device to collect detailed images of the power tower. For example, if a certain area of the target power tower is deformed due to a fault, the system will deliberately set a stopping point near the area when randomly planning the inspection route, so that the unmanned aerial vehicle can hover and collect images of the area, providing important basis for subsequent fault analysis and processing.

[0024] S30: According to the plurality of key points and the plurality of first stopping points, a first inspection fitness of the first inspection route is calculated and analyzed.

[0025] In detail, after completing the first inspection route planning of the target power tower and determining the plurality of first stopping points, the first inspection route is calculated and analyzed in combination with the plurality of labeled key points to evaluate its first inspection fitness. The fitness here is a comprehensive index, which considers multiple aspects, and the distance between the key points and the first stopping points is one of the important evaluation factors.

[0026] Specifically, the distance between each first stopping point and the corresponding key point is calculated and compared with the preset standard detection distance. Ideally, the closer the distance between the stopping point and the key point to the standard detection distance, the more accurate the unmanned aerial vehicle can approach the fault or important detection position during the inspection, thereby collecting clearer and more effective image data, and the detection quality is higher. For example, if the standard detection distance is set to 5 meters, when the distance between a certain first stopping point and the corresponding key point is exactly 5 meters, it means that the unmanned aerial vehicle can hover at the stopping point to collect images of the area where the key point is located at the best distance, and the detection quality is more ideal.

[0027] In addition to distance, other factors that may affect inspection efficiency and quality are also considered, such as the total length of the inspection route and flight time. After evaluating the adaptability of the first inspection route, if it is found that the inspection time is too long or there is room for optimization, the system will make optimization adjustments. By replanning the route and adjusting the location of stop points, the inspection time will be reduced, thereby improving overall inspection efficiency. This ensures more efficient and accurate power tower inspections while ensuring inspection quality.

[0028] S40: Optimize the inspection route to obtain the optimal inspection route, and control the drone to inspect the target power tower.

[0029] Specifically, after calculating and analyzing the fitness of the first inspection route, the route is further optimized based on the fitness evaluation results. This optimization process involves comprehensively adjusting the location and sequence of multiple first stop points within the first inspection route, as well as the inspection path. Using intelligent algorithms such as genetic algorithms and particle swarm optimization, the team explores route solutions that can further improve inspection efficiency and quality while meeting inspection planning constraints (such as safe flight fence restrictions).

[0030] Specific optimization goals include shortening the total inspection time, reducing the drone's flight distance, and ensuring the rational distribution of stopover points near key points. For example, if a certain inspection route is redundant or detours, the system will trim or replan it to allow the drone to reach the next stop more directly. If there are multiple stopover points near a key point but the distances are unevenly distributed, the stopover points will be adjusted to more evenly cover the area where the key point is located, thereby improving inspection quality.

[0031] After multiple rounds of iterative optimization, the optimal inspection route was ultimately determined. This route, while meeting flight safety constraints, enables the inspection of the target power tower to be completed in the shortest possible time and with the highest efficiency. The system then sends control commands to the drone, guiding it to execute the inspection along the optimal inspection route. This ensures that the drone accurately reaches each stop during flight, completing a comprehensive and efficient inspection of the target power tower.

[0032] In a preferred embodiment, fault data of a target power tower when a power fault occurs is obtained, and a preset safety flight fence is integrated and adjusted to obtain an adjusted safety flight fence sequence, and multiple key points are marked and obtained, including: Fault data when a power fault occurs on a target power tower is acquired, wherein the fault data includes power fault data and environmental data.

[0033] A preset safe flight fence of the target power tower is obtained, wherein the preset safe flight fence includes safe flight boundary coordinates.

[0034] The fault data and the preset safe flight fence are combined and input into a plurality of safe flight fence generators respectively to generate a plurality of adjusted safe flight fences as an adjusted safe flight fence sequence.

[0035] Key points are marked in the plurality of adjusted safe flight fences in the adjusted safe flight fence sequence to obtain a plurality of key points.

[0036] Optionally, when the target power tower has a power failure, a data acquisition process is started to collect data related to the failure, which includes power failure data and environmental data. The power failure data includes failure type (such as short circuit, open circuit, etc.), time of failure, abnormal changes in failure current or voltage, and other characteristic information; the environmental data involves weather conditions (such as wind speed, wind direction, rainfall, etc.), temperature, humidity, and other factors that may affect the state of the unmanned aerial vehicle and the power tower.

[0037] At the same time, the preset safe flight fence information of the target power tower is obtained, which is defined by safe flight boundary coordinates. These coordinates clearly define the safe flight range that the unmanned aerial vehicle should maintain under normal circumstances to ensure that the unmanned aerial vehicle does not collide with the power tower and its attached facilities during the inspection process.

[0038] The system combines the collected failure data with the preset safe flight fence information and inputs them into a plurality of safe flight fence generators. The safe flight fence generators are based on different algorithms or models to process and analyze the input data and generate a plurality of adjusted safe flight fences, which constitute an adjusted safe flight fence sequence. Each adjusted safe flight fence is adjusted according to the failure data and environmental data to adapt to the inspection requirements under the failure condition.

[0039] Finally, key points are marked in the plurality of adjusted safe flight fences in the adjusted safe flight fence sequence. These key points are usually the positions where the tower has a large deformation after the failure, the failure characteristics are obvious, or the areas that need to be focused on and data collected during the inspection process. For example, if the failure causes a serious tilt in a part of the tower, the system will mark the vertex or key turning point of the tilted part as a key point so that the unmanned aerial vehicle can accurately position and collect relevant images or data during the inspection to provide strong support for subsequent failure analysis and processing.

[0040] In a preferred embodiment, the construction step of the plurality of safe flight fence generators includes: In the inspection data of the power failure of the power tower, a sample preset safe flight fence set and a sample fault data set are collected, and a sample adjusted safe flight fence after different sample fault data is constructed to obtain a sample adjusted safe flight fence set.

[0041] The data in the sample preset safe flight fence set, the sample fault data set and the sample adjusted safe flight fence set is divided by a preset proportion to obtain first safe flight fence generation training data, and the division is continued to obtain Kth safe flight fence generation training data, K being a positive integer.

[0042] A network structure of K safe flight fence generators is constructed by using deep learning.

[0043] The network parameter training and optimization of the K safe flight fence generators are performed in sequence by using the first safe flight fence generation training data to the Kth safe flight fence generation training data, and the construction is completed after the test convergence.

[0044] Specifically, in the process of constructing multiple safe flight fence generators, a key sample set needs to be collected from the inspection data of the power failure of the power tower. A sample preset safe flight fence set is collected, which contains safe flight boundary coordinate information preset for the power tower in different scenarios. Meanwhile, a sample fault data set is collected, which covers various types of power failure, environmental parameters and other data when the failure occurs.

[0045] Further, based on the sample preset safe flight fence and the sample fault data, a sample adjusted safe flight fence corresponding to different sample fault data is further constructed, for example, when the failure type is short circuit and accompanied by strong wind environment, the preset safe flight fence is adjusted according to the failure characteristics and environmental factors to obtain the corresponding sample adjusted safe flight fence, and finally a sample adjusted safe flight fence set is formed.

[0046] Subsequently, the sample preset safe flight fence set, the sample fault data set and the sample adjusted safe flight fence set are divided. According to a preset proportion, for example, 70% of the data is used as training data, and the corresponding part is extracted from the three sets to construct first safe flight fence generation training data. By continuing to divide according to different proportions, second safe flight fence generation training data can be obtained, and until Kth safe flight fence generation training data (K being a positive integer) is obtained. This way of dividing data in batches helps to gradually optimize the performance of the generator.

[0047] After the data is prepared, a network structure of K safe flight fence generators is constructed using deep learning technology. The deep learning model can handle complex nonlinear relationships and is suitable for generating adjusted safe flight fences according to fault data and preset safe flight fences. The network parameters of the K safe flight fence generators are trained and optimized in turn using the first safe flight fence generation training data to the Kth safe flight fence generation training data. During the training process, the network parameters are continuously adjusted so that the generator can accurately generate adjusted safe flight fences that conform to the actual situation according to the input preset safe flight fences and fault data. When the generator reaches a state of convergence on the test data, that is, the output result is stable and meets expectations, it indicates that the K safe flight fence generators have been constructed and can be put into practical application. For example, when facing new power tower faults in the future, these generators can quickly generate accurate adjusted safe flight fences according to real-time fault data and preset safe flight fences, providing safety protection for unmanned aerial vehicle inspection.

[0048] In a preferred embodiment, key points are labeled in the plurality of adjusted safe flight fences in the sequence of adjusted safe flight fences, and a plurality of key points are obtained, including: In each adjusted safe flight fence, the distance between each adjusted safe flight boundary coordinate and the preset safe flight fence is calculated, and the adjusted safe flight boundary coordinate with the maximum distance is selected as a key point.

[0049] The plurality of key points of the plurality of adjusted safe flight fences in the sequence of adjusted safe flight fences are obtained through calculation and processing.

[0050] In detail, in the process of key point labeling for multiple adjusted safety flight fences in the adjusted safety flight fence sequence to obtain multiple key points, for each adjusted safety flight fence in the adjusted safety flight fence sequence, calculation work needs to be carried out. The specific calculation content is the distance between each adjusted safety flight boundary coordinate in the adjusted safety flight fence and the preset safety flight fence. For example, assuming that there are three adjusted safety flight boundary coordinates A, B and C in an adjusted safety flight fence, the distance between A and the preset safety flight fence, the distance between B and the preset safety flight fence, and the distance between C and the preset safety flight fence need to be calculated respectively. After the distance calculation is completed, the adjusted safety flight boundary coordinate with the maximum distance is selected from the multiple distances calculated, and it is determined as the key point of the adjusted safety flight fence. The key point with the maximum distance is the position point with the maximum deformation amplitude in the adjusted safety flight fence after the failure deformation of the tower, so that the position with the maximum deformation of the tower can be easily observed during the inspection, and the effectiveness of the inspection observation is improved. In the same way, the above operation is sequentially performed on each adjusted safety flight fence in the adjusted safety flight fence sequence, and finally the multiple key points corresponding to the multiple adjusted safety flight fences in the adjusted safety flight fence sequence are calculated and obtained.

[0051] In a preferred embodiment, a first inspection route for inspecting the target power tower is randomly planned according to the adjusted safety flight fence sequence, comprising: The union set of the adjusted safety flight fence sequence is obtained to obtain a fusion safety flight fence, and an inspection planning constraint is constructed, wherein the inspection planning constraint includes that the inspection route does not enter the fusion safety flight fence.

[0052] A first inspection route satisfying the inspection planning constraint is randomly planned, wherein the first inspection route is used to inspect the target power tower and includes multiple first stopping points.

[0053] Further, a union set operation is performed on the obtained adjusted safety flight fence sequence. For example, assuming that there are three fences in the adjusted safety flight fence sequence, namely fence 1, fence 2 and fence 3, the boundary ranges of the three fences are integrated, and after removing the overlapping parts, a unified fusion safety flight fence covering all fence boundaries is obtained.

[0054] Further, an inspection planning constraint is constructed based on the fusion safety flight fence, which clearly stipulates that the inspection route cannot enter the area defined by the fusion safety flight fence, which is an important prerequisite for ensuring the safety of the inspection flight and avoiding entering dangerous or restricted areas.

[0055] Subsequently, a first inspection route for inspecting the target power towers is generated by using a stochastic programming algorithm under the condition of meeting the inspection planning constraints. For example, the target power towers are distributed in a certain area, and the stochastic programming algorithm considers factors such as the positions of the towers and the fusion safety flight fence boundary to plan a route that meets the constraints and effectively covers the target power towers. The first inspection route includes a plurality of first stopping points, which can be set according to the actual distribution of the power towers and the inspection requirements to ensure that each target power tower can be fully inspected.

[0056] The significance of the stochastic route planning lies in that in the face of uncertain factors such as the complex distribution of target power towers in the inspection planning and the irregular fusion safety flight fence boundary, a first inspection route that meets the actual requirements can be quickly generated by efficiently exploring the feasible solution space under the premise of meeting the inspection planning constraints (i.e., the inspection route does not enter the fusion safety flight fence) through random sampling and algorithm iteration, avoiding being trapped in a local optimal solution, effectively balancing the calculation efficiency and the quality of the solution, and providing a reliable foundation for the efficient and safe execution of subsequent inspection tasks.

[0057] In a preferred embodiment, the first inspection fitness of the first inspection route is calculated and analyzed according to the plurality of key points and the plurality of first stopping points, including: The first inspection time of the UAV flying and inspecting according to the first inspection route is obtained, wherein the first inspection time includes the flight time and the stopping time at the plurality of first stopping points.

[0058] The ratio of the preset inspection time and the first inspection time is calculated to obtain the first time fitness.

[0059] The first detection fitness of the first inspection route is calculated according to the plurality of key points and the plurality of first stopping points.

[0060] The first inspection fitness is calculated and obtained according to the first time fitness and the first detection fitness.

[0061] Preferably, to calculate and analyze the first inspection fitness of the first inspection route, the first inspection time of the UAV flying and inspecting according to the first inspection route is obtained, which covers the flight time and the stopping time at the plurality of first stopping points. For example, if the UAV spends 2 hours in flight and a total of 1 hour in each first stopping point, the first inspection time is 3 hours.

[0062] Then, the ratio of the preset inspection time and the first inspection time is calculated to obtain the first time fitness. Assuming that the preset inspection time is 4 hours and the first inspection time is 3 hours, the ratio is 3 / 4, which reflects the efficiency of the inspection in the time dimension. The larger the ratio, the more efficient the inspection in the time dimension.

[0063] Meanwhile, a first detection fitness of the first inspection route is calculated according to the plurality of key points and the plurality of first stay points. The key points represent important boundary positions of safe flight, and the first stay points are associated with detection of the target power tower. By analyzing the spatial relationship distance factor between the key points and the stay points, the rationality of the inspection route in detecting the target can be quantitatively evaluated, and the first detection fitness is obtained.

[0064] The first inspection fitness is calculated by comprehensively considering the first time fitness and the first detection fitness. The first time fitness focuses on the time efficiency of the inspection, and reflects the completion of the inspection task in the time dimension. The first detection fitness focuses on the rationality of the inspection route for target detection, and reflects the performance of the inspection task in the detection effect. By assigning different weights to the two and combining them to calculate the first inspection fitness, the comprehensive performance of the inspection route can be comprehensively measured. The specific weight configuration method can be set according to the actual situation, which is not limited here.

[0065] The first inspection fitness serves as a comprehensive index for measuring the pros and cons of the first inspection route, and provides a clear basis for subsequent optimization of the inspection route, which helps to improve the inspection efficiency, ensure flight safety and improve the detection quality, and ensures that the inspection task can be completed efficiently and accurately.

[0066] In a preferred embodiment, the first detection fitness of the first inspection route is calculated according to the plurality of key points and the plurality of first stay points, comprising: Obtaining a standard detection distance.

[0067] Calculating the distances between the plurality of key points and the plurality of first stay points to obtain a plurality of first actual detection distances.

[0068] Calculating the absolute deviation coefficients of the plurality of first actual detection distances and the standard detection distance respectively, and calculating a plurality of detection standard approximation coefficients.

[0069] Calculating the mean value of the plurality of detection standard approximation coefficients to obtain the first detection fitness.

[0070] Specifically, when calculating the first detection fitness, the standard detection distance is first obtained, which is a pre-set reference distance for measuring the rationality of the detection effect, for example, 50 meters. Then, the distances between the plurality of key points and the plurality of first stay points are calculated to obtain a plurality of first actual detection distances. Assuming that there is a key point and a first stay point, the actual distance between them is calculated by spatial coordinates to be 60 meters, which is one of the first actual detection distances.

[0071] Then, the absolute deviation coefficients of the plurality of first actual detection distances and the standard detection distance are calculated respectively, and the calculation method is that the absolute difference value of the first actual detection distance and the standard detection distance is divided by the standard detection distance. Taking the above example as an example, the absolute difference value is 60-50=10 meters, and the absolute deviation coefficient is 10÷50=0.2.

[0072] On this basis, the detection standard approximation coefficient is calculated, which is 1 minus the absolute deviation coefficient, that is, 1-0.2=0.8, and the greater the value, the closer the first actual detection distance is to the standard detection distance, and the better the detection effect. The mean value of the plurality of detection standard approximation coefficients calculated for all key points and first stopping points is the first detection fitness.

[0073] For example, if there are 5 combinations of key points and first stopping points, 5 detection standard approximation coefficients are calculated as 0.8, 0.85, 0.75, 0.9, and 0.82, and the mean value is (0.8+0.85+0.75+0.9+0.82)÷5=0.824, which is the first detection fitness.

[0074] The first detection fitness as a quantitative index can intuitively reflect the rationality of the inspection route in terms of detection targets, provide clear basis for subsequent optimization and adjustment of the inspection route, help to improve the inspection effect, and ensure more accurate and efficient detection of the target power tower.

[0075] The unmanned aerial vehicle inspection planning method for power failure feature perception provided by the embodiment of the application has at least the following technical effects: 1. By acquiring the failure data when the target power tower fails, combining the preset safety flight fence, dynamically integrating and adjusting to generate an adjusted safety flight fence sequence, and further marking key points, the safety flight range of the unmanned aerial vehicle can be flexibly adjusted according to real-time failure data and environmental changes, ensuring that the unmanned aerial vehicle is safe and efficient in complex and variable power failure scenarios, and effectively improving the safety and adaptability of the inspection.

[0076] 2. A plurality of safety flight fence generators are used, a network model capable of intelligently generating an adjusted safety flight fence according to different failure data is constructed based on deep learning technology and a large amount of sample data for training and optimization, intelligent planning of the inspection route is realized, the optimal inspection route meeting the safety flight constraint can be quickly generated according to the failure feature and the environmental condition, and the planning efficiency and accuracy of the inspection route are greatly improved.

[0077] 3. By comprehensively considering the inspection time fitness and the detection fitness, the generated inspection route is evaluated in terms of comprehensive fitness, and the optimal inspection route is obtained by optimizing the inspection route, which not only pays attention to the time efficiency of the inspection, but also pays attention to the detection quality of the inspection, ensures that the unmanned aerial vehicle can comprehensively and accurately perceive the power failure feature in the inspection process, provides strong support for subsequent fault diagnosis and repair, and significantly improves the overall performance of the unmanned aerial vehicle inspection of the power failure feature perception.

[0078] Embodiment two: As Figure 2 shown, based on the same inventive concept of the power failure feature perception unmanned aerial vehicle inspection planning method provided in embodiment one, the embodiment of the application also provides an unmanned aerial vehicle inspection planning system for power failure feature perception, which comprises: A data acquisition module 11 is configured to acquire failure data of a target power tower when a power failure occurs, integrate and adjust a preset safe flight fence to obtain a sequence of adjusted safe flight fences, and label a plurality of key points.

[0079] A route planning module 12 is configured to construct an inspection planning constraint according to the sequence of adjusted safe flight fences, and randomly plan a first inspection route for inspecting the target power tower, wherein the first inspection route comprises a plurality of first stopping points.

[0080] An adaptability calculation module 13 is configured to calculate and analyze the first inspection fitness of the first inspection route according to the plurality of key points and the plurality of first stopping points.

[0081] An inspection control module 14 is configured to optimize the inspection route to obtain an optimal inspection route, and control the unmanned aerial vehicle to perform the inspection of the target power tower.

[0082] Further, the data acquisition module 11 is further configured to perform the following steps: acquire failure data of a target power tower when a power failure occurs, wherein the failure data comprises power failure data and environmental data; acquire a preset safe flight fence of the target power tower, wherein the preset safe flight fence comprises safe flight boundary coordinates; combine the failure data and the preset safe flight fence, and input them into a plurality of safe flight fence generators respectively to generate a plurality of adjusted safe flight fences as a sequence of adjusted safe flight fences; and label a plurality of key points in the plurality of adjusted safe flight fences in the sequence of adjusted safe flight fences.

[0083] Further, the data acquisition module 11 is further configured to perform the following steps: In the inspection data of the power failure of the power tower, a sample preset safe flight fence set and a sample fault data set are collected, and a sample adjusted safe flight fence after different sample fault data is constructed to obtain a sample adjusted safe flight fence set; data in the sample preset safe flight fence set, the sample fault data set and the sample adjusted safe flight fence set is divided by a preset proportion to obtain first safe flight fence generation training data, and Kth safe flight fence generation training data is obtained by further division, K being a positive integer; a network structure of K safe flight fence generators is constructed by using deep learning; the K safe flight fence generators are respectively subjected to network parameter training and optimization by using the first safe flight fence generation training data to the Kth safe flight fence generation training data in turn, and are constructed after convergence is tested.

[0084] Further, the data acquisition module 11 is further configured to perform the following steps: In each adjusted safe flight fence, the distance between each adjusted safe flight boundary coordinate and the preset safe flight fence is calculated, and the adjusted safe flight boundary coordinate with the largest distance is selected as a key point; a plurality of key points of a plurality of adjusted safe flight fences in the adjusted safe flight fence sequence are obtained by calculation and processing.

[0085] Further, the route planning module 12 is further configured to perform the following steps: The union of the adjusted safe flight fence sequence is obtained to obtain a fusion safe flight fence, and a patrol planning constraint is constructed, wherein the patrol planning constraint includes that the patrol route does not enter the fusion safe flight fence; a first patrol route meeting the patrol planning constraint is randomly planned, wherein the first patrol route is used for patrol of the target power tower and includes a plurality of first stopping points.

[0086] Further, the fitness calculation module 13 is further configured to perform the following steps: A first patrol time of the unmanned aerial vehicle flying and patrolling according to the first patrol route is obtained, wherein the first patrol time includes a flight time and a stopping time at the plurality of first stopping points; a ratio of a preset patrol time and the first patrol time is calculated to obtain a first time fitness; a first detection fitness of the first patrol route is calculated according to the plurality of key points and the plurality of first stopping points; and a first patrol fitness is calculated according to the first time fitness and the first detection fitness.

[0087] Further, the fitness calculation module 13 is further configured to perform the following steps: A standard detection distance is obtained; distances between the plurality of key points and the plurality of first stay points are calculated to obtain a plurality of first actual detection distances; absolute deviation coefficients of the plurality of first actual detection distances and the standard detection distance are respectively calculated, and a plurality of detection standard approximation coefficients are calculated; a mean value of the plurality of detection standard approximation coefficients is calculated to obtain a first detection fitness.

[0088] Through the foregoing detailed description of the method for unmanned aerial vehicle inspection planning based on power failure feature perception, those skilled in the art can clearly understand the system for unmanned aerial vehicle inspection planning based on power failure feature perception in the embodiments. For the system disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple, and relevant parts can be found in the method part.

[0089] The above description of disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A UAV inspection planning method based on power fault feature perception, characterized by: A drone inspection planning system for power fault feature perception includes: Obtain fault data when a power failure occurs on a target power tower, perform integrated adjustments to the preset safety flight fence, obtain an adjusted safety flight fence sequence, and mark multiple key points; Constructing inspection planning constraints according to the adjusted safety flight fence sequence, and randomly planning a first inspection route for inspecting the target power tower, wherein the first inspection route includes a plurality of first stop points; Calculating and analyzing a first inspection fitness of the first inspection route based on the multiple key points and the multiple first stop points; Optimize the inspection route, obtain the optimal inspection route, and control the drone to inspect the target power tower.

2. The method for planning inspection by drones based on power fault feature perception according to claim 1 is characterized in that: Obtain fault data when a power failure occurs on the target power tower, perform integrated adjustments to the preset safety flight fence, obtain the adjusted safety flight fence sequence, and mark multiple key points, including: Acquire fault data when a power fault occurs on a target power tower, wherein the fault data includes power fault data and environmental data; Acquire a preset safe flight fence of the target power tower, wherein the preset safe flight fence includes safe flight boundary coordinates; Combining the fault data with the preset safety flight fence and inputting the combined data into a plurality of safety flight fence generators to generate a plurality of adjusted safety flight fences as an adjusted safety flight fence sequence; Key points are marked on a plurality of adjusted safety flight fences in the adjusted safety flight fence sequence to obtain a plurality of key points.

3. The method for planning inspection by drones based on power fault feature perception according to claim 2 is characterized in that: The steps of constructing the multiple safety flight fence generators include: In the inspection data of power failures occurring on power towers, a sample preset safety flight fence set and a sample fault data set are collected, and sample adjusted safety flight fences after different sample fault data are constructed to obtain a sample adjusted safety flight fence set; Dividing a preset proportion of data within the sample preset safety flight fence set, the sample fault data set, and the sample adjusted safety flight fence set to obtain a first safety flight fence generation training data, and continuing to divide to obtain a Kth safety flight fence generation training data, where K is a positive integer; Using deep learning, we build a network structure of K safe flight fence generators; The training data is generated from the first safety flight fence to the Kth safety flight fence in sequence, and the network parameters of the K safety flight fence generators are trained and optimized respectively, and the construction is completed after the test converges.

4. The method for planning inspection by drones based on power fault feature perception according to claim 2 is characterized in that: Key points of the multiple adjusted safety flight fences in the adjusted safety flight fence sequence are marked to obtain multiple key points, including: In each adjusted safety flight fence, calculating the distance between each adjusted safety flight boundary coordinate and the preset safety flight fence, and selecting the adjusted safety flight boundary coordinate with the largest distance as the key point; The calculation process obtains a plurality of key points of adjusting the safety flight fences in the safety flight fence adjustment sequence.

5. The method for planning inspection by drones based on power fault feature perception according to claim 1 is characterized in that: Establishing inspection planning constraints according to the adjusted safety flight fence sequence, and randomly planning a first inspection route for inspecting the target power tower, including: Obtaining the union of the adjusted safety flight fence sequences, obtaining a fused safety flight fence, and constructing inspection planning constraints, wherein the inspection planning constraints include that the inspection route does not enter the fused safety flight fence; A first inspection route that satisfies the inspection planning constraints is generated by random planning, wherein the first inspection route is used to inspect the target power pole tower and includes a plurality of first stop points.

6. The method for planning inspection by drones based on power fault feature perception according to claim 1 is characterized in that: Calculating and analyzing the first inspection fitness of the first inspection route according to the multiple key points and the multiple first stop points includes: Obtaining a first inspection time for the drone to perform a flight inspection along the first inspection route, wherein the first inspection time includes the flight time and the stay time at the plurality of first stay points; Calculate the ratio of the preset inspection time to the first inspection time to obtain the first time adaptability; Calculating a first detection fitness of the first inspection route according to the multiple key points and the multiple first stop points; A first inspection fitness is obtained by calculation according to the first time fitness and the first detection fitness.

7. The method for planning inspection by drones based on power fault feature perception according to claim 6 is characterized in that: Calculating a first detection fitness of the first inspection route according to the multiple key points and the multiple first stop points includes: Get the standard detection distance; Calculating distances between the plurality of key points and the plurality of first stop points to obtain a plurality of first actual detection distances; respectively calculating absolute deviation coefficients between the plurality of first actual detection distances and the standard detection distance, and calculating a plurality of detection standard approximation coefficients; An average of the plurality of detection standard approximation coefficients is calculated to obtain a first detection fitness.

8. A UAV inspection planning system based on power fault feature perception, characterized by: A method for planning drone inspections for implementing power fault feature perception according to any one of claims 1 to 7, the system comprising: The data acquisition module is used to obtain fault data when a power failure occurs on the target power tower, perform integrated adjustments to the preset safety flight fence, obtain the adjusted safety flight fence sequence, and mark multiple key points; a route planning module, configured to construct inspection planning constraints according to the adjusted safety flight fence sequence, and randomly plan a first inspection route for inspecting the target power tower, wherein the first inspection route includes a plurality of first stop points; a fitness calculation module, configured to calculate and analyze a first inspection fitness of the first inspection route based on the plurality of key points and the plurality of first stop points; The inspection control module is used to optimize the inspection route, obtain the optimal inspection route, and control the drone to inspect the target power tower.

9. The UAV inspection planning system for power fault feature perception according to claim 8 is characterized in that: Obtain fault data when a power failure occurs on the target power tower, perform integrated adjustments to the preset safety flight fence, obtain the adjusted safety flight fence sequence, and mark multiple key points, including: Acquire fault data when a power fault occurs on a target power tower, wherein the fault data includes power fault data and environmental data; Acquire a preset safe flight fence of the target power tower, wherein the preset safe flight fence includes safe flight boundary coordinates; Combining the fault data with the preset safety flight fence and inputting the combined data into a plurality of safety flight fence generators to generate a plurality of adjusted safety flight fences as an adjusted safety flight fence sequence; Key points are marked on a plurality of adjusted safety flight fences in the adjusted safety flight fence sequence to obtain a plurality of key points.

10. The UAV inspection planning system for power fault feature perception according to claim 9 is characterized in that: The steps of constructing the multiple safety flight fence generators include: In the inspection data of power failures occurring on power towers, a sample preset safety flight fence set and a sample fault data set are collected, and sample adjusted safety flight fences after different sample fault data are constructed to obtain a sample adjusted safety flight fence set; Dividing a preset proportion of data within the sample preset safety flight fence set, the sample fault data set, and the sample adjusted safety flight fence set to obtain a first safety flight fence generation training data, and continuing to divide to obtain a Kth safety flight fence generation training data, where K is a positive integer; Using deep learning, we build a network structure of K safe flight fence generators; The training data is generated from the first safety flight fence to the Kth safety flight fence in sequence, and the network parameters of the K safety flight fence generators are trained and optimized respectively, and the construction is completed after the test converges.

Images