Power system inspection unmanned aerial vehicle cooperative positioning method and system

Abstract

The invention discloses an electric power system inspection unmanned aerial vehicle cooperative positioning method and system, relates to the field of radio positioning, and aims to improve the positioning precision and cooperative capability of an unmanned aerial vehicle group in an electric power inspection process. By introducing a hierarchical positioning mechanism of the piloting unmanned aerial vehicle and the common unmanned aerial vehicle, the piloting unmanned aerial vehicle is responsible for providing high-precision positioning data, and the common unmanned aerial vehicle performs positioning calibration through communication with the piloting unmanned aerial vehicle. According to the method, a GPS, an inertial navigation system and a visual SLAM technology are combined, environmental factors such as wind speed, illumination intensity and the like are comprehensively considered, and calibration frequency and time window duration are dynamically adjusted by calculating credibility of each unmanned aerial vehicle. According to the method, the adaptability and the collaborative operation capability of the unmanned aerial vehicle group in a complex environment are remarkably enhanced, the dependence on the performance of a single unmanned aerial vehicle is reduced, high-precision positioning can still be kept in an environment with weak or changeable GPS signals, and the method has a wide application prospect.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of radio positioning, in particular to a power system inspection unmanned aerial vehicle cooperative positioning method and system. BACKGROUND

[0002] With the rapid development of unmanned aerial vehicle technology, unmanned aerial vehicles have been widely used in many industries, especially in the power industry, and have gradually become an essential tool for important tasks such as power grid inspection, transmission line monitoring, and post-disaster recovery. Unmanned aerial vehicles are efficient, flexible, and fast, and can replace manual high-altitude inspection, greatly improving the efficiency and safety of power grid operation and maintenance. However, in complex power grid environments, the positioning accuracy, flight stability, and cooperative work capability of unmanned aerial vehicles still face many challenges, especially in harsh weather conditions or areas with strong electromagnetic interference, where existing unmanned aerial vehicle positioning technologies cannot meet the high requirements for accuracy, real-time performance, and reliability of power grid inspection tasks.

[0003] In Chinese Patent Publication No. CN 114973036 B, a GNSS / inertial navigation / wireless base station fusion-based unmanned aerial vehicle three-dimensional positioning method is disclosed. Based on information geometry theory, the satellite navigation, inertial navigation, and radio navigation information carried by the unmanned aerial vehicle are converted into different information probability models, solving the problem of different information format of heterogeneous navigation sources during fusion. In addition, by calculating the navigation source information accuracy probability function and fusing multiple probability density functions, the positioning result is obtained. However, these methods often require complex algorithms and a large amount of computing resources, and may be affected by environmental changes, differences in unmanned aerial vehicle performance, and other factors in actual application, resulting in unstable cooperative effect. SUMMARY

[0004] The present application aims to improve the positioning accuracy of unmanned aerial vehicle groups, reduce the dependence on the performance of individual unmanned aerial vehicles, and enable unmanned aerial vehicles to adapt to changing working environments, such as GPS signal-limited environments such as urban canyons and indoor environments, as well as outdoor environments under different weather conditions.

[0005] To solve the above technical problems, the technical solution adopted by the present application is as follows: A power system inspection unmanned aerial vehicle cooperative positioning method, comprising: dividing a group of unmanned aerial vehicles with a number greater than 3 into a lead unmanned aerial vehicle and ordinary unmanned aerial vehicles; the lead unmanned aerial vehicle simultaneously positions through GPS positioning technology, an inertial navigation system, and visual SLAM; the ordinary unmanned aerial vehicles correct their own positioning according to the distance between themselves and each lead unmanned aerial vehicle; and the unmanned aerial vehicles plan a path according to the current positioning and complete the collection of power device information after reaching the target position.

[0006] Preferably, the specific step of dividing the UAV group with more than three UAVs into the leader UAV and the common UAV includes: setting a time window with a time length of T for the UAV group with more than three UAVs, calculating the credibility of all UAVs at the beginning of any time window, and taking the top three UAVs with the highest credibility as the leader UAV of the current time window.

[0007] Preferably, the credibility of the UAV is obtained according to the signal-to-noise ratio of the GPS signal of the UAV, the measured wind speed and the measured light intensity.

[0008] Preferably, the time length T of the time window is determined by a genetic algorithm, and specifically includes: establishing a fitness function to balance the relationship between the time window length and the positioning error, and obtaining the optimal time window parameters through selection, crossover and mutation operations in population iteration.

[0009] Preferably, the fitness function of the genetic algorithm is a weighted sum function of the time window length and the positioning error, wherein the weight coefficient of the positioning error is greater than the weight coefficient of the time window length.

[0010] Preferably, the specific method of positioning by the GPS positioning technology, the inertial navigation system and the visual SLAM is as follows: The GPS data, the inertial navigation data and the visual SLAM data are weighted and averaged as the positioning data of the leader UAV; The weight of the GPS data is determined based on the high and low of the signal-to-noise ratio; The weight of the inertial navigation data is determined according to the change of the wind speed; The weight of the visual SLAM data is determined according to the light intensity.

[0011] Preferably, the specific method of correcting the positioning of the common UAV according to the distance between the common UAV and each leader UAV is that the common UAV obtains the distance between each leader UAV through a distance measuring module, combines the positioning data of the leader UAV, and calculates the self-calibration coordinates by using a trilateration algorithm.

[0012] Preferably, the cooperative positioning method of the power system inspection UAV can also dynamically adjust the calibration frequency. The dynamic adjustment of the calibration frequency calculates an environmental credibility index according to the average GPS signal-to-noise ratio, the average environmental wind speed and the average light intensity of the UAV group in the current time window, and linearly adjusts the calibration period between the preset maximum calibration interval and the minimum calibration interval based on the environmental credibility index.

[0013] A cooperative positioning system of a power system inspection UAV, which executes the cooperative positioning method of the power system inspection UAV according to any one of claims 1 to 8, comprising: A leader UAV selection module for dividing the leader UAV and the common UAV. The navigation unmanned aerial vehicle positioning module acquires accurate positioning data of the navigation unmanned aerial vehicle. The common unmanned aerial vehicle positioning module acquires accurate positioning data of the common unmanned aerial vehicle.

[0014] Preferably, the power system inspection unmanned aerial vehicle cooperative positioning system further comprises: A wind speed sensor acquires wind speed information. A light sensor acquires light information. A camera acquires visual information and performs image acquisition. An infrared thermal imager detects overheated components.

[0015] The above techniques improve the positioning accuracy of the unmanned aerial vehicle group, reduce the dependence on the performance of a single unmanned aerial vehicle, and enable the unmanned aerial vehicle to adapt to variable working environments, such as GPS signal limited environments such as urban canyons, indoor environments, and outdoor environments under different weather conditions. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 is a flowchart of the power system inspection unmanned aerial vehicle cooperative positioning method.

[0017] Figure 2 is a schematic diagram of the power system inspection unmanned aerial vehicle cooperative positioning system. DETAILED DESCRIPTION

[0018] The following is a specific embodiment of the present application, which will be described in detail in combination with the drawings.

[0019] Example 1 The positioning process of the present application is shown in Figure 1 and includes the following steps.

[0020] Setting of the time window: For a group of unmanned aerial vehicles (more than three unmanned aerial vehicles), a time window with a duration of T is first set, at the beginning of each time window, the reliability of all unmanned aerial vehicles is calculated, and the top three unmanned aerial vehicles with the highest reliability are selected as the navigation unmanned aerial vehicles of the current time window.

[0021] Positioning of the navigation unmanned aerial vehicle: Each navigation unmanned aerial vehicle is positioned by three techniques: GPS positioning, inertial navigation system (INS), and visual SLAM.

[0022] At the current position, the positioning data of GPS, INS, and visual SLAM are acquired respectively, and the weights of these data are calculated to obtain accurate positioning of the navigation unmanned aerial vehicle.

[0023] Positioning of the common unmanned aerial vehicle: For each regular UAV except the lead UAV, its current position data is first obtained through GPS technology.

[0024] Within the time window, distance measurements are taken with the three lead UAVs through wireless communication every certain time (referred to as t), and the position of the regular UAV is calibrated based on the accurate position data of the lead UAVs to obtain more accurate position data.

[0025] Calculation of credibility: The credibility calculation of each UAV involves multiple factors, including the signal-to-noise ratio of the GPS signal, wind speed, and light intensity.

[0026] By calculating the scores of these factors, the overall credibility of each UAV can be obtained.

[0027] Dynamic adjustment of calibration frequency: The calibration frequency (i.e., the value of t) is dynamically adjusted based on the overall credibility of the UAV swarm. If the credibility is high, the calibration frequency (t) can be appropriately increased to reduce the communication and calculation frequency; conversely, the frequency is reduced to ensure higher positioning accuracy.

[0028] Next, these steps will be introduced in more detail.

[0029] For a UAV swarm with more than three UAVs, a time window with a duration of T is set, which is a pre-set time period for performing specific tasks or calculations within a specific time interval, where the value of T is selected by a genetic algorithm.

[0030] The duration T of the time window is selected by a genetic algorithm, which is a heuristic search algorithm that simulates the process of biological evolution. It iterates through a population of candidate solutions through selection, crossover, mutation, and other operations to find the optimal or near-optimal solution to a problem. This algorithm draws on principles of natural selection and genetics, allowing the algorithm to evolve and adapt during the search process, thereby demonstrating strong search capabilities and high efficiency in complex optimization problems. Genetic algorithms are particularly suitable for solving complex, multi-peak, non-linear, or high-dimensional optimization problems that are difficult to solve using traditional mathematical methods. The specific operations are as follows: Step one: set the population size and randomly generate x candidate T values. The randomly generated x candidate T values form the initial population, and the maximum number of iterations is set. Step two: for any candidate T value, use the formula Calculate the fitness value F of the candidate T value, where T is the size of the candidate T value, E represents the average error of the unmanned aerial vehicle positioning when the candidate T value is applied to the power system inspection unmanned aerial vehicle cooperative positioning, and α1 and α2 are weight coefficients of the size of the candidate T value and the average error E, respectively. The formula indicates that the optimization goal of the candidate T value is to maximize the size of the T value and minimize the average error E. The higher the fitness value F obtained by calculation is, the better the performance of the candidate T value is. Step three: apply the roulette wheel selection method to select the corresponding number of candidate T values with the best performance at a preset selection ratio, and group the selected candidate T values to form a selected individual set; Step four: apply a preset crossover probability to the candidate T values in the selected individual set to perform a crossover operation, generate a corresponding number of new candidate T values, and group the obtained new candidate T values to form a crossover individual set. Randomly select a candidate T value from the crossover individual set, and perform a small amplitude adjustment on the candidate T value according to a preset mutation probability. The crossover individual set after adjustment is the mutation individual set; Step five: merge the selected individual set and the mutation individual set to obtain a merged individual set, calculate the fitness of all candidate T values in the merged individual set, and select the x candidate T values with the best fitness as a new generation population. Step six: repeat steps three to five to iteratively update the population until the maximum number of iterations is reached. The candidate T value with the best fitness in the last generation population is the optimal T value.

[0031] At the beginning of any time window, calculate the credibility of all unmanned aerial vehicles. The credibility is an index for measuring the performance and reliability of the unmanned aerial vehicle. The top three unmanned aerial vehicles in terms of credibility are selected as the lead unmanned aerial vehicles in the current time window. The lead unmanned aerial vehicles serve as leaders in the group, and their selection is crucial for the cooperative work of the entire group. High credibility of the lead unmanned aerial vehicles helps to improve the positioning accuracy and reliability of the entire unmanned aerial vehicle group.

[0032] Calculate the credibility of all unmanned aerial vehicles. The specific operation is as follows: Obtain the GPS signal signal-to-noise ratio SNR(i), measured wind speed V(i), and measured light intensity L(i) of all unmanned aerial vehicles, i=1, 2, …, n; n represents the number of unmanned aerial vehicles in the unmanned aerial vehicle group. Use the formula Calculate the first score A SNR(i) of the signal-to-noise ratio of the ith unmanned aerial vehicle, where SNR(i) min and SNR(i) max are the minimum and maximum values of the GPS signal signal-to-noise ratio SNR(i) of all unmanned aerial vehicles, respectively. Use the formula to calculate the first score A of the wind speed of the ith unmanned aerial vehicle.V(i) wherein V(i) min and V(i) max are the minimum and maximum values respectively among all the wind speed V(i) measured by the UAVs; The first score of light A of the i-th UAV is calculated by the formula L(i) wherein L(i) min and L(i) max are the minimum and maximum values respectively among all the light intensity L(i) measured by the UAVs; The reliability of the i-th UAV is calculated by the formula A(i) = 0.6A SNR(i) + 0.2A V(i) + 0.2A L(i) .

[0033] For any one of the lead UAV, simultaneously through GPS positioning technology, inertial navigation system and visual SLAM positioning; GPS is a satellite-based navigation system that can provide location and time information worldwide; inertial navigation system determines the position change of the UAV by measuring its acceleration and rotation; INS does not rely on external signals, so it is very useful in environments where GPS signals are blocked; visual SLAM uses visual sensors such as cameras to simultaneously construct an environmental map and position the UAV; visual SLAM is particularly effective in indoor or GPS signal deficient environments; at the current positioning time point, GPS positioning data, INS positioning data and visual SLAM positioning data are obtained respectively; the weights of GPS positioning data, INS positioning data and visual SLAM positioning data are calculated respectively; each positioning technology has its advantages and limitations, therefore, the system will calculate the weight of each positioning data according to the current environmental conditions and sensor performance; the weight reflects the relative importance of each data in the final positioning result; by assigning weights to different positioning data, the system can optimize the positioning result, reduce errors, and improve the accuracy and robustness of positioning; based on the obtained weights, the accurate positioning data of the current lead UAV is calculated; the fused accurate positioning data provides a more reliable position estimate that integrates information from multiple sensors; this method can reduce the impact of single sensor failure or error on the positioning result, and improve the positioning performance of the UAV in various environments.

[0034] The weights of GPS positioning data, INS positioning data and visual SLAM positioning data are calculated as follows: for any one of the three lead UAVs, based on the current lead UAV's obtained GPS signal-to-noise ratio SNR, measured wind speed V and measured light intensity L, the weight ω1 of GPS positioning data is calculated by the formula , wherein SNR max and SNRmin These are the preset maximum and minimum reference signal-to-noise ratios, respectively; using the formula... Calculate the weights for acquiring INS location data, ω2, where V max and V min These are the preset maximum and minimum reference wind speeds; using the formula... Calculate the weights ω3 of the visual SLAM localization data, where L max and L min The maximum reference light intensity and minimum light intensity are preset respectively.

[0035] The precise positioning data of the navigation drone is calculated based on the acquired weights. The specific operation is as follows: For any navigation drone, based on the weights ω1 of the GPS positioning data, ω2 of the INS positioning data, and ω3 of the visual SLAM positioning data acquired by the current navigation drone, the formula P = ω1P is used. GPS +ω2P INS +ω3P SLAM Calculate and obtain the precise positioning data P of the current navigation drone, where P GPS P INS and P SLAM These are GPS positioning data, INS positioning data, and visual SLAM positioning data, respectively.

[0036] For any ordinary drone other than the lead drone, GPS positioning technology is used for positioning. GPS positioning data is acquired at the current positioning time. While GPS provides basic location information, it can be affected by various factors (such as signal blockage and multipath effects), resulting in suboptimal positioning accuracy. Within the current time window, every time interval t, the distance to the three lead drones is acquired via wireless communication. Based on the precise positioning data of the three lead drones, a calibration positioning is performed to obtain the precise positioning data for the current ordinary drone. Calibration positioning allows ordinary drones to use the high-precision position information of the lead drones to correct their own GPS positioning, thereby improving their positioning accuracy. This method is particularly suitable for environments with unstable GPS signals or severe multipath effects. Precise positioning data is crucial for the collaborative operation of drone swarms, ensuring that each member of the swarm accurately knows its location, enabling efficient and safe mission execution.

[0037] The specific steps for calibration and positioning are as follows: For any ordinary drone, the distances between the current drone and three lead drones are obtained using wireless communication technology. Based on the precise positioning data of the three lead drones, the precise positioning data of the current drone is obtained using trilateration.

[0038] At the start of any given time window, the reliability of the drone swarm is calculated. This reliability is a comprehensive indicator reflecting the swarm's reliability and efficiency in collaborative operations. By calculating this indicator, the system can assess the overall performance of the drone swarm and make corresponding adjustments. The value of 't' is adjusted based on the acquired drone swarm reliability. 't' is the time interval for distance measurement and calibration positioning between the ordinary drone and the lead drone. The system adjusts this time interval according to the drone swarm's reliability. If the drone swarm's reliability is high, it indicates good collaborative positioning accuracy and stability, allowing for an increase in the value of 't' to reduce communication and computation frequency, thus saving energy and computing resources. If the drone swarm's reliability is low, it indicates a need for more frequent calibration to improve positioning accuracy and stability, therefore the value of 't' is decreased, increasing the calibration frequency. Dynamically adjusting the value of 't' makes the collaborative operation of the drone swarm more flexible and efficient, optimizing resource utilization while ensuring positioning accuracy, and improving the overall system's adaptability and response speed.

[0039] The reliability of the drone swarm is calculated, and the value of t is adjusted based on the obtained reliability. The specific operation is as follows: Based on the GPS signal-to-noise ratio (SNR) (i), measured wind speed (V) (i), and measured light intensity (L) (i) of all drones acquired at the beginning of the current time window, the formula is used... Calculate and obtain the average signal-to-noise ratio Using formula Calculate and obtain average wind speed Using formula Calculate average illumination Then, the obtained average signal-to-noise ratio was used. Average wind speed and average light Calculate and obtain the second score B of the signal-to-noise ratio. SNR Wind speed second rating B V And lighting second rating B L ; Calculate the second score B for signal-to-noise ratio SNR The formula is Calculate wind speed, second score B V The formula is Calculate the second score of illumination B L The formula is Finally, the formula B = 0.6B is used. SNR +0.2B V +0.2B L Calculate and obtain the credibility B of the drone swarm; based on the obtained credibility B of the drone swarm, use the formula t = t min +B(t max -t min ) Calculate the duration t, where tmax and t min These are the preset longest and shortest intervals, respectively, where t represents the calibration and positioning of a regular drone every t time interval within the current time window.

[0040] A smart positioning and collaborative system for unmanned aerial vehicles, such as Figure 2 As shown, it includes: The navigation drone selection module includes a credibility calculation unit and a selection unit. The credibility calculation unit is used to calculate the credibility of all drones at the start of any time window. The selection unit is used to select the top three drones in terms of credibility as the navigation drones for the current time window.

[0041] The navigation drone positioning module is used to simultaneously locate any navigation drone using GPS positioning technology, inertial navigation system and visual SLAM. At the current positioning time, it acquires GPS positioning data, INS positioning data and visual SLAM positioning data respectively; calculates the weights of GPS positioning data, INS positioning data and visual SLAM positioning data respectively, and calculates the accurate positioning data of the current navigation drone based on the acquired weights. The ordinary drone positioning module is used to locate any ordinary drone other than the lead drone using GPS positioning technology. It acquires GPS positioning data at the current positioning time point. Within the current time window, every time interval t, it acquires the distance to the three lead drones through wireless communication technology. Based on the accurate positioning data of the three lead drones, it performs a calibration positioning and acquires the accurate positioning data of the current ordinary drone.

[0042] Wind speed sensors help assess the impact of wind on the environment and adjust the calibration frequency. Wind speed sensors measure the wind speed in the drone's environment. During drone positioning, wind speed is a crucial environmental factor that directly affects the drone's flight status and positioning accuracy. Through wind speed sensors, the system can monitor wind speed changes in real time and adjust the positioning algorithm accordingly.

[0043] Signal-to-noise ratio (SNR) score: Wind speed measurement data is used to calculate the drone's "wind speed score." The wind speed score is a relative value calculated based on the wind speed data of all drones, used to evaluate the stability and reliability of each drone. If the wind speed is too high, it may cause the drone's flight path to become unstable, thus affecting positioning accuracy.

[0044] Calibration Frequency Adjustment: By monitoring wind speed in real time, the system can dynamically adjust the calibration frequency of ordinary drones. If the ambient wind speed is high, the calibration frequency may be increased (the t-value decreased) to improve positioning accuracy and avoid interference from wind speed on drone positioning. The illumination sensor provides data on ambient light intensity, optimizing the performance of visual SLAM. The illumination sensor measures the ambient light intensity, especially in environments with large light variations, such as indoors or in shaded areas. Light intensity directly affects the performance of visual SLAM (Simultaneous Localization and Mapping), as visual SLAM relies on images of the surrounding environment acquired by a camera. Illumination Score: Data from the illumination sensor is used to calculate the drone's "illumination score." This score is evaluated by comparing the illumination intensity of each drone in the drone swarm. Areas with higher or more stable illumination improve the positioning accuracy of visual SLAM, while insufficient illumination can lead to inaccurate visual SLAM. Positioning Data Weighting: The measurement results from the illumination sensor help weight the positioning data for visual SLAM. If the illumination intensity is low, the weight of the visual SLAM data may be reduced, thus relying on GPS and inertial navigation system (INS) data to provide more stable positioning results.

[0045] Cameras, through visual SLAM technology, not only help drones achieve precise localization but also provide 3D maps of the surrounding environment, enhancing their adaptability to complex environments. Cameras are primarily used in visual SLAM technology to capture image information of the surrounding environment for localization and mapping. Visual SLAM relies on camera-acquired image information and uses image processing to calculate the drone's current position and the map of its surrounding environment. Precise Localization: The image data captured by the camera, combined with other sensors (such as GPS and INS), helps the drone achieve high-precision localization, especially in environments with weak or no GPS signals (such as indoors or urban canyons). Visual SLAM can compensate for the shortcomings of GPS positioning, providing more stable and accurate position estimates. Environmental Perception and Map Building: Cameras also help drones perceive their environment and build real-time 3D environmental maps. This not only assists in precise localization but also improves the drone's obstacle avoidance capabilities and adaptability to environmental changes.

[0046] After obtaining its own location, the drone plans a route to the target point based on the current location. The specific method is as follows: (1) Global coarse planning (offline stage) Generate an initial reference path (such as A*, RRT*) for each UAV to avoid known static obstacles, reduce real-time computation, and provide a general direction.

[0047] (2) Distributed Model Predictive Control (DMPC, Online Phase) Each drone optimizes its future trajectory within the next few seconds based on its own and neighboring information, avoids obstacles in real time, and dynamically adjusts its path.

[0048] (3) Obstacle avoidance using the artificial potential field method (APF) Attractive force: The target point generates an attractive force on the drone, and the potential field function is U_att = 1 / 2k_att*d 2 Repulsive force: Other drones / obstacles generate a repulsive force, and the potential field function is: U_rep = 1 / 2k_rep*(1 / d_safe - 1 / d_ij) 2 (if d_ij < d_safe, otherwise 0) where d_ij is the distance to the neighbor and d_safe is the safety threshold; Direction of the resultant force: The drone moves along the direction of the potential field gradient descent.

[0049] After reaching the target point, the drone can quickly obtain all-round data of the power tower by carrying high-precision sensors, and achieve: Defect identification (rust, insulator damage, bolt loosening, etc.); 3D modeling (evaluating structural deformation and tower foundation settlement); Disaster emergency (rapid assessment after lightning strikes, icing, and wildfires).

[0050] It should be understood that those of ordinary skill in the art can make improvements or transformations based on the above description, and all such improvements and transformations shall fall within the protection scope of the appended claims of this invention. The parts not described in detail in this specification belong to the prior art well-known to those of ordinary skill in the art.

Claims

1. A cooperative positioning method for unmanned aerial vehicles (UAVs) used for power system inspection, characterized in that, include: A drone swarm with more than 3 drones is divided into a lead drone and ordinary drones. The lead drones are positioned using GPS, inertial navigation, and visual SLAM. The ordinary drones correct their own positioning based on their distance from each lead drone. The drones plan their paths based on their current positioning and collect information about the power equipment after reaching the target location.

2. The collaborative positioning method for power system inspection drones according to claim 1, characterized in that, The specific steps for dividing a drone swarm with more than 3 drones into lead drones and ordinary drones include: for a drone swarm with more than 3 drones, setting a time window of duration T, at the beginning of any time window, calculating the credibility of all drones, and taking the top three drones in terms of credibility as the lead drones for the current time window.

3. The collaborative positioning method for power system inspection drones according to claim 2, characterized in that, The reliability of the drone is determined based on the signal-to-noise ratio of the drone's GPS signal, measured wind speed, and measured light intensity.

4. The collaborative positioning method for power system inspection drones according to claim 2, characterized in that, The duration T of the time window is determined by optimization using a genetic algorithm, specifically including: establishing a fitness function to balance the relationship between the time window length and the positioning error, and obtaining the optimal time window parameters through selection, crossover, and mutation operations in population iteration.

5. The collaborative positioning method for power system inspection drones according to claim 4, characterized in that, The fitness function of the genetic algorithm is a weighted sum of the time window length and the positioning error, wherein the positioning error weight coefficient is greater than the time window length weight coefficient.

6. The collaborative positioning method for power system inspection drones according to claim 1, characterized in that, The specific method for positioning using GPS positioning technology, inertial navigation system and visual SLAM is as follows: The GPS data, inertial navigation data, and visual SLAM data are weighted and averaged to obtain the positioning data for the navigation drone; The weights of the GPS data are determined based on the signal-to-noise ratio. The weights of the inertial navigation data are determined based on changes in wind speed; The weights of the visual SLAM data are determined based on the light intensity.

7. The collaborative positioning method for power system inspection drones according to claim 1, characterized in that, The specific method by which the ordinary drone corrects its own positioning based on the distance between itself and each lead drone is as follows: the ordinary drone obtains the distance between itself and each lead drone through the ranging module, and calculates its own calibration coordinates by combining the positioning data of the lead drones and using a trilateration algorithm.

8. The collaborative positioning method for power system inspection drones according to any one of claims 1 to 7, characterized in that, The power system inspection drone collaborative positioning method can also dynamically adjust the calibration frequency. The dynamic adjustment of the calibration frequency is based on the average GPS signal-to-noise ratio, average ambient wind speed and average light intensity of the drone swarm within the current time window to calculate the environmental reliability index. Based on the environmental reliability index, the calibration cycle is linearly adjusted between the preset maximum calibration interval and the minimum calibration interval.

9. A cooperative positioning system for power system inspection drones, comprising executing the cooperative positioning method for power system inspection drones as described in any one of claims 1 to 8, characterized in that, include: The pilot drone selection module distinguishes between pilot drones and regular drones; The drone positioning module acquires precise positioning data for the drone. A standard drone positioning module that acquires precise positioning data for standard drones.

10. A collaborative positioning system for power system inspection drones according to claim 9, characterized in that, Also includes: Wind speed sensor to acquire wind speed information; A light sensor acquires light information; A camera acquires visual information and captures images. Infrared thermal imager for detecting overheated components.

Images