A method and system for generating wind turbine inspection routes based on drones

Abstract

The present invention discloses a method for generating a wind turbine inspection route based on an unmanned aerial vehicle. Through completely autonomous flight, the longitude and latitude of the wind turbine and the altitude of the hub center point are automatically remeasured each time an inspection is performed to avoid errors caused by longitude and latitude. A classification model is added when identifying the yaw angle and blade phase angle of the wind turbine, and the recognition rate is improved through redundancy; after arriving directly in front of the hub, the final correction is performed to ensure that there is no deviation in the position of the wind turbine, solving the problem of measurement and longitude and latitude deviation in the prior art patents. At the same time, a curve fitting equation for blade pre-bending is introduced to solve the problem that the blade will exceed the camera screen when inspecting the leading and trailing edges. This application automatically obtains the key parameters, yaw angle and blade phase angle of the wind turbine through a drone, and detects the blade pre-bending through a curve fitting algorithm, and finally generates an accurate drone blade inspection route.

Description

Technical Field

[0001] The present invention relates to the field of new energy, and in particular to a method and system for generating a wind turbine inspection route based on an unmanned aerial vehicle (UAV), which can be used for automatically generating a route in a UAV wind turbine inspection. Background Art

[0002] At present, most of the devices that capture wind energy are wind turbines composed of towers, wind turbine chassis, and wind turbine blades. The wind turbine blades need to rotate continuously under the action of wind power to complete energy conversion. Its safety and reliability are the key to ensuring the normal operation of wind turbines. Regular inspections of wind turbine blades can timely check faults and confirm fault information, and perform maintenance and repairs on wind turbine blades in advance according to the fault level. Wind turbine blade inspections are a key step to ensure the long-term normal operation of wind turbines.

[0003] For the inspection of wind turbine blades, the operation and maintenance personnel mainly use handheld telescopes to check whether there are any abnormalities on the blade surface. The work intensity of manual inspection is high and the efficiency is low. Manual inspection based on naked eye observation and experience judgment cannot guarantee the accuracy of fault detection, and is accompanied by high inspection costs. In recent years, some wind farms have configured drones for blade inspection, but the operation of drone inspection is relatively complicated, and some routes are difficult to operate manually. In response to this, some wind turbine inspection schemes have recently introduced automatic route generation algorithms, which greatly improve the inspection efficiency. However, the current automatic route generation algorithm requires manual input of wind turbine parameters (such as wind turbine longitude and latitude, tower height, etc.), and there will be certain errors when calculating the yaw angle of the wind turbine, which greatly affects the accuracy of subsequent route generation. In the method and system for generating wind turbine inspection routes based on drones CN115480589A, the longitude and latitude of the wind turbine and the tower need to be provided additionally. In actual flight, due to the lack of network RTK signals in some areas, there will be certain deviations every time the drone is turned on, and re-measurement is required. However, the patent does not provide a corresponding solution. In addition, the wind turbine yaw angle and blade phase angle are affected by the segmentation image algorithm, and there is no other redundant method. Finally, the influence of blade pre-bending is not taken into account in the route design. Summary of the invention

[0004] The technical problem to be solved by the present invention is to ensure the reliability of the equipment and provide a method for generating a wind turbine inspection route based on a drone. Through completely autonomous flight, the longitude and latitude of the wind turbine and the altitude of the center point of the hub are automatically remeasured each time an inspection is carried out to avoid errors caused by longitude and latitude. A classification model is added when identifying the yaw angle and blade phase angle of the wind turbine to improve the recognition rate through redundancy. After arriving directly in front of the hub, the final correction is performed to ensure that there is no deviation in the position of the wind turbine, which solves the problem of deviation in measurement and longitude and latitude in the prior art patents. At the same time, a curve fitting equation for blade pre-bending is introduced to solve the problem that the blades will exceed the camera screen when inspecting the leading and trailing edges. The present application automatically obtains the key parameters, yaw angle and blade phase angle of the wind turbine through a drone, and detects the blade pre-bending through a curve fitting algorithm, and finally generates an accurate drone blade inspection route.

[0005] The technical solution of this application is:

[0006] A method for generating a wind turbine inspection route based on a drone, wherein the wind turbine comprises a wind tower and an impeller arranged at the top of the wind tower, wherein the impeller comprises a hub and three blades evenly distributed along the circumference of the hub, and the drone is equipped with a pan / tilt with a camera and a GPS positioning system, and is characterized in that the method comprises the following steps:

[0007] S1, define the longitude and latitude waypoints of the wind turbine, the longitude and latitude waypoints of the wind turbine include the first longitude and latitude waypoint of the wind turbine and the second longitude and latitude waypoint of the wind turbine, the longitude and latitude of the first longitude and latitude waypoint of the wind turbine are the longitude and latitude (lng1, lat1) where the UAV is located, the height of the first longitude and latitude waypoint of the wind turbine is higher than the surrounding obstacles, adjust the gimbal to shoot horizontally, rotate the UAV nose direction on the spot, so that the wind turbine tower appears completely in the picture captured by the camera; at this time, the position of the UAV is A, and the longitude and latitude are (lkng1, lat1); the height of the longitude and latitude waypoint of the second wind turbine is the same as that of the first wind turbine, keep the shooting posture and nose direction unchanged, the longitude and latitude of the second wind turbine longitude and latitude waypoint is the longitude and latitude corresponding to when the camera shooting picture is facing the center of the tower, at this time, the position of the UAV is F, and the longitude and latitude are (lng2, lat2), calculate the longitude and latitude of the wind turbine tower (lng3, lat3), and the position of the wind turbine tower is D;

[0008] S2, based on the acquired longitude and latitude of the wind turbine tower (lng3, lat3), generate a route for the drone to reach the top of the wind turbine, wherein the route for the drone to reach the top of the wind turbine includes four waypoints; the first waypoint is the position F where the drone is located; the longitude and latitude of the second waypoint are the same as those of the first waypoint (lng1, lat1), and the height is the height h1 of the center of the wind turbine hub; the longitude and latitude of the third waypoint are (lng1, lat1), and the height h2 is greater than the sum of the height h1 of the center of the wind turbine hub and the length l of the wind turbine blade; the longitude and latitude of the fourth waypoint are the longitude and latitude of the wind turbine tower (lng3, lat3), and the height of the fourth waypoint is the same as that of the third waypoint, which is h2; after arriving at the fourth waypoint, the gimbal is adjusted to shoot vertically downward to obtain a bird's-eye view of the wind turbine;

[0009] S3, calculating the yaw angle α of the wind turbine based on the top view of the wind turbine;

[0010] S4, based on the calculated yaw angle α of the wind turbine, generate an initial route to the front of the hub; the initial route to the front of the hub includes two waypoints, the first waypoint of the initial route to the front of the hub has a height of h2, and the longitude and latitude are the longitude and latitude of the position in front of the hub; the longitude and latitude of the second waypoint of the initial route to the front of the hub are the same as those of the first waypoint of the initial route to the front of the hub, and the height is the height h1 of the center of the hub, and the position of the second waypoint of the initial route to the front of the hub is Q, and when arriving at position Q, the gimbal is adjusted to horizontal shooting, and the nose direction is adjusted to the yaw angle of the wind turbine -180°;

[0011] S5, obtain the front view of the wind turbine hub from the camera in the gimbal, and generate a calibrated hub front route based on the hub recognition algorithm. The calibrated hub front route contains two waypoints. The first waypoint of the calibrated hub front route is position Q, the nose direction and shooting posture remain unchanged, and the second waypoint of the calibrated hub front route is the center of the hub located at the center of the picture taken by the camera, and the front view of the wind turbine hub is obtained based on the second waypoint of the calibrated hub front route;

[0012] S6, calculating the fan blade phase angle β and blade pre-bend based on the hub front view obtained in step S5;

[0013] S7, generating an inspection route based on the longitude and latitude (lng3, lat3) of the wind turbine tower, the wind turbine yaw angle α, the wind turbine blade phase angle β, the tower height, the blade pre-bend and the wind turbine blade length l.

[0014] Preferably, calculating the longitude and latitude (lng3, lat3) of the wind turbine tower in S1 specifically includes the following steps:

[0015] 101) At the first wind turbine longitude and latitude waypoint, based on the tower recognition algorithm, the wind turbine tower appears completely in the picture acquired by the camera, and the wind turbine tower image I1 of the first wind turbine longitude and latitude waypoint is acquired. At this time, the position of the tower center point in the wind turbine tower image I1 is (μ1, v1);

[0016] 102) Based on the wind turbine tower image I1 and combined with the camera internal parameters, the deflection angle c of the wind turbine relative to the direction of the UAV nose is calculated;

[0017] 103) The flight direction is perpendicular to the nose of the aircraft, the shooting posture and the direction of the nose are kept unchanged, the latitude and longitude waypoints of the second wind turbine corresponding to the time when the camera shooting picture is facing the center of the tower are obtained, and the tower center image I2 is obtained. At this time, the position of the tower center point in the tower center image I2 is (μ2, v2);

[0018] 104) Based on the longitude and latitude (lng1, lat1) of the drone's position A, the deflection angle c of the wind turbine relative to the drone's nose and the longitude and latitude of the drone's position F (lng2, lat2), the longitude and latitude (lng3, lat3) corresponding to the position D of the wind turbine tower are calculated.

[0019] Step S2 specifically includes the following steps:

[0020] 201) The UAV continues to fly upwards so that there are no obstacles above the UAV. During the ascent, the image transmission image and the UAV height are used as parameters to identify the cabin and the hub; 202) When the cabin recognition detects the appearance of a target in the image, the UAV height is corrected, and the center point of the cabin is located in the center of the image. The corresponding UAV height is the height h'1 of the center of the wind turbine hub after correction.

[0021] S3 specifically includes the following steps:

[0022] 301) Input the wind turbine top view into the wind turbine yaw angle classification model to obtain the output category of the wind turbine yaw angle and the corresponding confidence conf α1 , and converted to the interval [-180°, 180°] according to the category, and the angle α1 of the classified estimated yaw angle is obtained, where

[0023] i is the output category of the wind turbine yaw angle, i is a positive integer in the interval [0, M-1]. The top view of the wind turbine is divided into M categories (-180 degrees to 180 degrees), and the angle interval corresponding to each category is k degrees. Based on the wind turbine yaw angle classification model, the current top view is input to obtain the output category, and the angle interval corresponding to the category is calculated. After the category is output, the angle interval is calculated.

[0024] 302) using a segmentation algorithm to segment the top view of the wind turbine into a binary image containing only the wind turbine area, and clustering all straight lines according to their angles;

[0025] 303) Perform straight line detection on the wind turbine top view, exclude the straight lines that intersect with the edge of the picture, obtain two straight lines representing the nacelle and the wind turbine blades respectively, take the hub area as the coordinate origin, and the lower left as the positive direction of the x-axis and y-axis, and calculate the angle between the x-axis and the straight line where the nacelle is located, which is the yaw angle α2 of the wind turbine on the wind turbine top view;

[0026] 304) The hub position is determined based on the intersection of the two straight lines, and the yaw angle α' of the cabin is calculated based on the hub as the origin. Based on α1 and α2, the comprehensive confidence of the cabin yaw angle is calculated as conf α =0.5*(conf α1 +conf α2 ). Let x = |α1-α2| where When the values ​​of α1 and α2 are equal, the comprehensive confidence of the cabin yaw angle is the highest. When the difference between α1 and α2 is 90°, the comprehensive confidence of the cabin yaw angle is the lowest. If the comprehensive confidence is greater than the set cabin yaw angle threshold, the cabin yaw angle α'=α1 based on the hub as the origin; if it is less than the cabin yaw angle threshold, the cabin yaw angle α'=α2 based on the hub as the origin; conf α1 Represents the comprehensive confidence (probability) of the cabin yaw angle given by the classification algorithm;

[0027] 305) Based on the yaw angle α' of the cabin when the hub is the origin, add the yaw angle ρ of the pan / tilt to obtain the corrected output value α of the fan yaw angle pan / tilt out , convert the wind turbine yaw angle output value into a fixed interval to obtain the wind turbine yaw angle α.

[0028] Step 305) specifically includes the following steps: Taking due north as 0°, the yaw angle of the PTZ is ρ, then the yaw angle of the wind turbine PTZ correction output value α out =90°+α'+ρ, the yaw angle of the wind turbine is corrected by the output value α out Make limit correction to obtain the yaw angle of the wind turbine Limit the wind turbine yaw angle α to (-180,180].

[0029] The step S5 of calibrating the wheel hub center position specifically includes the following steps:

[0030] 501), no one flies to the front of the hub, and there are no other obstacles during the flight;

[0031] 502), the camera captures the image, performs wheel recognition, and obtains the wheel image I3;

[0032] 503), obtain the center point (x1, y1) of the wheel hub recognition frame, compare it with the center position (x2, y2) of the wheel hub image I3, and obtain the flight direction of the controlled drone and distance Get the front view I4 of the fan hub;

[0033] 504), control the UAV to fly based on the flight direction and distance of step 503), and execute steps 502) and 503) repeatedly until the distance between the center point of the picture and the center point of the hub identification frame is less than the set threshold. At this time, the position of the UAV is the waypoint directly in front of the calibrated hub center.

[0034] Step S6 calculates the fan blade phase angle and specifically comprises the following steps:

[0035] 601a) The fan hub front view is divided into U categories, each category corresponds to an angle interval of K degrees, and a blade phase angle classification model is trained; after obtaining the fan hub front view, it is input into the blade phase angle classification model to obtain the output category of the fan blade phase angle and the corresponding confidence conf β1 , and converted to the angle interval according to the category, and the classified estimated blade phase angle β1=i*K is obtained, where i is the output category of the fan blade phase angle and i is a positive integer in the interval [0,U-1];

[0036] 602a) inputting the current front view based on the blade phase angle classification model, obtaining the output category, and calculating the corresponding angle interval;

[0037] 603a) searching for blades from the four sides of the fan hub front view toward the center in sequence, and obtaining 3 or 4 endpoints where the fan body extends to the edge of the image;

[0038] 604a) Take the hub as the coordinate origin, connect the hub and the endpoint of the image edge to calculate the angle, set the hub position in the fan hub front view I4 to (μ3, v3), and the endpoint to (μ4, v4), then the angle is

[0039] 605a) Exclude the straight line of the tower part and select the angle β2 between the first blade and the tower in the counterclockwise direction. The confidence of the angle β2 is conf β2 ;

[0040] 606a) calculate the fan blade phase angle β,

[0041] Based on β1 and β2, calculate the comprehensive confidence conf of the fan blade phase angle β =0.5*(conf β1 +conf β2 ). Let x = |β1-β2| where When the values ​​of β1 and β2 are the same, the comprehensive confidence is the highest, and when the difference between β1 and β2 is 60°, the comprehensive confidence is the lowest. If the comprehensive confidence is greater than the threshold of the fan blade phase angle, the algorithm output is β = β1, and when it is less than the threshold of the fan blade phase angle, the algorithm output is β = β2.

[0042] The calculation of blade pre-bending in S6 specifically includes the following steps:

[0043] 601b) The front view of the fan hub is segmented into blades, and a two-dimensional rectangular coordinate system is established with the hub center as the origin, the phase angle direction as the positive direction of the x-axis, and 90 degrees counterclockwise of the x-axis as the positive direction of the y-axis.

[0044] 602b) taking points along the positive direction of the x-axis according to the set spacing, and calculating the y-value corresponding to the leaf area to generate a number of data points;

[0045] According to the phase angle β of the fan blade, a two-dimensional rectangular coordinate system is established with the hub center as the origin, the phase angle direction as the positive direction of the x-axis, and 90 degrees counterclockwise of the x-axis as the positive direction of the y-axis.

[0046] 603b) curve fitting all data points, and converting the obtained curve analytical function from the two-dimensional rectangular coordinate system Convert to hub rectangular coordinates Two-dimensional rectangular coordinate system is the coordinate system rotated according to the phase angle;

[0047] Because the pre-bending of the three blades is the same, we only need to find the pre-bending analytical formula of one blade. In , points are taken along the positive direction of the x-axis according to the set spacing, and the mean value of the y-value corresponding to the leaf area is calculated to generate several data points.

[0048] 604b) Use the four-parameter equation Curve fitting is performed, where d represents the baseline of blade pre-bending, a represents the maximum value of blade pre-bending, c represents the position parameter of the inflection point, and b represents the slope parameter of the curve; the data points are brought in, and the expected error threshold ε is set. During the iteration process, the mean error ε1 between the true value and the estimated value of all points is counted, and the parameters are adjusted according to the error mean ε1 until the error mean is controlled within the expected threshold, ε1<ε, and the corresponding analytical expression is the equation for curve fitting; since the blade pre-bending causes the blade to bend, the blade pre-bending is represented by the curve equation.

[0049] Step S7 specifically includes the following steps:

[0050] 701), based on the wind turbine yaw angle α, wind turbine blade phase angle β, the wind turbine's inherent parameters impeller pitch angle θ and blade cone angle Calculate the inspection route: In the hub rectangular coordinates Under this condition, the angle of the first blade becomes β'=270°-β. According to the impeller pitch angle, the corresponding rotation matrix in this coordinate system is: Get the first blade direction vector The angles between the other two blades and the first blade are known, and the blade pre-bending design is the same. The direction vectors of the second and third blades are obtained by replacing the blade angles; the coordinate mapping expressions before and after the conversion are substituted into the curve equation of the blade pre-bending to obtain the rectangular coordinates of each blade at the hub. The curve fitting analytical formula below;

[0051] 702), the blade length l is known, in the coordinate system When inspecting the leading edge, the blade pre-bend line equation obtained by coordinate system transformation is The x variable is in [0,l*v β ] and divide the values ​​into n equal parts to generate n coordinate systems. Waypoints where ε g is the distance between the UAV and the blade when inspecting the leading edge of the blade. The UAV passes through the waypoints in sequence to obtain the blade inspection route;

[0052] When inspecting the trailing edge, the waypoint is ε h is the distance from the blade when inspecting the trailing edge of the blade; for the leeward side, the waypoint is The windward waypoint is ε m It is the distance between the windward side of the blade and the blade when inspecting it.

[0053] 703), combined with the longitude and latitude of the wind turbine tower (lng3, lat3), generate a route starting from the initial position of the drone; the wind turbine yaw angle α and the corrected hub height h'1 are known, and the blade pre-bend is based on the hub rectangular coordinate system The analytical expression of the hub rectangular coordinate system Convert to the tower rectangular coordinate system based on the bottom of the wind turbine tower The corresponding transformation matrix is: The tower rectangular coordinate system Then convert it into longitude, latitude and altitude; transform the coordinate system and add the influence of the wheel center point height and yaw angle.

[0054] 704), when inspecting the fan blades, inspect the leading edge, trailing edge, windward side (pressure side) and leeward side (suction side) of the blades.

[0055] A computing system comprises: one or more processors, a memory and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include instructions for executing any one of the methods according to claims 1 to 9.

[0056] The beneficial effects achieved by the present invention are:

[0057] The present invention discloses a method for generating a wind turbine inspection route based on a drone. The method does not require input of wind turbine parameters. Through a wind turbine component (tower, chassis, hub, etc.) recognition algorithm and a yaw angle classification model, an accurate inspection route is automatically planned, which includes waypoints on the inspection route, the nose direction of the drone, the gimbal angle and photo-taking actions at each waypoint, etc. No human intervention is required during the entire flight process, and the method can be used for drone wind turbine blade inspection tasks in any scenario.

[0058] This application uses a completely autonomous flight method to automatically re-measure the longitude and latitude of the wind turbine and the altitude of the center point of the hub every time an inspection is carried out to avoid errors caused by longitude and latitude. A classification model is added when identifying the yaw angle of the wind turbine and the phase angle of the blades, and the recognition rate is improved through redundancy. After arriving in front of the hub, the final correction is performed to ensure that the position of the wind turbine is not deviated, which solves the problem of deviation in measurement and longitude and latitude in the prior art patents. At the same time, the curve fitting equation of the blade pre-bend is introduced to solve the problem that the blade will exceed the camera screen when inspecting the leading and trailing edges. In steps S1 and S2, this application calculates the longitude and latitude of the tower and the height of the hub through automatic real-time recognition and multi-view geometry methods, effectively avoiding the error problems that may occur when relying on GPS positioning. This method makes the position calculation more accurate and lays a solid foundation for subsequent operations. In step S3, this application combines traditional image processing technology with advanced deep learning classifier algorithms to calculate the yaw angle of the wind turbine. This hybrid method significantly improves the robustness of yaw angle recognition and maintains a high degree of accuracy even in complex environments. In step S5, the position of the drone is adjusted through real-time recognition technology to ensure that the wind turbine hub is always in the center of the picture. This step not only further corrects the height and position of the hub center point, but also effectively eliminates the errors that may occur in the previous calculation process, improving the overall accuracy. In step S7, combined with camera parameters such as blade pre-bending, an inspection route for the four surfaces corresponding to a blade is generated, and then the remaining inspection routes can be generated by simply modifying the corresponding blade phase angle, making the route more universal and adaptable to different wind turbine shutdown angles. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1This is a schematic diagram of the positions of the drone and wind turbine tower when detecting longitude and latitude waypoints;

[0060] FIG. 2( a ) is a schematic diagram of a plane for calculating the latitude and longitude azimuth distance;

[0061] Figure 2(b) is a schematic diagram of calculating the latitude and longitude azimuth distance longitude and latitude;

[0062] Figure 3 It is a schematic diagram of the normalized imaging plane of the tower;

[0063] Figure 4 (a) is a top view of the fan;

[0064] Figure 4 (b) Binary image of the fan top view;

[0065] Figure 4(c) shows the windward angle identification process of the wind turbine;

[0066] Figure 5 (a) is a front view of a fan hub;

[0067] Figure 5(b) Binary image of the front view of the fan hub;

[0068] Fig. 5(c) Wind turbine blade phase angle identification process;

[0069] Figure 6 (a) Schematic diagram of fan blade phase angle;

[0070] The binary image corresponding to the blade in Figure 6(b) rotates clockwise to establish a two-dimensional rectangular coordinate system

[0071] Figure 7 Wheel hub rectangular coordinate system

[0072] Figure 8 Tower rectangular coordinate system based on the bottom of the wind turbine tower

[0073] Fig. 9 Blade pre-bending waypoint generation;

[0074] Figure 10 (a) Leading and trailing edge routes;

[0075] Figure 10(b) Windward and leeward routes. DETAILED DESCRIPTION

[0076] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0077] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0078] A method for generating a wind turbine inspection route based on a drone, wherein the wind turbine comprises a wind tower and an impeller arranged at the top of the wind tower, wherein the impeller comprises a hub and three blades evenly distributed along the circumference of the hub, and the drone is equipped with a pan / tilt with a camera and a GPS positioning system, and comprises the following steps:

[0079] S1, define the longitude and latitude waypoints of the wind turbine, the longitude and latitude waypoints of the wind turbine include the first longitude and latitude waypoint of the wind turbine and the second longitude and latitude waypoint of the wind turbine, the longitude and latitude of the first longitude and latitude waypoint of the wind turbine are the longitude and latitude (lng1, lat1) of the drone, the height of the first longitude and latitude waypoint of the wind turbine is higher than the surrounding obstacles, adjust the gimbal to shoot horizontally, rotate the drone head in situ to adjust the direction of the wind turbine tower so that the wind turbine tower appears completely in the picture captured by the camera; at this time, the position of the drone is A, and the longitude and latitude are (lng1, lat1); the height of the longitude and latitude waypoint of the second wind turbine is the same as that of the first wind turbine, keep the shooting posture and the direction of the head unchanged, the longitude and latitude of the second wind turbine longitude and latitude waypoint is the longitude and latitude corresponding to when the camera shooting picture faces the center of the tower, at this time the position of the drone is F, the longitude and latitude are (lng2, lat2), calculate the longitude and latitude of the wind turbine tower (lng3, lat3), and the position of the wind turbine tower is D;

[0080] S2, based on the acquired longitude and latitude of the wind turbine tower (lng3, lat3), generate a route for the drone to reach the top of the wind turbine, wherein the route for the drone to reach the top of the wind turbine includes four waypoints; the first waypoint is the position F where the drone is located; the longitude and latitude of the second waypoint are the same as those of the first waypoint (lng1, lat1), and the height is the height h1 of the center of the wind turbine hub; the longitude and latitude of the third waypoint are (lng1, lat1), and the height h2 is greater than the sum of the height h1 of the center of the wind turbine hub and the length l of the wind turbine blade; the longitude and latitude of the fourth waypoint are the longitude and latitude of the wind turbine tower (lng3, lat3), and the height of the fourth waypoint is the same as that of the third waypoint, which is h2; after arriving at the fourth waypoint, the gimbal is adjusted to shoot vertically downward to obtain a bird's-eye view of the wind turbine;

[0081] S3, calculating the yaw angle α of the wind turbine based on the top view of the wind turbine;

[0082] S4, based on the calculated yaw angle α of the wind turbine, generate an initial route to the front of the hub; the initial route to the front of the hub includes two waypoints, the first waypoint of the initial route to the front of the hub has a height of h2, and the longitude and latitude are the longitude and latitude of the position in front of the hub; the longitude and latitude of the second waypoint of the initial route to the front of the hub are the same as those of the first waypoint of the initial route to the front of the hub, and the height is the height h1 of the center of the hub, and the position of the second waypoint of the initial route to the front of the hub is Q, and when arriving at position Q, the gimbal is adjusted to horizontal shooting, and the nose direction is adjusted to the yaw angle of the wind turbine -180°;

[0083] S5, as shown in FIG5, obtains a front view of the wind turbine hub from the camera in the gimbal, and generates a calibrated hub front route based on the hub recognition algorithm. The calibrated hub front route includes two waypoints. The first waypoint of the calibrated hub front route is position Q, the nose direction and shooting posture remain unchanged, and the second waypoint of the calibrated hub front route is the center of the hub located at the center of the picture taken by the camera, and the wind turbine hub front view is obtained based on the second waypoint of the calibrated hub front route;

[0084] S6, calculating the fan blade phase angle β and blade pre-bend based on the hub front view obtained in step S5;

[0085] S7, generating an inspection route based on the longitude and latitude (lng3, lat3) of the wind turbine tower, the wind turbine yaw angle α, the wind turbine blade phase angle β, the tower height, the blade pre-bend and the wind turbine blade length l.

[0086] Calculating the longitude and latitude (lng3, lat3) of the wind turbine tower in S1 specifically includes the following steps:

[0087] 101) At the first wind turbine longitude and latitude waypoint, based on the tower recognition algorithm, the wind turbine tower appears completely in the picture acquired by the camera, and the wind turbine tower image I1 of the first wind turbine longitude and latitude waypoint is acquired. At this time, the position of the tower center point in the wind turbine tower image I1 is (μ1, v1);

[0088] 102) Based on the wind turbine tower image I1 and combined with the camera internal parameters, the deflection angle c of the wind turbine relative to the direction of the UAV nose is calculated;

[0089] 103) The flight direction is perpendicular to the nose of the aircraft, the shooting posture and the direction of the nose are kept unchanged, the latitude and longitude waypoints of the second wind turbine corresponding to the time when the camera shooting picture is facing the center of the tower are obtained, and the tower center image I2 is obtained. At this time, the position of the tower center point in the tower center image I2 is (μ2, v2);

[0090] 104) Based on the longitude and latitude (lng1, lat1) of the drone's position A, the deflection angle c of the wind turbine relative to the drone's nose direction and the longitude and latitude of the drone's position F (lng2, lat2), calculate the longitude and latitude (lng3, lat3) corresponding to the position D where the wind turbine tower is located;

[0091] A is the camera position of the drone, corresponding to the longitude and latitude (lng1, lat1), D is the center point of the tower, and the position of the tower on the normalized imaging plane of the wind turbine tower image I1 is C; the position of the tower center in the image is ((μ1, v1)), and the camera internal parameters (f x ,f y ,c x ,cy ), then the coordinates of point C (x c ,y c ,1) is: Specifically, f x and f y It is the product of the focal length F of the camera and the physical dimensions dx and dy of a single pixel in the x-axis and y-axis directions, that is, f x =F*dx,f y =F*dy, dx and dy represent the actual length units occupied by a pixel in the x direction and y direction respectively. x and c y The table shows the x-axis and y-axis translation of the origin. Taking A as the origin, the coordinates of A are (0,0,0). On the two-dimensional plane, B is the point on the normalized imaging plane where the center of the image is. According to the definition of the normalized imaging plane, the distance L between A and B is AB =1, so ∠BAC = atan(L BC ).like Figure 1 As shown, A is the location of the first wind turbine latitude and longitude waypoint, B is the location of the center point of the picture on the normalized imaging plane at this time, and C is the location of the tower on the normalized imaging plane ( Figure 3 shown) on the position;

[0092] Keep the direction of the drone's nose unchanged, and fly the drone in a direction perpendicular to the nose to the second waypoint, such as Figure 1 As shown in the figure, F is the location of the drone camera corresponding to the waypoint, and the center of the tower is at the center of the picture. The longitude and latitude corresponding to the drone are (lng2, lat2). According to the longitude and latitude of A and F and ∠BAC, the real distance between A and D is calculated. Get the azimuth angle a of point D relative to point A.

[0093] As shown in Figure 2(a), the azimuth of point D relative to point A is known to be a, and the actual distance between AD and point A is L AD , then the translation distances from point A to point D are L AD *sin(a),L AD *cos(a), the longitude and latitude of point A is known to be (lng1, lat1), then as shown in Figure 2(b), the radius of the current latitude is arc = ARC*cos(lat1), where the average radius of the earth is 6371.393 km (this number is the average distance from the center of the earth to all points on the earth's surface). Then the longitude and latitude of point D is (lng3, lat3), where lng3 = lng1 + L AD *sin(a) / [ARC*cos(lat1)*2π / 360],lat3=lat1+L AD*cos(a) / (ARC*2π / 360).

[0094] Step S2 specifically includes the following steps:

[0095] 201) The drone continues to fly upwards so that there are no obstacles above the drone (such as wind turbine blades, wires, etc.). During the ascent, the image transmission image and the drone altitude are used as parameters to identify the cabin and the wheel hub;

[0096] 202) When the cabin recognition detects that a target appears in the picture, the height of the drone is corrected, the center point of the cabin is located at the center of the picture, and the corresponding drone height is the height h'1 of the center of the wind turbine hub after correction.

[0097] The specific steps in S3 include:

[0098] 301) Input the wind turbine top view into the wind turbine yaw angle classification model to obtain the corresponding output category and the corresponding confidence conf α1 , and converted to the interval [-180°, 180°] according to the category, and the angle α1 of the classified estimated yaw angle is obtained, where i is the output category. In this embodiment, i is a positive integer in the interval [0,71]. In this embodiment, the wind turbine top view is divided into 72 categories (-180 degrees to 180 degrees), and the angle interval corresponding to each category is 5 degrees. Based on the wind turbine yaw angle classification model, the current top view is input, the output category is obtained, and the angle interval corresponding to the category is calculated. After the category is output, the angle interval is calculated. In this embodiment, the category identified by the wind turbine yaw angle is 2, and the wind turbine yaw angle is between 10 and 15 degrees.

[0099] 302) The top view of the wind turbine is segmented into a binary image containing only the wind turbine area using a segmentation algorithm, and all straight lines are clustered according to their angles. As shown in FIG4(b), two straight lines are finally obtained, and the position corresponding to the hub area can be determined according to the intersection of the straight lines.

[0100] 303) Perform straight line detection on the wind turbine top view, exclude the straight lines that intersect with the edge of the picture, obtain two straight lines representing the cabin and the wind turbine blades respectively, take the hub area as the coordinate origin, and the lower left as the positive direction of the x and y axes, calculate the angle between the x axis and the straight line where the cabin is located, which is the yaw angle α2 of the wind turbine on the wind turbine top view. 304) Determine the hub position based on the intersection of the two straight lines, and calculate the cabin yaw angle α' based on the hub as the origin. Based on α1 and α2, the comprehensive confidence of the cabin yaw angle is calculated as conf α =0.5*(conf α1 +conf α2 ). Let x = |α1-α2| where When the values ​​of α1 and α2 are equal, the comprehensive confidence of the cabin yaw angle is the highest. When the difference between α1 and α2 is 90°, the comprehensive confidence of the cabin yaw angle is the lowest. If the comprehensive confidence is greater than the set cabin yaw angle threshold, the cabin yaw angle α'=α1. When it is less than the cabin yaw angle threshold, the cabin yaw angle α'=α2. α1 Represents the comprehensive confidence (probability) of the cabin yaw angle given by the classification algorithm;

[0101] 305) According to the cabin yaw angle α', plus the yaw angle ρ of the PTZ, the wind turbine yaw angle PTZ correction output value α is obtained out And convert the wind turbine yaw angle output value into a fixed interval to obtain the wind turbine yaw angle α.

[0102] Step 305) specifically includes the following steps: Taking due north as 0°, the yaw angle of the PTZ is ρ, then the yaw angle of the wind turbine PTZ correction output value α out =90°+α'+ρ, the yaw angle of the wind turbine is corrected by the output value α out Make limit correction to obtain the yaw angle of the wind turbine Clamp the result to (-180,180].

[0103] The step S5 of calibrating the wheel hub center position specifically includes the following steps:

[0104] 501), no one flies to the front of the hub, and there are no other obstacles during the flight;

[0105] 502), the camera captures the image, performs wheel recognition, and obtains the wheel image I3;

[0106] 503), obtain the center point (x1, y1) of the wheel hub recognition frame, compare it with the center position (x2, y2) of the wheel hub image I3, and obtain the flight direction of the controlled drone and distance Obtain the front view I4 of the fan hub.

[0107] In this embodiment, the width of the image captured by the camera is w, and the height is h, then the center of the image is x2=w / 2, y2=h / 2. Because the hub should appear in the center of the image, after the hub is identified in the image, the deviation between the hub and the center of the image is calculated; the position of the drone is adjusted according to the deviation value so that the hub is in the center of the front view of the wind turbine hub.

[0108] 504), control the UAV to fly based on the flight direction and distance of step 503), and execute steps 502) and 503) repeatedly until the distance between the center point of the picture and the center point of the hub identification frame is less than the set threshold. At this time, the position of the UAV is the waypoint directly in front of the calibrated hub center.

[0109] Step S6 calculates the fan blade phase angle and specifically comprises the following steps:

[0110] 601a) In this embodiment, the front view of the wind turbine hub is divided into 24 categories (0 degrees to 120 degrees), and the angle interval corresponding to each category is 5 degrees, and the blade phase angle classification model is trained. After obtaining the front view of the wind turbine hub, it is input into the blade phase angle classification model to obtain the corresponding output category and the corresponding confidence conf β1 , and converted into the interval [0°, 120°] according to the category, the classified estimated blade phase angle β1 = i*5 is obtained, where i is the output category and is a positive integer in the interval [0, 23].

[0111] like Figure 5(b) and 5(c) After obtaining the front view of the wind turbine hub, the center of the hub is the center of the picture. In order to reduce the influence of other parts, the segmentation algorithm is used to obtain a binary image containing only the wind turbine area. Based on the binary image, the blades are searched from the four sides to the center in sequence to obtain the intersection points of the blades or tower extending to the four sides of the image, usually 4 intersection points. When the wind turbine is in or approximately in a positive 'Y' or reverse 'Y' shutdown, one of the blades is in a vertical state and coincides with the wind turbine tower, resulting in an intersection of 3.

[0112] 602a) Inputting the current front view based on the blade phase angle classification model, obtaining the output category, and calculating the corresponding angle interval.

[0113] 603a) Searching for blades from the four sides of the fan hub front view toward the center in sequence, and obtaining 3 or 4 endpoints where the fan body extends to the edge of the image.

[0114] 604a) Take the hub as the coordinate origin, connect the hub and the endpoint of the image edge to calculate the angle, set the hub position in the fan hub front view I4 to (μ3, v3), and the endpoint to (μ4, v4), then the angle is

[0115] Take the hub center as the origin and connect the origin with each intersection point respectively.

[0116] 605a) Given that the angle between the three blades is 120°, exclude the straight line of the tower part and select the angle β2 between the first blade and the tower in the counterclockwise direction. The confidence of angle β2 is conf β2 ;

[0117] 606a) calculates the fan blade phase angle β based on the results of 602a) and 605a), Figure 6(a).

[0118] Based on β1 and β2, calculate the comprehensive confidence conf of the fan blade phase angle β =0.5*(confβ1 +conf β2 ). Let x = |β1-β2| where When the values ​​of β1 and β2 are the same, the comprehensive confidence is the highest, and when the difference between β1 and β2 is 60°, the comprehensive confidence is the lowest. If the comprehensive confidence is greater than the threshold of the fan blade phase angle, the algorithm output is β = β1, and when it is less than the threshold of the fan blade phase angle, the algorithm output is β = β2.

[0119] The calculation of blade pre-bending in S6 specifically includes the following steps:

[0120] 601b) The front view of the fan hub is segmented into blades, and a two-dimensional rectangular coordinate system is established with the hub center as the origin, the phase angle direction as the positive direction of the x-axis, and 90 degrees counterclockwise of the x-axis as the positive direction of the y-axis.

[0121] 602b) Take points along the positive direction of the x-axis according to the set spacing, and calculate the y value corresponding to the leaf area to generate a number of data points.

[0122] According to the phase angle β of the fan blade, as shown in Figure 6(b), a two-dimensional rectangular coordinate system is established with the hub center as the origin, the phase angle direction as the positive direction of the x-axis, and 90 degrees counterclockwise from the x-axis as the positive direction of the y-axis.

[0123] 603b) Curve fit all data points, such as Figure 7 As shown, the obtained curve analytical function is converted from a two-dimensional rectangular coordinate system Convert to hub rectangular coordinates Two-dimensional rectangular coordinate system It is a coordinate system after the phase angle rotation. In this embodiment, the blade direction is set to the x-axis for the convenience of calculation and value taking.

[0124] Because the pre-bending of the three blades is the same, we only need to find the pre-bending analytical formula of one blade. In , points are taken along the positive direction of the x-axis according to the set spacing, and the mean value of the y-value corresponding to the leaf area is calculated to generate several data points.

[0125] 604b) Use the four-parameter equation Perform curve fitting, where d represents the baseline of blade pre-bending, a represents the maximum value of blade pre-bending, c represents the position parameter of the inflection point, and b represents the slope parameter of the curve. This expression is based on the shape of the S-shaped curve. When x is small, the value of the denominator is close to 1, and the curve is close to the baseline d; when x is large, the value of the denominator approaches infinity, and the curve is close to the maximum value a. The c parameter controls the position of the inflection point of the curve, and the d parameter controls the slope of the curve. Bring in all data points and set the expected threshold ε. During the iteration process, the mean error ε1 between the true value and the estimated value of all points is counted, and the parameters are adjusted according to the error mean ε1 until the error mean is controlled within the expected threshold, ε1<ε. At this time, the corresponding analytical expression is the equation for curve fitting; since the blade pre-bending causes the blade to bend, the blade pre-bending is represented by the curve equation. ε represents the expected error threshold.

[0126] Step S7 specifically includes the following steps:

[0127] 701), based on the wind turbine yaw angle α, wind turbine blade phase angle β, the wind turbine's inherent parameters impeller pitch angle θ and blade cone angle To calculate the inspection route. Establish a two-dimensional rectangular coordinate system according to the blade phase angle direction This is to generate several data points to be fitted, and to convert the blade coordinate system to a unified hub rectangular coordinate system in the subsequent process of generating the route. For the convenience of calculation, we use Figure 7 The figure is based on the hub rectangular coordinate system In the hub rectangular coordinates Under this condition, the angle of the first blade becomes β'=270°-β. According to the impeller pitch angle, the corresponding rotation matrix in this coordinate system is: Get the first blade direction vector The angles between the other two blades and the first blade are known, and the blade pre-bending design is the same. The direction vectors of the second and third blades are obtained by replacing the blade angles. Substituting the coordinate mapping expressions before and after the conversion into the curve equation of the blade pre-bending, the curve fitting analytical expression of each blade in the hub coordinate system is obtained.

[0128] Two-dimensional rectangular coordinate system It is a two-dimensional coordinate system, the hub rectangular coordinates and tower rectangular coordinate system They are all three-dimensional coordinate systems.

[0129] 702), the blade length l is known, in the coordinate system When inspecting the leading edge, the blade pre-bend line equation z obtained by coordinate system transformation is The x-axis and y-axis correspond to The y-axis and z-axis of the x variable are in [0,l*v β ] is divided into n equal parts, such as Fig. 9 As shown, n coordinate systems can be generated Waypoints where v β is the first blade direction vector, ε g is the distance between the UAV and the blade when inspecting the leading edge of the blade. The UAV passes through these waypoints in sequence to obtain the blade inspection route. Similarly, when inspecting the trailing edge, the waypoint is ε h is the distance from the blade when inspecting the trailing edge of the blade. For the leeward side, the waypoint is The windward waypoint is ε m is the distance from the windward side of the blade when inspecting it. In this case, x can be set to 0.

[0130] 703), establish the tower rectangular coordinate system based on the bottom of the wind turbine tower Combined with the longitude and latitude of the wind turbine tower (lng3, lat3), a route is generated starting from the initial position of the drone, as shown in Figure 10. The wind turbine yaw angle α and the corrected hub height h'1 are known, and the blade pre-bend is based on the hub rectangular coordinate system. The analytical expression of the hub rectangular coordinate system Convert to the tower rectangular coordinate system based on the bottom of the wind turbine tower The corresponding transformation matrix is ​​as follows: All waypoints are represented by projection coordinates. In this embodiment, waypoint x = [x0, x1, x2] T , the corresponding projection coordinates are x'=[x0,x1,x2,1] T . Convert to the tower rectangular coordinate system Then convert it to longitude and latitude. Convert the coordinate system and add the influence of the hub center point height and yaw angle. This application uses ENU to convert the tower rectangular coordinate system Reconvert to latitude, longitude and altitude.

[0131] 704), when inspecting the wind turbine blades, each blade needs to be inspected for the leading edge, trailing edge, windward side (pressure side) and leeward side (suction side). In this embodiment, the blades are numbered as blade 1, blade 2 and blade 3, respectively, and the inspection order can be as follows: blade 1-leading edge, blade 1-trailing edge, blade 1-leeward side, blade 1-windward side, blade 2-leading edge, blade 2-trailing edge, blade 2-leeward side, blade 2-windward side, blade 3-leading edge, blade 3-trailing edge, blade 3-leeward side, blade 3-windward side. Among them, the inspection generally starts from the front of the hub center, as shown in G1 of Figure 10 (a), and arrives at G2 after inspecting blade 1-leading edge, and the waypoint used is the waypoint v calculated in 2. g After safely reaching the trailing edge tip of blade 1 through waypoints G3, G4, and G5, start inspecting the trailing edge of blade 1, using waypoint v h Arrive at G6. Then, go to the starting position of the inspection blade 1-leeward side through G6, G7 and H1 directly behind the cabin, and start the inspection from the root of the blade from H1 to H2, as shown in Figure 10(b). Go through H2, H3, H4, H5 to the top of the windward side. Then, go through H5 and H6 to complete the inspection, using the waypoint v r For blade 2 and blade 3, the corresponding blade phase angles are modified to calculate the corresponding waypoints.

[0132] A computing system comprises: one or more processors, a memory and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include instructions for executing any one of the methods according to claims 1 to 9.

[0133] The above are only preferred embodiments of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

[0134] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

[0135] In the description provided herein, a large number of specific details are described. However, it is understood that embodiments of the present invention can be practiced without these specific details. In some instances, well-known methods, structures and techniques are not shown in detail so as not to obscure the understanding of this description.

[0136] Similarly, it should be understood that in order to streamline the present disclosure and aid in understanding one or more of the various inventive aspects, in the above description of exemplary embodiments of the present invention, various features of the present invention are sometimes grouped together into a single embodiment, figure, or description thereof. However, this disclosed method should not be interpreted as reflecting the intention that the claimed invention requires more features than those expressly recited in each claim. Rather, as reflected in the claims, inventive aspects lie in less than all of the features of the individual embodiments previously disclosed. Therefore, the claims that follow the detailed description are hereby expressly incorporated into the detailed description, with each claim itself serving as a separate embodiment of the present invention.

[0137] Those skilled in the art will appreciate that the modules or units or groups of the devices in the examples disclosed herein may be arranged in the device as described in the embodiment, or alternatively may be located in one or more devices different from the devices in the example. The modules in the foregoing examples may be combined into one module or may be divided into multiple submodules.

[0138] Those skilled in the art will appreciate that the modules in the devices in the embodiments may be adaptively changed and arranged in one or more devices different from the embodiments. The modules or units or groups in the embodiments may be combined into one module or unit or group, and in addition they may be divided into a plurality of submodules or subunits or subgroups. All features disclosed in this specification (including the accompanying claims, abstracts and drawings) and all processes or units of any method or device so disclosed may be combined in any combination, except that at least some of such features and / or processes or units are mutually exclusive. Unless otherwise expressly stated, each feature disclosed in this specification (including the accompanying claims, abstracts and drawings) may be replaced by an alternative feature providing the same, equivalent or similar purpose.

[0139] In addition, those skilled in the art will appreciate that, although some embodiments described herein include certain features included in other embodiments but not other features, the combination of features of different embodiments is meant to be within the scope of the present invention and form different embodiments. For example, in the claims below, any one of the claimed embodiments may be used in any combination.

[0140] In addition, some of the embodiments are described herein as methods or combinations of method elements that can be implemented by a processor of a computer system or by other devices that perform the functions. Therefore, a processor with necessary instructions for implementing the method or method elements forms a device for implementing the method or method elements. In addition, the elements described herein of the device embodiments are examples of devices for implementing the functions performed by the elements for the purpose of implementing the invention.

[0141] The various techniques described herein may be implemented in combination with hardware or software, or a combination thereof. Thus, the method and apparatus of the present invention, or some aspects or portions of the method and apparatus of the present invention may be in the form of program codes (i.e., instructions) embedded in a tangible medium, such as a floppy disk, a CD-ROM, a hard disk drive, or any other machine-readable storage medium, wherein when the program is loaded into a machine such as a computer and executed by the machine, the machine becomes a device for practicing the present invention.

[0142] In the case where the program code is executed on a programmable computer, the computing device generally includes a processor, a storage medium readable by the processor (including volatile and non-volatile memory and / or storage elements), at least one input device, and at least one output device. The memory is configured to store the program code; the processor is configured to execute the method of the present invention according to the instructions in the program code stored in the memory.

[0143] By way of example and not limitation, computer readable media include computer storage media and communication media. Computer readable media include computer storage media and communication media. Computer storage media stores information such as computer readable instructions, data structures, program modules or other data. Communication media generally embodies computer readable instructions, data structures, program modules or other data in a modulated data signal such as a carrier wave or other transmission mechanism, and includes any information delivery medium. Any combination of the above is also included within the scope of computer readable media.

[0144] As used herein, unless otherwise specified, the use of ordinal numbers "first," "second," "third," etc. to describe common objects merely indicates that different instances of similar objects are involved, and is not intended to imply that the objects so described must have a given order in time, space, order, or in any other manner.

[0145] Although the present invention has been described according to a limited number of embodiments, it will be apparent to those skilled in the art, with the benefit of the above description, that other embodiments may be envisioned within the scope of the invention thus described. In addition, it should be noted that the language used in this specification is selected primarily for readability and teaching purposes, rather than for explaining or defining the subject matter of the present invention. Therefore, many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the appended claims. The disclosure of the present invention is illustrative, not restrictive, with respect to the scope of the present invention, which is defined by the appended claims.

Claims

1. A method for generating a wind turbine inspection route based on a drone, wherein the wind turbine comprises a wind tower and an impeller arranged at the top of the wind tower, wherein the impeller comprises a hub and three blades evenly distributed along the circumference of the hub, and the drone is equipped with a pan / tilt with a camera and a GPS positioning system, wherein: The following steps are involved: S1, define the longitude and latitude waypoints of the wind turbine, the longitude and latitude waypoints of the wind turbine include the first longitude and latitude waypoint of the wind turbine and the second longitude and latitude waypoint of the wind turbine, the longitude and latitude of the first longitude and latitude waypoint of the wind turbine are the longitude and latitude (lng1, lat1) of the drone, the height of the first longitude and latitude waypoint of the wind turbine is higher than the surrounding obstacles, adjust the gimbal to shoot horizontally, rotate the drone head in situ to adjust the direction of the wind turbine tower so that the wind turbine tower appears completely in the picture captured by the camera; at this time, the position of the drone is A, and the longitude and latitude are (lng1, lat1); the height of the longitude and latitude waypoint of the second wind turbine is the same as that of the first wind turbine, keep the shooting posture and the direction of the head unchanged, the longitude and latitude of the second wind turbine longitude and latitude waypoint is the longitude and latitude corresponding to when the camera shooting picture faces the center of the tower, at this time the position of the drone is F, the longitude and latitude are (lng2, lat2), calculate the longitude and latitude of the wind turbine tower (lng3, lat3), and the position of the wind turbine tower is D; S2, based on the acquired longitude and latitude of the wind turbine tower (lng3, lat3), generate a route for the drone to reach the top of the wind turbine, wherein the route for the drone to reach the top of the wind turbine includes four waypoints; the first waypoint is the position F where the drone is located; the longitude and latitude of the second waypoint are the same as those of the first waypoint (lng1, lat1), and the height is the height h1 of the center of the wind turbine hub; the longitude and latitude of the third waypoint are (lng1, lat1), and the height h2 is greater than the sum of the height h1 of the center of the wind turbine hub and the length l of the wind turbine blade; the longitude and latitude of the fourth waypoint are the longitude and latitude of the wind turbine tower (lng3, lat3), and the height of the fourth waypoint is the same as that of the third waypoint, which is h2; after arriving at the fourth waypoint, the gimbal is adjusted to shoot vertically downward to obtain a bird's-eye view of the wind turbine; S3, calculating the yaw angle α of the wind turbine based on the top view of the wind turbine; S4, generating an initial route to the front of the hub according to the calculated yaw angle α of the wind turbine; The initial route to the front of the hub contains two waypoints. The first waypoint of the initial route to the front of the hub has a height of h2, and its longitude and latitude are the longitude and latitude of the position in front of the hub. The second waypoint of the initial route to the front of the hub has the same longitude and latitude as the first waypoint of the initial route to the front of the hub, and its height is h1, the height of the hub center. The position of the second waypoint of the initial route to the front of the hub is Q. When arriving at position Q, the gimbal is adjusted to horizontal shooting, and the nose direction is adjusted to the wind turbine yaw angle of -180°. S5, obtaining a front view of the wind turbine hub from the camera in the gimbal, and generating a calibrated hub front route based on the hub recognition algorithm; the calibrated hub front route includes two waypoints, the first waypoint of the calibrated hub front route is position Q, the nose direction and shooting posture remain unchanged, the second waypoint of the calibrated hub front route is the center of the hub located at the center of the picture taken by the camera, and the wind turbine hub front view is obtained based on the second waypoint of the calibrated hub front route; S6, calculating the fan blade phase angle β and blade pre-bend based on the hub front view obtained in step S5; S7, generating an inspection route based on the longitude and latitude (lng3, lat3) of the wind turbine tower, the wind turbine yaw angle α, the wind turbine blade phase angle β, the tower height, the blade pre-bend and the wind turbine blade length l; 301) Input the wind turbine top view into the wind turbine yaw angle classification model to obtain the output category of the wind turbine yaw angle and the corresponding confidence conf α1 , and converted to the interval [-180°, 180°] according to the category, and the angle α1 of the classified estimated yaw angle is obtained, where i is the output category of the wind turbine yaw angle, i is a positive integer in the interval [0, M-1], and the wind turbine top view is divided into M categories, and the angle interval corresponding to each category is k degrees; based on the wind turbine yaw angle classification model, the current top view is input, the output category is obtained, and the angle interval corresponding to the category is calculated; after the category is output, the corresponding angle interval is calculated; 302) using a segmentation algorithm to segment the top view of the wind turbine into a binary image containing only the wind turbine area, and clustering all straight lines according to their angles; 303) Perform straight line detection on the wind turbine top view, exclude the straight lines that intersect with the edge of the picture, obtain two straight lines representing the nacelle and the wind turbine blades respectively, take the hub area as the coordinate origin, and the lower left as the positive direction of the x-axis and y-axis, and calculate the angle between the x-axis and the straight line where the nacelle is located, which is the yaw angle α2 of the wind turbine on the wind turbine top view; 304) The hub position is determined according to the intersection of the two straight lines, and the yaw angle α' of the cabin is calculated based on the hub as the origin; based on α1 and α2, the comprehensive confidence of the cabin yaw angle is calculated as conf α =0.5*(conf α1 +conf α2 ); Let x = |α1-α2| When the values ​​of α1 and α2 are equal, the comprehensive confidence of the cabin yaw angle is the highest. When the difference between α1 and α2 is 90°, the comprehensive confidence of the cabin yaw angle is the lowest. If the comprehensive confidence is greater than the set cabin yaw angle threshold, the cabin yaw angle α' based on the hub as the origin is α1. If it is less than the cabin yaw angle threshold, the cabin yaw angle α' based on the hub as the origin is α2. α1 Represents the comprehensive confidence of the cabin yaw angle given by the classification algorithm; 305) Based on the yaw angle α' of the nacelle when the hub is the origin, add the yaw angle ρ of the PTZ to obtain the corrected output value α of the wind turbine yaw angle PTZ out , convert the wind turbine yaw angle output value into a fixed interval to obtain the wind turbine yaw angle α.

2. The method for generating a wind turbine inspection route based on a drone according to claim 1, characterized in that: The calculation of the longitude and latitude (lng3, lat3) of the wind turbine tower in S1 specifically includes the following steps: 101) At the first wind turbine longitude and latitude waypoint, based on the tower recognition algorithm, the wind turbine tower appears completely in the picture acquired by the camera, and the wind turbine tower image I1 of the first wind turbine longitude and latitude waypoint is acquired. At this time, the position of the tower center point in the wind turbine tower image I1 is (μ1, v1); 102) Based on the wind turbine tower image I1 and combined with the camera internal parameters, the deflection angle c of the wind turbine relative to the direction of the UAV nose is calculated; 103) The flight direction is perpendicular to the nose of the aircraft, the shooting posture and the direction of the nose are kept unchanged, the latitude and longitude waypoints of the second wind turbine corresponding to the time when the camera shooting picture is facing the center of the tower are obtained, and the tower center image I2 is obtained. At this time, the position of the tower center point in the tower center image I2 is (μ2, v2); 104) Based on the longitude and latitude (lng1, lat1) of the drone's position A, the deflection angle c of the wind turbine relative to the drone's nose and the longitude and latitude of the drone's position F (lng2, lat2), the longitude and latitude (lng3, lat3) corresponding to the position D of the wind turbine tower are calculated.

3. The method for generating a wind turbine inspection route based on a drone according to claim 1, characterized in that: Step S2 specifically includes the following steps: 201) The drone continues to fly upwards so that there are no obstacles above the drone. During the ascent, the image transmission picture and the drone altitude are used as parameters to identify the cabin and the wheel hub; 202) When the cabin and hub recognition detect that a target appears in the picture, the drone height is corrected, the cabin or the cabin center point is located in the center of the picture, and the corresponding drone height is the height h'1 of the center of the wind turbine hub after correction.

4. The method for generating a wind turbine inspection route based on a drone according to claim 1, characterized in that: Step 305) specifically includes the following steps: Taking due north as 0°, the yaw angle of the PTZ is ρ, then the yaw angle of the wind turbine PTZ correction output value α out =90°+α'+ρ, the yaw angle of the wind turbine is corrected by the output value α out Make limit correction to obtain the yaw angle of the wind turbine Limit the wind turbine yaw angle α to (-180,180].

5. The method for generating a wind turbine inspection route based on a drone according to claim 1, characterized in that: The step S5 of calibrating the wheel hub center position specifically includes the following steps: 501), no one flies to the front of the hub, and there are no other obstacles during the flight; 502), the camera captures the image, performs wheel recognition, and obtains the wheel image I3; 503), obtain the center point (x1, y1) of the wheel hub recognition frame, compare it with the center position (x2, y2) of the wheel hub image I3, and obtain the flight direction of the controlled UAV and distance Obtain the front view I4 of the fan hub; 504), control the UAV to fly based on the flight direction and distance of step 503), and execute steps 502) and 503) repeatedly until the distance between the center point of the picture and the center point of the hub identification frame is less than the set threshold. At this time, the position of the UAV is the waypoint directly in front of the calibrated hub center.

6. The method for generating a wind turbine inspection route based on a drone according to claim 1, characterized in that: Step S6 calculates the fan blade phase angle and specifically comprises the following steps: 601a) The fan hub front view is divided into U categories, each category corresponds to an angle interval of K degrees, and a blade phase angle classification model is trained; after obtaining the fan hub front view, it is input into the blade phase angle classification model to obtain the output category of the fan blade phase angle and the corresponding confidence conf β1 , and converted to the angle interval according to the category, and the classified estimated blade phase angle β1=i*K is obtained, where i is the output category of the fan blade phase angle and i is a positive integer in the interval [0,U-1]; 602a) inputting the current front view based on the blade phase angle classification model, obtaining the output category, and calculating the corresponding angle interval; 603a) searching for blades from the four sides of the fan hub front view toward the center in sequence, and obtaining 3 or 4 endpoints where the fan body extends to the edge of the image; 604a) Take the hub as the coordinate origin, connect the hub and the endpoint of the image edge to calculate the angle, set the hub position in the fan hub front view I4 to (μ3, v3), and the endpoint to (μ4, v4), then the angle is 605a) Exclude the straight line of the tower part and select the angle β2 between the first blade and the tower in the counterclockwise direction. The confidence of the angle β2 is conf β2 ; 606a) calculate the fan blade phase angle β, Based on β1 and β2, calculate the comprehensive confidence conf of the fan blade phase angle β =0.5*(conf β1 +conf β2 ); Let x = |β1-β2| When the values ​​of β1 and β2 are the same, the comprehensive confidence is the highest, and when the difference between β1 and β2 is 60°, the comprehensive confidence is the lowest; if the comprehensive confidence is greater than the set threshold of the fan blade phase angle, the algorithm output result is β=β1, and when it is less than the threshold of the fan blade phase angle, the algorithm output is β=β2.

7. The method for generating a wind turbine inspection route based on a drone according to claim 1, characterized in that: The calculation of blade pre-bending in S6 specifically includes the following steps: 601b) The front view of the fan hub is segmented into blades, and a two-dimensional rectangular coordinate system is established with the hub center as the origin, the phase angle direction as the positive direction of the x-axis, and 90 degrees counterclockwise of the x-axis as the positive direction of the y-axis. 602b) taking points along the positive direction of the x-axis according to the set spacing, and calculating the y-value corresponding to the leaf area to generate a number of data points; 603b) curve fitting all data points, and converting the obtained curve analytical function from the two-dimensional rectangular coordinate system Convert to hub rectangular coordinates Two-dimensional rectangular coordinate system is the coordinate system rotated according to the phase angle; 604b) Use the four-parameter equation Curve fitting is performed, where d represents the baseline of blade pre-bending, a represents the maximum value of blade pre-bending, c represents the position parameter of the inflection point, and b represents the slope parameter of the curve; the data points are brought in, and the expected error threshold ε is set. During the iteration process, the mean error ε1 between the true value and the estimated value of all points is counted, and the parameters are adjusted according to the error mean ε1 until the error mean is controlled within the expected threshold, ε1<ε, at which point the corresponding analytical expression is the equation for curve fitting; since the blade pre-bending causes the blade to bend, the blade pre-bending is represented by the curve equation.

8. The method for generating a wind turbine inspection route based on a drone according to claim 1, characterized in that: Step S7 specifically includes the following steps: 701), based on the wind turbine yaw angle α, wind turbine blade phase angle β, the wind turbine's inherent parameters impeller pitch angle θ and blade cone angle Calculate the inspection route: In the hub rectangular coordinates Under this condition, the angle of the first blade becomes β'=270°-β; according to the impeller pitch angle, in the hub rectangular coordinate The corresponding rotation matrix is: Get the first blade direction vector The angles between the other two blades and the first blade are known, and the blade pre-bends are the same. By replacing the blade angles, the direction vectors of the second and third blades are obtained; the coordinate mapping expressions before and after the conversion are substituted into the curve equation of the blade pre-bend to obtain the rectangular coordinates of each blade at the hub The curve fitting analytical formula below; 702), the blade length l is known, in the coordinate system In the inspection of the leading edge, the blade pre-bend line equation z = f(x) is obtained by coordinate system transformation. , the x variable is in [0,l*v β ] is divided into n equal parts to generate n coordinate systems Waypoints where ε g is the distance between the UAV and the blade when inspecting the leading edge of the blade. The UAV passes through the waypoints in sequence to obtain the blade inspection route; where n represents the x variable in [0, l*v β ] is divided into n equal parts and the generated value is based on the coordinate system The number of waypoints; When inspecting the trailing edge, the waypoint is ε h is the distance from the blade when inspecting the trailing edge of the blade; for the leeward side, the waypoint is The windward waypoint is ε m The distance between the blade and the windward side of the blade when inspecting; 703), combined with the longitude and latitude of the wind turbine tower (lng3, lat3), generate a route starting from the initial position of the drone; the wind turbine yaw angle a and the corrected hub height h'1 are known, and the blade pre-bend is based on the hub rectangular coordinate system The analytical expression of the hub rectangular coordinate system Convert to the tower rectangular coordinate system based on the bottom of the wind turbine tower The corresponding transformation matrix is: The tower rectangular coordinate system Then convert to latitude, longitude and altitude; 704), when inspecting the fan blades, the leading edge, trailing edge, windward side and leeward side of the blades are inspected respectively.

9. A computing system, characterized in that: include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include instructions for executing any one of the methods according to claims 1 to 8.

Images