Path planning method and system for inspecting the front and back sides of wind turbine blades while the turbine is shut down.

Abstract

This invention proposes a path planning method and system for inspecting the front and back of wind turbine blades while the turbine is shut down. The method includes: a camera gimbal acquiring images of the wind turbine hub and blade root within its field of view based on its pitch and rotation angles; identifying the blade root outline and hub position through image segmentation, further obtaining the hub center point and blade root center point, and then calculating the direction vector of the blade in the image; calculating the direction vector of the blade root in space to obtain the wind turbine attitude; calculating the wind turbine orientation; calculating the spatial coordinates of the path center point and three blade tip points based on the wind turbine attitude; initially generating a path point sequence along the blade length based on the calculated path center point coordinates and blade tip point coordinates; and a drone flying along the generated path point sequence along the blade length, dynamically correcting the flight path by real-time blade tracking to ensure the blade remains within the camera's field of view.

Description

Technical Field

[0001] This invention belongs to the field of wind turbine blade inspection technology based on UAVs, and particularly relates to a path planning method and system for inspecting the front and back of the blades when the wind turbine is shut down. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] Wind turbine blades are susceptible to environmental erosion and mechanical fatigue during long-term operation, leading to surface damage (such as cracks and corrosion). Traditional inspection methods rely on manual climbing or fixed cameras, which are inefficient, risky, and lack comprehensive coverage.

[0004] In recent years, drone inspection technology has been gradually applied. For example, in existing technologies, drones fly to the top of wind turbines and determine the orientation of the wind turbines by taking pictures of the top view, or drones fly around wind turbines and determine the orientation of the wind turbines by the blade length from different angles.

[0005] All of the above methods require the drone to perform additional actions, thus reducing its range. Furthermore, all these methods require detecting the leaf tip position, which is difficult to detect in images, compromising detection accuracy. In summary, existing solutions have the following specific problems:

[0006] Drones typically need a long time to determine the attitude of wind turbines, which reduces the drone's range and makes the attitude recognition process cumbersome.

[0007] The drones rely on preset paths, causing them to fly along fixed paths and making them unable to adapt to slight rotations or bends when the blades are stopped.

[0008] Existing methods mostly target one side of the blade (front or back), which limits the coverage and makes it difficult to achieve comprehensive detection.

[0009] The displacement of the blades due to wind or structural deformation causes the inspection path to deviate, which cannot be dynamically corrected and still requires manual intervention. Summary of the Invention

[0010] To overcome the shortcomings of the prior art, this invention provides a path planning method for inspecting the front and back of wind turbine blades while the turbine is shut down. This method enables unmanned aerial vehicle (UAV) inspections that can autonomously plan paths, dynamically correct deviations, and cover both sides of the blades.

[0011] To achieve the above objectives, one or more embodiments of the present invention provide the following technical solutions:

[0012] Firstly, a path planning method for inspecting the front and back sides of wind turbine blades while the turbine is shut down is disclosed, including:

[0013] The camera gimbal acquires images of the wind turbine hub and blade roots within its field of view based on its pitch and rotation angles.

[0014] The blade root contour and hub position are identified by image segmentation, and the hub center point and blade root center point are obtained. Then the direction vector of the blade in the image is calculated.

[0015] The direction vector of the blade root in space is calculated to obtain the wind turbine attitude; the orientation of the wind turbine is calculated.

[0016] Based on the spatial coordinates of the center point of the wind turbine attitude calculation path and the tips of the three blades;

[0017] Based on the calculated coordinates of the path center point and the spatial coordinates of the blade tip, a preliminary sequence of path points along the blade length is generated.

[0018] The drone flies along a generated sequence of path points along the blade's length and dynamically corrects its flight path by tracking the blade in real time, ensuring that the blade remains within the camera's field of view.

[0019] As a further technical solution, the pitch and rotation angles of the camera gimbal are obtained as follows:

[0020] The camera lens of the camera gimbal on the drone is facing the direction of the wind turbine nacelle;

[0021] Based on the drone's flight altitude, the wind turbine's hub height, the drone's GPS coordinates, and the wind turbine's GPS coordinates, calculate the camera gimbal's pitch angle α, and adjust the camera gimbal's pitch angle to α.

[0022] The rotation angle β of the camera gimbal is calculated based on the GPS coordinates of the UAV and the wind turbine, and the rotation angle of the camera gimbal is adjusted to β.

[0023] As a further technical solution, the pitch angle calculation process of the camera gimbal is as follows:

[0024]

[0025] Among them, the drone's flight altitude h 无人机 Wind turbine hub height h 风电 UAV GPS coordinates (φ) 无人机 ,λ 无人机 (and GPS coordinates of the wind turbine) φ represents latitude, λ represents longitude, and d represents the straight-line distance between the drone and the wind turbine, which can be directly calculated from the GPS coordinates of the drone and the wind turbine.

[0026] As a further technical solution, the calculation process for the rotation angle β of the camera gimbal is as follows:

[0027]

[0028] Among them, the UAV GPS coordinates (φ) 无人机 ,λ 无人机 ) and GPS coordinates of wind turbine (φ) 风电 ,λ 风电 ), where φ is the latitude, λ is the longitude, and d is the straight-line distance between the drone and the wind turbine, which can be directly calculated from the GPS coordinates of the drone and the wind turbine.

[0029] As a further technical solution, the step of calculating the camera optical axis direction vector based on the lens direction is also included:

[0030] Establish a fixed coordinate system OXYZ, with the positive Z-axis pointing vertically upward, the positive X-axis pointing due east, and the positive Y-axis pointing due south.

[0031] The camera optical axis direction vector is obtained by rotating the true north direction vector as described above.

[0032] The calculated direction vector of the camera's optical axis is the normal vector of the lens plane.

[0033] As a further technical solution, when calculating the attitude of the wind turbine, let the direction vector of the wind turbine blade root in the fixed coordinate system be... The optical axis direction vector of the camera is calculated as follows: By obtaining the direction vector of the wind turbine blade root in a fixed coordinate system, and solving for the direction vector of the blade root in space, the pose of the wind turbine can be obtained.

[0034] Secondly, a path planning system for inspecting the front and back of the blades while the wind turbine is shut down was disclosed, including:

[0035] The image acquisition module is configured such that the camera gimbal acquires images of the wind turbine hub and blade root within its field of view based on its pitch and rotation angles.

[0036] The image segmentation module is configured to: identify the blade root contour and hub position through image segmentation, and further obtain the hub center point and blade root center point, and then calculate the direction vector of the blade in the image;

[0037] The wind turbine attitude acquisition module is configured to: calculate the direction vector of the blade root in space to obtain the wind turbine attitude; and calculate the orientation of the wind turbine.

[0038] The path point sequence generation module along the blade length is configured to: calculate the spatial coordinates of the path center point and the three blade tip points based on the wind turbine attitude.

[0039] Based on the calculated coordinates of the path center point and the spatial coordinates of the blade tip, a preliminary sequence of path points along the blade length is generated.

[0040] The inspection module is configured such that the UAV flies along a generated sequence of path points along the blade length and dynamically corrects the flight path by tracking the blade in real time, ensuring that the blade is always in the camera's field of view.

[0041] The above one or more technical solutions have the following beneficial effects:

[0042] The technical solution of this invention relies on simple conditions: it only requires obtaining a simple photo of the wind turbine and the perspective of the photo taken by the drone, and the photo must include a portion of the wind turbine hub and three blades.

[0043] The technical solution of this invention is a multimodal positioning technology that integrates GPS coordinates, machine vision, and mathematical calculations to accurately locate the wind turbine orientation and blade position.

[0044] The present invention provides a dynamic path correction solution: combining real-time tracking of blade position to adapt to dynamic changes in blade position.

[0045] The technical solution of this invention achieves seamless connection between the front and back sides: by calculating the back coordinates and adjusting the gimbal angle, it can achieve double-sided detection of the blade in a single flight, simplifying the process and improving accuracy.

[0046] The closed-loop inspection logic of this invention's technical solution automatically switches between blades and their front and back sides, reducing manual intervention and improving efficiency.

[0047] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0048] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0049] Figure 1 This is a flowchart illustrating the method for generating UAV inspection paths for wind turbines according to this application.

[0050] Figure 2 This is a schematic diagram of a wind turbine generator according to this application;

[0051] Figure 3 This is a schematic diagram of the location of the wind turbine taken by the drone in this application;

[0052] Figure 4 A schematic diagram of a photograph taken by the UAV during pose recognition in this application;

[0053] Figure 5 This is a schematic diagram of the center point of the wind turbine blade root and the center point of the hub in a photo taken by the UAV in this application.

[0054] Figure 6 This is a schematic diagram of the drone path planning in this application. Detailed Implementation

[0055] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0056] It should be noted that the terminology used herein is for the purpose of describing particular implementations only and is not intended to limit the exemplary implementations of the present invention.

[0057] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0058] Example 1

[0059] See Figure 1-6 As shown, to achieve fully automated intelligent inspection of wind turbines, this embodiment discloses a path planning method for inspecting the front and back sides of the blades when the wind turbine is shut down, including:

[0060] By using images captured by drones to calculate the orientation of wind turbines and the position of their blades, and further planning the inspection path for the drones, the operation of "autonomous flight and automatic planning" can be achieved.

[0061] The drone used in this example is the DJI Mavic 3T, which comes with a gimbal and camera, and is equipped with an RTK high-precision positioning device. This drone also has built-in obstacle avoidance capabilities.

[0062] The selected drone takes off and flies to a certain altitude. Based on the drone's coordinates and the wind turbine's coordinates, the gimbal angle is adjusted so that the wind turbine's hub and blade root are within the camera's field of view, allowing for clear imaging of the hub and blade root.

[0063] There are two methods for image processing: one is to directly rely on the drone's computing module for recognition, and the other is to transmit the captured images to a ground station for processing. Either method is acceptable, depending on the capabilities of the computing module used in the selected drone.

[0064] Image segmentation techniques, such as YOLOv8, are used to identify the wheel hub and blade roots in a photograph. Keypoint recognition technology is then used to calculate the center points of the wheel hub and blade roots, as well as the direction vector of the blade roots within the photograph.

[0065] The direction vector of the blade in the spatial coordinate system is calculated based on the direction vector of the camera's optical axis when the photo is taken by the drone, and the direction vector of the blade root in the photo.

[0066] The pose of the wind turbine is obtained by identifying the front and back of the wind turbine in the photo using a neural network and combining the direction vector of the blades.

[0067] The drone inspection path points are generated based on the calculated pose and wind turbine coordinates.

[0068] The drone flies along the path points and tracks the blades in real time, dynamically correcting its flight path to ensure that the blades remain in the camera's field of view.

[0069] The above method, when implemented in practice, includes:

[0070] Step 1: Determine the attitude of the wind turbine

[0071] Step S1-1: The drone flies to a certain altitude. Initialize the pitch and rotation angles of the camera gimbal. The initial values ​​are both 0°, meaning the camera lens is facing due north, and the camera lens focal length is the default value l0.

[0072] Step S1-2: Based on the drone's flight altitude h 无人机 Wind turbine hub height h 风电 UAV GPS coordinates (φ) 无人机 ,λ 无人机 ) and GPS coordinates of wind turbine (φ) 风电 ,λ 风电 (Note that φ is latitude and λ is longitude) Calculate the camera's pitch angle α. Adjust the camera's pitch angle to α.

[0073]

[0074] Where d is the straight-line distance between the drone and the wind turbine. For smaller wind turbines, the effect of the Earth's surface curvature can be ignored. x = Rcosφcosλ, y = Rcosφsinλ, z = Rsinφ, where R is the average radius of the Earth.

[0075] Steps S1-3: Calculate the camera gimbal rotation angle β based on the UAV's GPS coordinates and the wind turbine's GPS coordinates. Adjust the camera gimbal rotation angle to β:

[0076]

[0077] Step S1-4: After steps S1-1 to S1-3, the camera is now facing the wind turbine nacelle. The nacelle is identified using a neural network, and the camera's focal length is adjusted based on the proportion of the nacelle frame's length / height to the image, ensuring a clear view of the nacelle, hub, and blade roots. The adjusted focal length l1 = min(k × l0 × min(width)). 图像 / width 机舱 height 图像 / width 机舱 ),l max ), where k is the focal length adjustment factor, which is set according to the actual situation, l max This is the camera's maximum focal length. If the wind turbine blades are blocked by the tower, the drone needs to adjust its position and repeat the previous process until the nacelle, hub, and blade roots can be clearly captured.

[0078] Step S1-5: Calculate the camera optical axis direction vector based on the lens orientation:

[0079] Establish a fixed coordinate system OXYZ, with the positive Z-axis pointing vertically upward, the positive X-axis pointing due east, and the positive Y-axis pointing due south.

[0080] The camera's optical axis direction vector is the true north direction vector. This is obtained after rotation in steps S1-2 and S1-3. Therefore, the camera optical axis direction vector can be obtained as follows:

[0081]

[0082] The calculated direction vector of the camera's optical axis is the normal vector of the lens plane.

[0083] Steps S1-6: Identify the blade root contour and hub position using image segmentation techniques (e.g., YOLOv8), and further obtain the hub center point and blade root center point using keypoint recognition techniques. Calculate the blade direction vector in the image for subsequent steps. Assume the pixel coordinates of the center point of a certain blade root in the image are p. 叶根i =(x 叶根i ,y 叶根i The pixel coordinates of the center point of the wheel hub in the image are p. 轮毂 =(x 轮毂 ,y 轮毂 Its direction vector is Where i = 1, 2, 3.

[0084] Steps S1-7: Use a neural network (such as ResNet18 network) to determine the front and back of the wind turbine in the image, that is, determine which quadrant in the XOY plane the wind turbine is pointing to.

[0085] The input is a photo of the wind turbine taken by the camera at this moment;

[0086] The feature extraction layer can use residual network modules (such as convolutional layers of ResNet18) or a combination of traditional convolutional modules, with each convolutional layer followed by a batch normalization layer and a ReLU activation function;

[0087] The classifier consists of a global average pooling layer, a fully connected layer, and a Softmax output layer. The output dimensions correspond to four directional categories: the first quadrant (the wind turbine faces the upper left in the photo), the second quadrant (the wind turbine faces the upper right in the photo), the third quadrant (the wind turbine faces the lower right in the photo), and the fourth quadrant (the wind turbine faces the lower left in the photo).

[0088] In this implementation example, the neural network used is a trained ResNet18 network. During training, images of multiple wind turbines when the wind turbines are shut down are collected. The images must include the hub and a portion of the three blades. The collected images are preprocessed to remove duplicates and images that do not meet the requirements. Then, they are labeled. The labeling is based on the orientation of the wind turbine in the identified image. The orientation is labeled as follows: first quadrant (wind turbine facing the upper left in the image), second quadrant (wind turbine facing the upper right in the image), third quadrant (wind turbine facing the lower right in the image), and fourth quadrant (wind turbine facing the lower left in the image). Specifically, when the wind turbine is facing the first quadrant, the image is labeled as (1 0 0 0); when the wind turbine is facing the second quadrant, the image is labeled as (0 1 0 0); when the wind turbine is facing the third quadrant, the image is labeled as (0 01 0); and when the wind turbine is facing the fourth quadrant, the image is labeled as (0 0 0 1).

[0089] The annotated images are divided into training and testing sets. The training set is used to train the ResNet18 network to obtain the trained ResNet18 network. The testing set is then used to test the network until it meets the requirements, thus obtaining the final trained ResNet18 network.

[0090] Steps S1-8: Calculate the wind turbine attitude. Assume the direction vector of the wind turbine blade root in the fixed coordinate system is... The optical axis direction vector of the camera calculated in step S1-5 is: Therefore, the direction vector of the wind turbine blade root in a fixed coordinate system can be expressed by the following formula, where a, b, and c are unknowns to be determined.

[0091]

[0092] At the same time, the angles between these three directional vectors are 120° to each other, therefore we can obtain...

[0093]

[0094] Right now

[0095]

[0096] By solving these three equations, we can obtain the values ​​of the unknowns a, b, and c, and then obtain the direction vector of the blade root in space, thus obtaining the orientation of the wind turbine.

[0097] Steps S1-9: Calculate the direction vector of the wind turbine's orientation. The direction vector of the wind turbine's orientation in a fixed coordinate system. The calculation can be performed using the following formula. Since the three blade direction vectors are in the same plane, and this plane is perpendicular to the horizontal plane, any two direction vectors can be selected for calculation, and these direction vectors must lie within the horizontal plane.

[0098]

[0099] Step S1-10: Calculate the orientation of the wind turbine. Calculate the angle γ between the orientation vector and true north.

[0100]

[0101] Then, combine steps S1-7 to determine the orientation of the wind turbine. Taking a camera facing north as an example, if the wind turbine in the photo points to the first quadrant, the wind turbine's orientation is north-northeast (γ); if it points to the second quadrant, the wind turbine's orientation is north-northeast (360°-γ); if it points to the third quadrant, the wind turbine's orientation is north-northeast (180°+γ); if it points to the fourth quadrant, the wind turbine's orientation is north-northeast (180°-γ).

[0102] Step Two: Path Planning and Dynamic Adjustment

[0103] Step S2-1: Assume the wind turbine blade length is l. The spatial coordinates of the wind turbine pose path center point and the three blade tips obtained in steps S1-1-S1-10 can be calculated. The calculation method is as follows:

[0104] Coordinates of the center point of the wind turbine's front path: Coordinates of the center point of the path on the back of the wind turbine: Where P 风电 =(x 风电 ,y 风电 ,h 风电 In this case, k is a positive parameter that can be set according to the size of the wind turbine. It's necessary to ensure the drone is at a certain distance from the turbine while still being able to clearly capture images of blade surface damage. Generally, k can be twice the nacelle length. Determining the front and back sides requires combining steps S1-7.

[0105] Spatial coordinates of the tips of the three blades on the front of the wind turbine: Spatial coordinates of the tips of the three blades on the back of the wind turbine: Where i = 1, 2, 3.

[0106] Step S2-2: Based on the calculated coordinates of the path center point and the spatial coordinates of the blade tip, a preliminary sequence of path points along the blade length is generated. The number of path points can be adjusted.

[0107] Taking the inspection of a blade's front side by a drone as an example, the coordinates of the center point of the wind turbine's front path calculated according to step S2-1 are as follows: The spatial coordinates of the tip of the front blade are The coordinates of the generated path points are Where m is the number of path points generated, and n is the path point index.

[0108] Step S2-3: The UAV flies along the generated path points and dynamically corrects the flight path by tracking the blades in real time to ensure that the blades are always in the camera's field of view.

[0109] The drone detects the position of the blade outline in the image using machine vision. When the blade outline is on the left side of the image, the drone flies to the left. When the blade outline is on the right side of the image, the drone flies to the right. The drone dynamically corrects its flight path by combining generated path points with image detection to ensure that the blade outline is always in the center of the image.

[0110] S2-4: During the inspection process, the drone continuously monitors the images for blade tip or hub features. When these features are present, it indicates that the inspection of that surface has been completed. The drone then finds the nearest path point to inspect another blade or the other side of a blade. This process continues until the inspection is finished.

[0111] Step 3: Closed-loop control and safety mechanisms

[0112] Step S3-1: Mark the completed blades with a counter. If the inspection is interrupted (e.g., due to strong wind interference), the drone will automatically return to the safe point and record the progress.

[0113] This invention provides an automatic path planning method for unmanned aerial vehicle (UAV) inspection of the front and back surfaces of wind turbine blades while the turbine is shut down. This method uses machine vision to determine the turbine's orientation and blade position, combined with dynamic path correction technology to achieve full coverage inspection of both sides of the blades, solving the problems of rigid paths and incomplete coverage in existing technologies. This invention is applicable to the automated operation and maintenance of wind turbines and features high efficiency, safety, and strong adaptability.

[0114] Example 2

[0115] The purpose of this embodiment is to provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the above-described method.

[0116] Example 3

[0117] The purpose of this embodiment is to provide a computer-readable storage medium.

[0118] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the steps of the above method.

[0119] Example 4

[0120] The purpose of this embodiment is to provide a path planning system for inspecting the front and back sides of wind turbine blades while the turbine is shut down, including:

[0121] The image acquisition module is configured such that the camera gimbal acquires images of the wind turbine hub and blade root within its field of view based on its pitch and rotation angles.

[0122] The image segmentation module is configured to: identify the blade root contour and hub position through image segmentation, and further obtain the hub center point and blade root center point, and then calculate the direction vector of the blade in the image;

[0123] The wind turbine attitude acquisition module is configured to: calculate the direction vector of the blade root in space to obtain the wind turbine attitude; and calculate the orientation of the wind turbine.

[0124] The path point sequence generation module along the blade length is configured to: calculate the spatial coordinates of the path center point and the three blade tip points based on the wind turbine attitude.

[0125] Based on the calculated coordinates of the path center point and the spatial coordinates of the blade tip, a preliminary sequence of path points along the blade length is generated.

[0126] The inspection module is configured such that the UAV flies along a generated sequence of path points along the blade length and dynamically corrects the flight path by tracking the blade in real time, ensuring that the blade is always in the camera's field of view.

[0127] Example 5

[0128] The purpose of this embodiment is to provide a computer program product containing instructions that, when run on a computer, cause the computer to perform the methods and functions involved in any of the above embodiments.

[0129] The steps and methods involved in the apparatus of the above embodiments correspond to those in Embodiment 1. For specific implementation details, please refer to the relevant description section of Embodiment 1. The term "computer-readable storage medium" should be understood as a single medium or multiple media including one or more instruction sets; it should also be understood as including any medium capable of storing, encoding, or carrying an instruction set for execution by a processor and enabling the processor to perform any of the methods in this invention.

[0130] Those skilled in the art will understand that the modules or steps of the present invention described above can be implemented using general-purpose computer devices. Optionally, they can be implemented using computer-executable program code, thereby allowing them to be stored in a storage device for execution by a computer device, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. The present invention is not limited to any particular combination of hardware and software.

[0131] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A method for path planning of inspecting the front and back surfaces of a blade in a fan shutdown state, characterized in that, The method comprises the following steps: The camera holder obtains an image of the fan hub and the blade root part in the field of view based on the pitch angle and the rotation angle of the camera holder; the pitch angle calculation process of the camera holder is as follows: Wherein, the unmanned aerial vehicle flight height , the wind turbine hub height , the unmanned aerial vehicle GPS coordinate and the wind turbine GPS coordinate , is the latitude, is the longitude, d is the straight-line distance between the unmanned aerial vehicle and the wind turbine, which can be directly calculated through the GPS coordinates of the unmanned aerial vehicle and the wind turbine; The rotation angle of the camera holder is The calculation process is: Wherein, the GPS coordinates of the unmanned aerial vehicle The GPS coordinates of the wind turbine , is the latitude, is the longitude, d is the straight-line distance between the unmanned aerial vehicle and the wind turbine, which can be directly calculated by the GPS coordinates of the unmanned aerial vehicle and the wind turbine. The image segmentation module is configured to identify the blade root profile and the hub position through image segmentation, and further obtain the hub center point and the blade root center point, and then calculate the direction vector of the blade in the picture; The wind turbine posture acquisition module is configured to calculate the direction vector of the blade root in space to obtain the posture of the wind turbine; and calculate the orientation of the wind turbine; The path center point and the spatial coordinates of the three blade tip points are calculated based on the posture of the wind turbine; According to the calculated path center point coordinates and the spatial coordinates of the blade tip points, a path point sequence along the length of the blade is initially generated; The unmanned aerial vehicle flies along the generated path point sequence along the length of the blade, and dynamically corrects the flight path by tracking the blade in real time to ensure that the blade is always in the field of view of the camera.

2. The method of claim 1, wherein the method further comprises: determining a first path for the inspection of the first side of the blade; determining a second path for the inspection of the second side of the blade; and determining a third path for the inspection of the first side of the blade. The camera holder obtains an image of the fan hub and the blade root part in the field of view based on the pitch angle and the rotation angle of the camera holder; the pitch angle calculation process of the camera holder is as follows: The camera lens of the camera holder carried by the unmanned aerial vehicle is directed to the direction of the wind turbine cabin; According to the unmanned aerial vehicle flight height, the wind turbine hub height, the unmanned aerial vehicle GPS coordinate and the wind turbine GPS coordinate, the pitch angle of the camera holder is calculated , and the pitch angle of the camera holder is adjusted ; According to the GPS coordinates of the unmanned aerial vehicle and the GPS coordinates of the wind turbine, the rotation angle of the camera holder is calculated , and the rotation angle of the camera holder is adjusted .

3. The method of claim 1, wherein the method further comprises: determining a first path for the inspection of the first side of the blade; determining a second path for the inspection of the second side of the blade; and determining a third path for the inspection of the first side of the blade. The method further comprises the step of calculating the direction vector of the camera optical axis based on the direction of the lens: A fixed coordinate system OXYZ is established, the positive direction of the Z axis is vertically upward, the positive direction of the X axis is east, and the positive direction of the Y axis is south; The direction vector of the camera optical axis is obtained by rotating the positive north direction vector, and the direction vector of the camera optical axis is obtained based on this; The direction vector of the calculated camera optical axis is the normal vector of the lens plane.

4. The method of claim 1, wherein the method further comprises: determining whether the fan is in a stopped state; and if the fan is in the stopped state, determining whether the fan is in a cleaning mode. When the posture of the wind turbine is calculated, the direction vector of the wind turbine blade root in the fixed coordinate system is ; based on the calculated optical axis direction vector of the camera , the direction vector of the wind turbine blade root in the fixed coordinate system is obtained, the direction vector of the blade root in space is solved, the approximate direction of the wind turbine is judged in combination with the neural network, and then the accurate posture of the wind turbine is obtained.

5. A path planning system for inspecting the front and back surfaces of the blades in a fan shutdown state, characterized by, The method comprises the following steps: The image acquisition module is configured to obtain an image of the fan hub and the blade root part in the field of view based on the pitch angle and the rotation angle of the camera holder; the pitch angle calculation process of the camera holder is as follows: Wherein, the unmanned aerial vehicle flight height , the wind turbine hub height , the unmanned aerial vehicle GPS coordinate and the wind turbine GPS coordinate , is the latitude, is the longitude, d is the straight-line distance between the unmanned aerial vehicle and the wind turbine, which can be directly calculated through the GPS coordinates of the unmanned aerial vehicle and the wind turbine; The rotation angle of the camera holder is The calculation process is: Wherein, the GPS coordinates of the unmanned aerial vehicle The GPS coordinates of the wind turbine , is the latitude, is the longitude, d is the straight-line distance between the unmanned aerial vehicle and the wind turbine, which can be directly calculated by the GPS coordinates of the unmanned aerial vehicle and the wind turbine. The image segmentation module is configured to identify the blade root profile and the hub position through image segmentation, and further obtain the hub center point and the blade root center point, and then calculate the direction vector of the blade in the picture; The wind turbine posture acquisition module is configured to calculate the direction vector of the blade root in space to obtain the posture of the wind turbine; and calculate the orientation of the wind turbine; The path center point and the spatial coordinates of the three blade tip points are calculated based on the posture of the wind turbine; According to the calculated path center point coordinates and the spatial coordinates of the blade tip points, a path point sequence along the length of the blade is initially generated; The inspection module is configured to fly the unmanned aerial vehicle along the generated path point sequence along the length of the blade, and dynamically correct the flight path by tracking the blade in real time to ensure that the blade is always in the field of view of the camera.

6. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to implement the method of any one of claims 1 to 4.

7. A computer device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the program to implement the steps of the method of any one of claims 1-4.

8. A computer-readable storage medium having stored thereon a computer program, characterized in that, The program is executed by the processor to perform the steps of the method of any one of claims 1-4.

Images