Wind driven generator inspection method, device and equipment based on unmanned aerial vehicle and medium

Abstract

The invention relates to the technical field of unmanned aerial vehicle equipment, and discloses a wind driven generator inspection method, device and equipment based on an unmanned aerial vehicle and a medium. Firstly, when the unmanned aerial vehicle flies to a cruise initial position and the electric quantity is in a preset range, a fan top image is acquired; then, based on the fan top image, the unmanned aerial vehicle is controlled to fly to the fan blade center point of the wind driven generator, and a fan swept surface image is collected; then, determining the actual error distance of the swept surface based on the fan swept surface image; and under the condition that the actual error distance of the swept surface meets the preset condition of the swept surface, determining the reference flight direction of the component of the wind driven generator. Then, based on the component safety distance, the component inspection image overlapping rate and the component inspection shooting times, a component single-step moving distance sequence set and the component total moving distance are determined; and finally, performing image acquisition based on the reference flight direction of the component, the single-step moving distance sequence set of the component and the total moving distance of the component to obtain a component inspection image set.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of unmanned aerial vehicle equipment, and particularly relates to a wind turbine inspection method and device based on an unmanned aerial vehicle, equipment and a medium. BACKGROUND

[0002] With the development of unmanned aerial vehicle technology, it has become a common practice to use unmanned aerial vehicles to inspect wind power equipment.

[0003] In the related art, manual operation of unmanned aerial vehicle patrol flight partially solves the high-altitude safety problem, but the operation complexity is high, professional pilots are needed for control, the flight path depends on manual planning, and it is difficult to dynamically adjust under complex weather conditions, which can easily lead to image blur or missed detection. Therefore, a new wind turbine inspection method based on an unmanned aerial vehicle is needed. SUMMARY

[0004] The embodiments of the present specification aim to at least partially solve one of the technical problems in the related art. To this end, the embodiments of the present specification propose a wind turbine inspection method, device, equipment and medium based on an unmanned aerial vehicle.

[0005] The embodiments of the present specification provide a wind turbine inspection method based on an unmanned aerial vehicle, the method comprising: acquiring a wind turbine top image when the unmanned aerial vehicle flies to a cruise initial position and the power is within a preset range, wherein the cruise initial position comprises the coordinates and a safe height of the wind turbine; controlling the unmanned aerial vehicle to fly to a wind turbine blade center point of the wind turbine based on the wind turbine top image and collecting a wind turbine swept surface image; determining a swept surface actual error distance based on the wind turbine swept surface image; determining a component reference flight direction of the wind turbine when the swept surface actual error distance meets a swept surface preset condition; determining a component single step movement distance sequence set and a component total movement distance based on a component safety distance, a component inspection image overlap rate and a component inspection shooting frequency; performing image acquisition based on the component reference flight direction, the component single step movement distance sequence set and the component total movement distance to obtain a component inspection image set.

[0006] The embodiments of the present specification provide a wind turbine inspection device based on an unmanned aerial vehicle, the device comprising: a wind turbine top image acquisition module configured to acquire a wind turbine top image when the unmanned aerial vehicle flies to a cruise initial position and the power is within a preset range, wherein the cruise initial position comprises the coordinates and a safe height of the wind turbine; A wind turbine swept surface image acquisition module is used to control the UAV to fly to the center point of the wind turbine blade of the wind turbine and acquire a wind turbine swept surface image based on the wind turbine top image; A swept surface actual error distance determination module, configured to determine a swept surface actual error distance based on the wind turbine swept surface image; a component reference flight direction determination module, configured to determine a component reference flight direction of the wind turbine when the actual error distance of the swept surface meets a preset condition of the swept surface; A component movement data determination module is used to determine a component single-step movement distance sequence set and a component total movement distance based on a component safety distance, a component inspection image overlap rate, and a component inspection shooting number; The component inspection image acquisition module is used to acquire images based on the component reference flight direction, the component single-step movement distance sequence set and the component total movement distance to obtain a component inspection image set.

[0007] An embodiment of this specification provides a computer device, a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the method described in any one of the above embodiments are implemented.

[0008] The embodiments of this specification provide a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the method described in any one of the above embodiments are implemented.

[0009] An embodiment of this specification provides a computer program product, which includes instructions. When the instructions are executed by a processor of a computer device, the computer device can perform the steps of the method described in any of the above embodiments.

[0010] In the above embodiment of the present specification, first, the unmanned aerial vehicle is caused to fly to a cruising initial position and an image of the top of the wind turbine is acquired when the cruising initial position and the power are within a preset range, wherein the cruising initial position includes the coordinates of the wind turbine and a safe height. Subsequently, the unmanned aerial vehicle is caused to fly to the center point of the wind turbine blade of the wind turbine based on the image of the top of the wind turbine and an image of the swept surface of the wind turbine is acquired. Next, the actual error distance of the swept surface is determined based on the image of the swept surface of the wind turbine. When the actual error distance of the swept surface meets a preset condition of the swept surface, the reference flight direction of the component of the wind turbine is determined. Then, the component single-step movement distance sequence set and the component total movement distance are determined based on the component safe distance, the component inspection image overlap rate, and the component inspection shooting frequency. Finally, the image acquisition is performed based on the reference flight direction of the component, the component single-step movement distance sequence set, and the component total movement distance, and a set of component inspection images is obtained. The above implementation process fuses the environmental data in the wind turbine image and the multi-modal sensor information in real time, and combines a visual image recognition algorithm to dynamically optimize the flight trajectory and the shooting parameters of the unmanned aerial vehicle. While reducing the manual intervention, the detection accuracy and efficiency in a complex scene are improved, thereby effectively solving the contradiction between cost and efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0011] Figure 1 A flowchart of a wind turbine inspection method based on an unmanned aerial vehicle is provided for the embodiment of the present specification; Figure 2 A flowchart of causing the unmanned aerial vehicle to fly to the center point of the wind turbine blade and acquire the image of the swept surface of the wind turbine is provided for the embodiment of the present specification; Figure 3 A flowchart of determining the actual error distance of the swept surface is provided for the embodiment of the present specification; Figure 4a A flowchart of determining the foot coordinates of the blade is provided for the embodiment of the present specification; Figure 4b A plurality of key points corresponding to the image of the top of the wind turbine are provided for the embodiment of the present specification; Figure 5 A flowchart of determining the actual error distance of the top is provided for the embodiment of the present specification; Figure 6a A flowchart of correcting the actual error distance of the top is provided for the embodiment of the present specification; Figure 6b A top diagram after the actual error distance of the top is corrected is provided for the embodiment of the present specification; Figure 7 A flowchart of determining the axial flight direction of the wind wheel is provided for the embodiment of the present specification; Figure 8 A flowchart of acquiring the image of the windward surface of the wind turbine is provided for the embodiment of the present specification; Figure 9 A flowchart for determining the actual error distance of the windward surface according to an embodiment of the present specification is shown in FIG. 6; Figure 10a A flowchart for correcting the actual error distance of the windward surface according to an embodiment of the present specification is shown in FIG. 7; Figure 10b A schematic diagram of the windward surface after the actual error distance of the windward surface is corrected according to an embodiment of the present specification is shown in FIG. 8; Figure 11 A flowchart for determining the reference flight direction of the component of the wind turbine according to an embodiment of the present specification is shown in FIG. 9; Figure 12 A flowchart for determining the sequence set of the single-step moving distance of the blade and the total moving distance of the component according to an embodiment of the present specification is shown in FIG. 10; Figure 13 A flowchart for determining the sequence set of the single-step moving distance of the tower and the total moving distance of the tower according to an embodiment of the present specification is shown in FIG. 11; Figure 14a A flowchart for implementing the blade inspection according to an embodiment of the present specification is shown in FIG. 12; Figure 14b A schematic diagram of the wind turbine blade and tower inspection according to an embodiment of the present specification is shown in FIG. 13; Figure 15 A flowchart for implementing the tower inspection according to an embodiment of the present specification is shown in FIG. 14; Figure 16 A flowchart for collecting the image of the leeward surface of the wind turbine according to an embodiment of the present specification is shown in FIG. 15; Figure 17 A flowchart for determining the actual error distance of the leeward surface according to an embodiment of the present specification is shown in FIG. 16; Figure 18 A flowchart for correcting the actual error distance of the leeward surface according to an embodiment of the present specification is shown in FIG. 17; Figure 19 A schematic diagram of the wind turbine inspection device based on the unmanned aerial vehicle according to an embodiment of the present specification is shown in FIG. 18; Figure 20 A schematic diagram of the internal structure of the computer device according to an embodiment of the present specification is shown in FIG. 19. DETAILED DESCRIPTION

[0012] Embodiments of the present application are described in detail below with reference to the attached drawings, which are shown by way of example, and wherein like or similar elements are referred to by like reference numerals throughout the several views. The embodiments described below are examples intended to provide an explanation of the present application and are not intended in any way to restrict the present application.

[0013] In the related art, the artificial basket inspection method relies on the operation and maintenance personnel to climb the tower of the wind turbine (height exceeding 100 meters), and checks the blades, nacelle and other components through visual inspection or handheld devices (such as thermal imagers) at close range. However, in the artificial basket inspection method, the inspection time of each wind turbine is as long as 2-3 hours, and the annual average high-altitude accident rate is 0.3‰, which has a risk of falling. Since artificial inspection can only rely on limited observation, the missed detection rate is as high as 20%, especially for small cracks (such as cracks less than or equal to 2mm).

[0014] In the related art, the ground telescope / high-power lens detection method captures the blade images of the wind turbine through high-resolution equipment to achieve blade inspection. However, the ground telescope / high-power lens detection method is limited by light and weather, and the single coverage rate is less than 30%. In addition, the missed detection rate for small cracks is as high as 42%, and the annual effective detection window of offshore wind farms is usually less than 100 days. The data of the ground telescope / high-power lens detection method is highly subjective and lacks traceability.

[0015] In the related art, the fixed sensor online monitoring method deploys vibration / temperature sensors at key parts of the wind turbine, such as the gearbox and bearing, and transmits data in real time through the SCADA system. However, the fixed sensor online monitoring method only covers the internal components of the wind turbine and cannot detect external damage to the blades of the wind turbine (such as lightning marks and corrosion). In addition, it is incompatible with manual records and unmanned aerial vehicle data protocols, resulting in a fault prediction accuracy of less than 60%.

[0016] The above existing technologies have defects such as efficiency bottlenecks (such as dependence on downtime and low coverage), safety risks (such as high-altitude operations), fragmented data (such as difficulty in integrating multi-source information), and blind spots in blade diagnosis (such as difficulty in detecting external damage), resulting in an annual average damage rate of 9.5% for the blades of the wind turbine and high operating costs.

[0017] In the related art, manual operation of unmanned aerial vehicles for patrol flights partially solves the problem of high-altitude safety, but still has significant bottlenecks. On the one hand, the operation complexity is high: professional pilots are needed for control, the flight path relies on manual planning, and it is difficult to dynamically adjust in complex weather conditions (such as turbulence or sudden wind speed changes), which can easily cause image blurring or missed detection. On the other hand, the cost is high: a single professional unmanned aerial vehicle needs to be equipped with laser radar, thermal imager and other equipment, and the single flight endurance is less than or equal to 25 minutes, which frequently changes the battery and increases the overall operating cost.

[0018] Based on the above analysis, the embodiment of the present specification provides a UAV-based wind turbine inspection method. First, in the case that the UAV flies to the initial cruising position and the power is within the preset range, the top image of the wind turbine is obtained, wherein the initial cruising position includes the coordinates and the safe height of the wind turbine. Then, based on the top image of the wind turbine, the UAV is controlled to fly to the center point of the wind turbine blade of the wind turbine and collect the wind turbine sweeping surface image. Next, the actual error distance of the sweeping surface is determined based on the wind turbine sweeping surface image. In the case that the actual error distance of the sweeping surface meets the preset condition of the sweeping surface, the reference flight direction of the component of the wind turbine is determined. Then, based on the component safety distance, the component inspection image overlap rate and the component inspection shooting times, the component single-step moving distance sequence set and the component total moving distance are determined. Finally, based on the component reference flight direction, the component single-step moving distance sequence set and the component total moving distance, the image collection is performed to obtain the component inspection image set. The above implementation process fuses the environmental data in the wind turbine image and the multi-modal sensor information in real time, and combines the visual image recognition algorithm to dynamically optimize the flight trajectory and shooting parameters of the UAV. While reducing manual intervention, the detection accuracy and efficiency in complex scenes are improved, thereby effectively solving the contradiction between cost and efficiency.

[0019] The embodiment of the present specification provides a UAV-based wind turbine inspection method, which can include the following steps: Figure 1 S110, in the case that the UAV flies to the initial cruising position and the power is within the preset range, the top image of the wind turbine is obtained.

[0020] The initial cruising position includes the coordinates and the safe height of the wind turbine.

[0021] Specifically, before the inspection starts, the tower height , the blade length and the coordinates of the wind turbine are obtained by pre-measurement or other methods. Since there needs to be a safe distance between the UAV and the wind turbine to ensure the safety of the UAV, the safe distance between the top UAV and the wind turbine is set. The safe height . After the UAV vertically takes off and rises to the safe height , the top point flies to the coordinates of the wind turbine, and then the UAV nose is adjusted to the preset direction (such as the north direction) to start the inspection.

[0022] ​After the drone reaches its initial cruising position, it needs to determine whether the battery level is within a preset range (e.g., 25% to 100%) to determine whether the drone has enough power to return home and meet inspection requirements. If the battery level is not within the preset range, the drone is insufficient for subsequent inspections, resulting in a failed mission and the drone must be controlled to return home. If the battery level is within the preset range, it is necessary to determine whether the adaptive inspection parameters have been loaded. If not, they are loaded. If so, the drone's top camera's gimbal angle is adjusted to 0°, and image acquisition is performed to obtain an image of the wind turbine's top.

[0023] S120: Based on the image of the top of the wind turbine, control the drone to fly to the center point of the wind turbine blade and collect the image of the wind turbine's swept surface.

[0024] S130 : Determine an actual error distance of the swept surface based on the wind turbine swept surface image.

[0025] S140 : When the actual error distance of the swept surface meets a preset condition of the swept surface, determining a reference flight direction of components of the wind turbine.

[0026] S150 : Determine a component single-step moving distance sequence set and a component total moving distance based on a component safety distance, a component inspection image overlap rate, and a component inspection shooting number.

[0027] S160 , performing image acquisition based on the component reference flight direction, the component single-step movement distance sequence set, and the component total movement distance to obtain a component inspection image set.

[0028] Specifically, a wind turbine has two wind turbine swept areas. If at least one of them has not been inspected, image recognition and computational analysis are performed on the wind turbine top image to determine whether the drone is within the preset center range of the wind turbine top. If the drone is within the preset center range of the wind turbine top, the control algorithm is used to control the drone to fly to the center point of the wind turbine blades. The drone's nose is then adjusted to facilitate image acquisition, obtaining an image of the wind turbine swept area.

[0029] Next, the fan swept surface image is subjected to image recognition and calculation to determine the actual error distance of the swept surface. In the case where the actual error distance of the swept surface meets the preset condition of the swept surface, it is indicated that the unmanned aerial vehicle is at the actual blade center point at this time. Then, the fan swept surface image is analyzed and calculated to determine the component reference flight direction of the wind turbine. It is also necessary to calculate the component single-step moving distance sequence set and the component total moving distance by using the component safety distance, the component inspection image overlap rate and the component inspection shooting frequency, so as to control the flight distance of the unmanned aerial vehicle. Subsequently, the unmanned aerial vehicle is controlled to fly along the component reference flight direction according to the component single-step moving distance sequence set by using the control algorithm, and image acquisition is performed after completing a single-step moving distance, so as to finally obtain the component inspection image set. Finally, the unmanned aerial vehicle is controlled to fly along the component reference flight direction in the opposite direction according to the component total moving distance by using the control algorithm, and returns to the actual blade center point, so as to perform subsequent inspection or return.

[0030] It should be noted that, since the wind turbine has two fan swept surfaces, after the component inspection image set of one side of the fan swept surface is determined through the above process, the unmanned aerial vehicle is controlled to fly back to the cruising initial position, and then it is re-judged whether the electric quantity is within the preset range. If the electric quantity is not within the preset range, the unmanned aerial vehicle is controlled to return. If the electric quantity is within the preset range, subsequent operations are performed to obtain the component inspection image set of the other side of the fan swept surface, and then the unmanned aerial vehicle returns.

[0031] In the above embodiment, first, in the case where the unmanned aerial vehicle flies to the cruising initial position and the electric quantity is within the preset range, the fan top image is acquired, wherein the cruising initial position includes the coordinates and the safety height of the wind turbine. Subsequently, based on the fan top image, the unmanned aerial vehicle is controlled to fly to the fan blade center point of the wind turbine and acquire the fan swept surface image. Next, the actual error distance of the swept surface is determined based on the fan swept surface image. In the case where the actual error distance of the swept surface meets the preset condition of the swept surface, the component reference flight direction of the wind turbine is determined. Then, based on the component safety distance, the component inspection image overlap rate and the component inspection shooting frequency, the component single-step moving distance sequence set and the component total moving distance are determined. Finally, the component inspection image set is obtained by image acquisition based on the component reference flight direction, the component single-step moving distance sequence set and the component total moving distance. The above implementation process fuses the environmental data in the fan image and the multi-modal sensor information in real time, and combines the visual image recognition algorithm to dynamically optimize the flight trajectory and shooting parameters of the unmanned aerial vehicle. While reducing manual intervention, the detection accuracy and efficiency in complex scenes are improved, thereby effectively solving the contradiction between cost and efficiency.

[0032] In some embodiments, please refer to Figure 2, the top image of the fan corresponds to a top image center point, based on the top image of the fan, the unmanned aerial vehicle is controlled to fly to the center point of the fan blade of the wind turbine and collect the fan swept surface image, which can include the following steps: S210, based on the top image of the fan, the first target area contour is determined.

[0033] S220, based on the first target area contour, the blade foot coordinate is determined.

[0034] S230, based on the first target area contour, the blade foot coordinate and the top image center point, the top actual error distance is determined.

[0035] S240, in the case that the top actual error distance meets the top preset condition, the wind wheel axial flight direction is determined based on the blade foot coordinate.

[0036] S250, based on the wind wheel axial flight direction, the horizontal safety distance and the blade safety height, the unmanned aerial vehicle is controlled to fly to the center point of the fan blade and collect the fan swept surface image.

[0037] Specifically, the image recognition algorithm is used to identify the top image of the fan to determine the first target area contour. Next, based on the obtained first target area contour, the geometric calculation method is used to determine the blade foot coordinate. After determining the blade foot coordinate, the first target area contour and the top image center point are combined to convert the image distance into the actual distance to determine the top actual error distance. When the top actual error distance meets the top preset condition, that is, when the actual error is within the allowable range, it indicates that the position of the unmanned aerial vehicle is within the preset center range of the top of the wind turbine. Then, by analyzing the positional relationship of the blade foot coordinate, the wind wheel axial flight direction is determined by using the geometric principle and the heading planning algorithm. Finally, based on the determined wind wheel axial flight direction, combined with the horizontal safety distance and the blade safety height, the unmanned aerial vehicle is controlled to fly to the center point of the fan blade and collect the fan swept surface image.

[0038] In the above embodiment, based on the top image of the fan, the first target area contour is determined, based on the first target area contour, the blade foot coordinate is determined, based on the first target area contour, the blade foot coordinate and the top image center point, the top actual error distance is determined, in the case that the top actual error distance meets the top preset condition, the wind wheel axial flight direction is determined based on the blade foot coordinate, based on the wind wheel axial flight direction, the horizontal safety distance and the blade safety height, the unmanned aerial vehicle is controlled to fly to the center point of the fan blade and collect the fan swept surface image, which provides a data basis for subsequent blade and tower inspection.

[0039] In some embodiments, please refer to Figure 3, the fan swept surface image corresponds to a swept surface image center point, and determining the swept surface actual error distance based on the fan swept surface image can include the following steps: S310, identifying based on the fan swept surface image to determine a second target area contour.

[0040] S320, determining a wind wheel center structure centroid coordinate based on the second target area contour.

[0041] S330, determining the swept surface actual error distance based on the second target area contour, the wind wheel center structure centroid coordinate, and the swept surface image center point.

[0042] Specifically, the image recognition algorithm is used to identify the fan swept surface image to determine the second target area contour. Next, the second target area contour obtained is calculated by the centroid calculation formula to determine the wind wheel center structure centroid coordinate. After determining the wind wheel center structure centroid coordinate, the proportional conversion between the image distance and the actual distance is performed in combination with the second target area contour and the swept surface image center point to determine the swept surface actual error distance.

[0043] In the above embodiment, the second target area contour is determined based on the identification of the fan swept surface image, the wind wheel center structure centroid coordinate is determined based on the second target area contour, and the swept surface actual error distance is determined based on the second target area contour, the wind wheel center structure centroid coordinate, and the swept surface image center point, which provides a data basis for subsequent unmanned aerial vehicle swept surface correction.

[0044] In some embodiments, referring to Figure 4a The first target area contour includes a first blade segmentation contour, a second blade segmentation contour, and a nacelle segmentation contour. The blade foot coordinate is determined based on the calculation of the first target area contour, which can include the following steps: S410, determining a first blade centroid coordinate based on the calculation of the first blade segmentation contour.

[0045] S420, determining a second blade centroid coordinate based on the calculation of the second blade segmentation contour.

[0046] S430, determining a nacelle centroid coordinate based on the calculation of the nacelle segmentation contour.

[0047] S440, determining the blade foot coordinate based on the calculation of the first blade centroid coordinate, the second blade centroid coordinate, and the nacelle centroid coordinate.

[0048] Specifically, the image of the top of the wind turbine is recognized using an image recognition algorithm to determine the segmentation contours recognized in the image. When it is determined that the image of the top of the wind turbine includes the first blade segmentation contour, the second blade segmentation contour, and the cabin segmentation contour corresponding to the two largest blades, the first blade segmentation contour is calculated using the centroid calculation formula to determine the first blade centroid coordinates. The second blade segmentation contour is calculated using the centroid calculation formula to determine the second blade centroid coordinates. The cabin segmentation contour is calculated using the centroid calculation formula to determine the cabin centroid coordinates. Finally, the vertical projection method is used to calculate the distance between the cabin centroid coordinates and the line connecting the first blade centroid coordinates and the second blade centroid coordinates, thereby determining the blade vertical foot coordinates. If it is determined that the image of the top of the wind turbine does not include at least one of the first blade segmentation contour, the second blade segmentation contour, and the cabin segmentation contour corresponding to the two largest blades, the drone is controlled to return. For example, refer to Figure 4b , Figure 4b The green point in the figure is the centroid of the blade, the red point is the centroid of the nacelle, and the dark blue point at the centroid of the blade is the foot of the blade.

[0049] Exemplarily, based on the first blade segmentation profile Perform calculations to determine the coordinates of the first blade's center of mass ). Segment the contour based on the second leaf Perform calculations to determine the coordinates of the second blade's center of mass ). Based on the cabin segmentation contour Perform calculations to determine the coordinates of the center of mass of the cabin The center of mass coordinates of the cabin are calculated using the following formula: To the coordinates of the first blade's centroid and the coordinates of the second blade's centroid The vertical foot coordinate of the blade connecting the lines :

[0050]

[0051] in, As a parameter.

[0052] In the above embodiment, calculation is performed based on the first blade segmentation contour to determine the coordinates of the center of mass of the first blade, calculation is performed based on the second blade segmentation contour to determine the coordinates of the center of mass of the second blade, calculation is performed based on the cabin segmentation contour to determine the coordinates of the center of mass of the cabin, calculation is performed based on the coordinates of the center of mass of the first blade, the coordinates of the center of mass of the second blade and the coordinates of the center of mass of the cabin to determine the vertical foot coordinates of the blade, thereby providing a data basis for subsequent correction of the wind turbine top on the drone.

[0053] In some embodiments, see Figure 5determining the top actual error distance based on the first target region contour, the blade foot coordinate and the top image center point can include the following steps: S510, calculating based on the blade foot coordinate and the top image center point to determine the top image error distance.

[0054] S520, determining the top proportion coefficient based on the area of the cabin segmentation contour and the actual area of the cabin top.

[0055] S530, determining the top actual error distance based on the top image error distance and the top proportion coefficient.

[0056] Specifically, the top image center point is determined by calculating the center point of the fan top image. Then, the distance between the blade foot coordinate and the top image center point is calculated based on the geometric projection relationship, and the distance value between the foot and the image center point is obtained and taken as the top image error distance. The area of the cabin segmentation contour is calculated by the contour area calculation formula, and the area of the cabin segmentation contour is determined. Next, the calculated area of the cabin segmentation contour is compared with the actual area of the cabin top, and the area proportion conversion of the two is performed to determine the top proportion coefficient. Finally, the top actual error distance is determined by calculating the top image error distance and the top proportion coefficient. The actual area of the cabin top is an adaptive inspection parameter. For example, please refer to Figure 4b , Figure 4b The blue point in the middle of the sky is the top image center point.

[0057] For example, the top image center point is = ), the blade foot coordinate is , the area of the cabin segmentation contour is , and the actual area of the cabin top is . The top image center point and the top actual error distance are calculated by the following formula:

[0058]

[0059]

[0060]

[0061]

[0062] wherein, is the width of the fan top image, is the height of the fan top image, is the top image error distance, is the top proportion coefficient, The top actual error distance is the actual error distance of the top.

[0063] In the above embodiment, the top image error distance is determined based on the blade foot coordinates and the top image center point, the top proportion coefficient is determined based on the area of the cabin segmentation contour and the actual area of the top of the cabin, and the top actual error distance is determined based on the top image error distance and the top proportion coefficient, thereby providing a data basis for subsequent correction of the top of the wind turbine by the unmanned aerial vehicle.

[0064] In some embodiments, referring to Figure 6a , the method can further include the following steps: S610, in the case that the top actual error distance does not meet the top preset condition, the correction flight direction is determined based on the blade foot coordinates and the top image center point.

[0065] S620, after controlling the unmanned aerial vehicle to fly the top actual error distance based on the correction flight direction, the corrected top image of the wind turbine is collected.

[0066] The corrected top image of the wind turbine corresponds to a corrected image center point.

[0067] Specifically, in the case that the top actual error distance does not meet the top preset condition, it indicates that the position of the unmanned aerial vehicle is not within the preset center range of the top of the wind turbine, and there is a deviation phenomenon, so the position of the unmanned aerial vehicle needs to be corrected in order to improve the accuracy and reliability of the overall inspection process, and avoid the problem of inaccurate inspection data or missing key parts due to position deviation. Taking the preset direction of the unmanned aerial vehicle as a reference, the angle between the current position of the unmanned aerial vehicle and the blade foot coordinates is calculated according to the geometric relationship between the blade foot coordinates and the top image center point, and the correction flight direction is determined. It should be noted that clockwise is positive. Then, the unmanned aerial vehicle is controlled to fly the top actual error distance along the correction flight direction by using a control algorithm, and then reaches the required position, and then image collection is performed to obtain the corrected top image of the wind turbine. The corrected image center point is determined by center point calculation using the corrected top image of the wind turbine. The top preset condition can be less than the center point correction threshold .

[0068] For example, the correction flight direction is determined based on the blade foot coordinates and the top image center point = .

[0069] S630, the corrected target area contour is determined based on the corrected top image of the wind turbine.

[0070] ​S640, calculate based on the corrected target region contour to determine the corrected blade foot coordinate.

[0071] The corrected target region contour corresponds to the corrected engine room centroid coordinate.

[0072] Specifically, the image recognition algorithm is used to identify the corrected fan top image, and the segmentation contour recognized in the image can be determined. In the case where the corrected fan top image includes the segmentation contour corresponding to the two largest blades and the corrected engine room segmentation contour, the centroid calculation formula is used to calculate the two blade segmentation contours respectively to determine the centroid coordinates of the two blades respectively. The centroid calculation formula is used to calculate the corrected engine room segmentation contour to determine the corrected engine room centroid coordinate. Finally, the vertical projection method is used to calculate the distance between the corrected engine room centroid coordinate and the line connecting the two blade centroid coordinates, and then the corrected blade foot coordinate is determined. If it is determined that the corrected fan top image does not include at least one of the segmentation contour corresponding to the two largest blades and the corrected engine room segmentation contour, the unmanned aerial vehicle is controlled to return.

[0073] S650, based on the corrected target region contour, the corrected blade foot coordinate and the corrected image center point, determine the corrected top actual error distance.

[0074] Specifically, the distance between the corrected blade foot coordinate and the corrected image center point is calculated based on the geometric projection relationship, the distance value between the foot and the image center point is obtained, and the distance is taken as the corrected top image error distance. The corrected engine room segmentation contour is calculated by the contour area calculation formula to determine the area of the corrected engine room segmentation contour. Next, the area of the corrected engine room segmentation contour calculated is compared with the actual area of the engine room top, and the area proportion conversion of the two is performed to determine the corrected top proportion coefficient. Finally, the corrected top image error distance and the corrected top proportion coefficient are calculated to determine the corrected top actual error distance.

[0075] S660, based on the corrected top actual error distance and the top preset condition, in the case where the corrected top actual error distance does not meet the top preset condition and the iteration number does not meet the iteration preset number, the above process is iteratively executed until the corrected top actual error distance meets the top preset condition and the corrected engine room centroid coordinate is taken as the engine room centroid coordinate and the corrected blade foot coordinate is taken as the blade foot coordinate.

[0076] Specifically, the corrected top actual error distance is compared with the top preset condition. When the corrected top actual error distance does not meet the top preset condition, the number of times of position correction performed by the current UAV is determined, and the number of iterations is determined. If the number of iterations does not satisfy the preset number of iterations, the above process can be re-executed until the corrected top actual error distance meets the top preset condition, at which time the corrected nacelle centroid coordinates are taken as the nacelle centroid coordinates, and the corrected blade foot coordinates are taken as the blade foot coordinates. If the number of iterations satisfies the preset number of iterations, the UAV is controlled to return. The preset number of iterations can be the maximum number of corrections in a single inspection stage .

[0077] Exemplarily, referring to Figure 6b , Figure 6b the image in which the corrected top actual error distance meets the top preset condition.

[0078] In the above embodiments, the position of the UAV is corrected to achieve more accurate UAV flight movement and image acquisition during inspection.

[0079] In some embodiments, referring to Figure 7 , the wind wheel axial flight direction includes the windward flight direction and the leeward flight direction. When the top actual error distance meets the top preset condition, the wind wheel axial flight direction is determined based on the blade foot coordinates, which can include the following steps: S710, based on the nacelle centroid coordinates and the blade foot coordinates, the windward flight direction is determined.

[0080] S720, based on the blade foot coordinates and the nacelle centroid coordinates, the leeward flight direction is determined.

[0081] Specifically, when the top actual error distance meets the top preset condition, the angle between the current position of the UAV and the nacelle is calculated based on the geometric relationship between the nacelle centroid coordinates and the blade foot coordinates, with the UAV preset direction as the reference, to determine the windward flight direction. It should be noted that clockwise is positive.

[0082] With the UAV preset direction as the reference, the angle between the nacelle and the current position of the UAV is calculated based on the geometric relationship between the blade foot coordinates and the nacelle centroid coordinates, to determine the leeward flight direction. It should be noted that clockwise is positive.

[0083] It should be noted that the inspection sequence can be pre-set. According to the actual situation, the inspection of the windward side can be performed first, followed by the inspection of the leeward side, or vice versa.

[0084] In the above embodiment, the direction in which the unmanned aerial vehicle needs to fly is determined, so as to realize the control of the unmanned aerial vehicle, so as to subsequently complete the inspection of the wind turbine.

[0085] In some embodiments, referring to Figure 8 , the fan sweeping surface image includes a fan windward surface image, based on the wind wheel axial flight direction, the horizontal safety distance and the blade safety height, the unmanned aerial vehicle is controlled to fly to the fan blade center point and collect the fan sweeping surface image, which can include the following steps: S810, based on the windward surface flight direction, the unmanned aerial vehicle is controlled to fly horizontally for a horizontal safety distance, and then fly downward for a blade safety height, so that the unmanned aerial vehicle flies to the fan blade center point.

[0086] S820, based on the windward surface flight direction, the unmanned aerial vehicle is controlled to fly horizontally for a horizontal safety distance, and then fly downward for a blade safety height, so that the unmanned aerial vehicle flies to the fan blade center point.

[0087] The fan windward surface image corresponds to a windward surface image center point.

[0088] Specifically, the blade length The distance between the safety distance of the wind turbine The sum of the distances is the blade safety height. In order to ensure the safety of the unmanned aerial vehicle during the inspection process, the unmanned aerial vehicle is controlled to fly horizontally along the windward surface flight direction for a horizontal safety distance, so that the unmanned aerial vehicle can effectively avoid the fan blades and can be kept within a suitable range for detection during the flight process. The unmanned aerial vehicle can effectively avoid the blades of the wind turbine and can be kept within a suitable range for detection during the flight process. After completing the horizontal flight to reach the horizontal safety distance, the unmanned aerial vehicle is controlled to fly downward for a blade safety height, so that the unmanned aerial vehicle flies to the fan blade center point. After reaching the fan blade center point, the unmanned aerial vehicle needs to adjust the heading of the unmanned aerial vehicle based on the windward surface flight direction, so as to ensure that the heading is aligned with the windward surface of the wind turbine, and then image acquisition is performed to obtain the fan windward surface image. The fan windward surface image is used for center point calculation to determine the windward surface image center point. The horizontal safety distance is an adaptive inspection parameter.

[0089] In the above embodiment, based on the windward surface flight direction, the unmanned aerial vehicle is controlled to fly horizontally for a horizontal safety distance, and then fly downward for a blade safety height, so that the unmanned aerial vehicle flies to the fan blade center point, based on the windward surface flight direction, the heading of the unmanned aerial vehicle is adjusted and the fan windward surface image is collected, which provides a data basis for subsequent blade and tower inspection.

[0090] In some embodiments, referring to Figure 9 , the second target area outline includes a hub segmentation outline, based on the second target area outline, the wind wheel center structure centroid coordinates are determined, which can include the following steps: S910, calculate based on the hub segmentation contour to determine the hub centroid coordinates.

[0091] Specifically, in the process of inspecting the windward surface, the image recognition algorithm is used to identify the windward surface image of the wind turbine, and the segmentation contour identified in the image can be determined. In the case where the windward surface image of the wind turbine includes the segmentation contour of three blades, the segmentation contour of the tower drum and the hub segmentation contour, the hub segmentation contour is calculated by the centroid calculation formula to determine the hub centroid coordinates. If it is determined that the windward surface image of the wind turbine does not include at least one of the segmentation contour of three blades, the segmentation contour of the tower drum and the hub segmentation contour, the unmanned aerial vehicle is controlled to return.

[0092] Based on the second target area contour, the hub centroid coordinates and the swept surface image center point, the swept surface actual error distance can be determined, which can include the following steps: S920, calculate based on the hub centroid coordinates and the windward surface image center point to determine the windward surface image error distance.

[0093] S930, determine the windward surface proportionality coefficient based on the area of the hub segmentation contour and the actual area of the hub.

[0094] S940, determine the windward surface actual error distance based on the windward surface image error distance and the windward surface proportionality coefficient.

[0095] Specifically, based on the geometric projection relationship, the distance between the hub centroid coordinates and the windward surface image center point is calculated, the distance value between the hub and the image center point is obtained, and the distance is taken as the windward surface image error distance. The hub segmentation contour is calculated by the contour area calculation formula to determine the area of the hub segmentation contour. Next, the calculated area of the hub segmentation contour is compared with the actual area of the hub, and the area proportion of the two is calculated to determine the windward surface proportionality coefficient. Finally, the windward surface actual error distance is determined according to the calculation of the windward surface image error distance and the windward surface proportionality coefficient. The actual area of the hub is an adaptive inspection parameter.

[0096] Exemplarily, the windward surface image center point is = , the hub centroid coordinates are ), the area of the hub segmentation contour is , and the actual area of the hub is . The windward surface image center point and the windward surface actual error distance are calculated by the following formula:

[0097]

[0098]

[0099]

[0100]

[0101] in, is the image width of the windward side of the wind turbine, is the image height of the windward side of the wind turbine, is the windward image error distance, is the windward surface proportional coefficient, is the actual error distance on the windward side.

[0102] In the above embodiment, calculations are performed based on the hub segmentation contour to determine the hub center of mass coordinates, calculations are performed based on the hub center of mass coordinates and the center point of the windward surface image to determine the windward surface image error distance, the windward surface proportional coefficient is determined based on the area of ​​the hub segmentation contour and the actual area of ​​the hub, and the actual windward surface error distance is determined based on the windward surface image error distance and the windward surface proportional coefficient, thereby providing a data basis for subsequent windward surface correction of the drone.

[0103] In some embodiments, see Figure 10a , the method may further comprise the following steps: S1010: When the actual error distance of the windward surface does not meet the preset conditions of the swept surface, a calculation is performed based on the hub center of mass coordinates and the center point of the windward surface image to determine the windward correction flight direction.

[0104] S1020, after correcting the flight direction based on the windward direction to control the actual error distance of the windward side of the UAV, collect the corrected windward side image.

[0105] The corrected windward surface image corresponds to the center point of the windward corrected image.

[0106] Specifically, in the case that the actual error distance of the windward surface does not meet the preset condition of the swept surface, it indicates that the position of the unmanned aerial vehicle is not in the preset central range of the windward surface of the wind turbine, and there is a deviation phenomenon, so the position of the unmanned aerial vehicle needs to be corrected in order to improve the accuracy and reliability of the overall inspection process, and avoid the problem of inaccurate inspection data or missing key parts due to position deviation. Taking the preset direction of the unmanned aerial vehicle as the reference, the angle between the current position of the unmanned aerial vehicle and the hub centroid coordinates is calculated according to the geometric relationship between the hub centroid coordinates and the center point of the windward surface image, and the windward correction flight direction is determined. It should be noted that clockwise is positive. Then, the control algorithm is used to control the unmanned aerial vehicle to fly along the windward correction flight direction to the actual error distance of the windward surface, and then to the required position, and then image acquisition is performed to obtain the corrected windward surface image. The center point of the corrected windward surface image is determined by using the corrected windward surface image. The swept surface preset condition can be less than the center point correction threshold .

[0107] Exemplarily, based on the hub centroid coordinates ) and the center point of the windward surface image = , the windward correction flight direction is determined:

[0108] S1030, based on the corrected windward surface image, the windward correction target area contour is determined.

[0109] S1040, based on the windward correction target area contour, the corrected hub centroid coordinates and the center point of the windward correction image, the actual error distance of the corrected windward surface is determined.

[0110] The windward correction target area contour corresponds to the corrected hub centroid coordinates.

[0111] Specifically, the image recognition algorithm is used to recognize the corrected windward surface image, and the windward correction target area contour recognized in the image can be determined. In the case that the windward correction target area contour includes the segmentation contour of three blades and the corrected hub segmentation contour, the corrected hub centroid coordinates are determined by calculating the corrected hub segmentation contour through the centroid calculation formula.

[0112] The distance between the corrected hub centroid coordinate and the center point of the upwind corrected image is calculated based on the geometric projection relationship, the distance value between the hub and the center point of the image is obtained, and the distance is taken as the upwind face correction image error distance. The corrected hub segmentation contour is calculated by the contour area calculation formula, and the area of the corrected hub segmentation contour is determined. Next, the calculated area of the corrected hub segmentation contour is compared with the actual area of the hub, and the area ratio conversion of the two is performed to determine the upwind face correction proportion coefficient. Finally, the upwind actual error distance after correction is determined by calculating the upwind face correction image error distance and the upwind face correction proportion coefficient.

[0113] S1050, based on the comparison between the corrected upwind actual error distance and the sweep face preset condition, if the corrected upwind actual error distance does not meet the sweep face preset condition and the iteration number meets the upwind face iteration preset number, the above process is iteratively executed until the corrected upwind actual error distance meets the sweep face preset condition and the corrected upwind face image is taken as the fan upwind face image.

[0114] Specifically, the corrected upwind actual error distance is compared with the sweep face preset condition. If the corrected upwind actual error distance does not meet the sweep face preset condition, the number of position corrections that the current UAV has performed is determined to determine the iteration number. If the iteration number does not meet the upwind face iteration preset number, the above process can be re-executed until the corrected upwind actual error distance meets the sweep face preset condition, at which time the corrected upwind face image is taken as the fan upwind face image. If the iteration number meets the upwind face iteration preset number, the UAV is controlled to return. The upwind face iteration preset number can be the maximum correction number in a single inspection stage .

[0115] Exemplarily, please refer to Figure 10b , Figure 10b to the image whose corrected upwind actual error distance meets the sweep face preset condition.

[0116] In the above embodiments, the position of the UAV is corrected to achieve more accurate UAV flight movement and acquisition of inspection images.

[0117] In some embodiments, please refer to Figure 11 , the second target area further includes a third blade segmentation contour, a fourth blade segmentation contour, a fifth blade segmentation contour, and a tower cylinder segmentation contour, and determining the component reference flight direction of the wind turbine in the case that the sweep face actual error distance meets the sweep face preset condition can include the following steps: S1110, based on the third blade segmentation contour, the third blade centroid coordinate is calculated.

[0118] S1120, calculate based on the fourth blade segmentation contour, determine the fourth blade centroid coordinates.

[0119] S1130, calculate based on the fifth blade segmentation contour, determine the fifth blade centroid coordinates.

[0120] S1140, calculate based on the tower segmentation contour, determine the tower centroid coordinates.

[0121] S1150, based on the third blade centroid coordinates and the wind wheel center structure centroid coordinates, determine the third blade flight direction.

[0122] S1160, based on the fourth blade centroid coordinates and the wind wheel center structure centroid coordinates, determine the fourth blade flight direction.

[0123] S1170, based on the fifth blade centroid coordinates and the wind wheel center structure centroid coordinates, determine the fifth blade flight direction.

[0124] S1180, based on the tower centroid coordinates and the wind wheel center structure centroid coordinates, determine the tower flight direction.

[0125] Specifically, the image recognition algorithm is used to identify the windward surface image of the wind turbine, and the segmentation contour identified in the image can be determined. In the case where the windward surface image of the wind turbine includes the third blade segmentation contour, the fourth blade segmentation contour, the fifth blade segmentation contour, the tower segmentation contour and the wind wheel center structure segmentation contour, the third blade segmentation contour is calculated by the centroid calculation formula to determine the third blade centroid coordinates. The fourth blade segmentation contour is calculated by the centroid calculation formula to determine the fourth blade centroid coordinates. The fifth blade segmentation contour is calculated by the centroid calculation formula to determine the fifth blade centroid coordinates. The tower segmentation contour is calculated by the centroid calculation formula to determine the tower centroid coordinates.

[0126] The unmanned aerial vehicle preset direction is taken as the reference, and the angle at which the unmanned aerial vehicle needs to fly towards the third blade is calculated according to the geometric relationship between the third blade centroid coordinates and the wind wheel center structure centroid coordinates, to determine the third blade flight direction. The angle at which the unmanned aerial vehicle needs to fly towards the fourth blade is calculated according to the geometric relationship between the fourth blade centroid coordinates and the wind wheel center structure centroid coordinates, to determine the fourth blade flight direction. The angle at which the unmanned aerial vehicle needs to fly towards the fifth blade is calculated according to the geometric relationship between the fifth blade centroid coordinates and the wind wheel center structure centroid coordinates, to determine the fifth blade flight direction. The angle at which the unmanned aerial vehicle needs to fly towards the tower is calculated according to the geometric relationship between the tower centroid coordinates and the wind wheel center structure centroid coordinates, to determine the tower flight direction. It should be noted that clockwise is positive.

[0127] It should be noted that in the process of inspecting the windward surface, the wind wheel center structure centroid coordinate is the hub centroid coordinate, and in the process of inspecting the leeward surface, the wind wheel center structure centroid coordinate is the cabin centroid coordinate.

[0128] In the above embodiment, the inspection direction of the blade and the tower drum is determined, so that the subsequent unmanned aerial vehicle can more accurately obtain the inspection image.

[0129] In some embodiments, referring to Figure 12 , the component safety distance includes the blade length, the component inspection image overlap rate includes the blade inspection image overlap rate, the component inspection shooting frequency includes the blade inspection shooting frequency, and based on the component safety distance, the component inspection image overlap rate and the component inspection shooting frequency, the component single-step moving distance sequence set and the component total moving distance can include the following steps: S1210, determine the single-shot coverage length of the blade based on the blade length, the blade inspection image overlap rate and the blade inspection shooting frequency.

[0130] S1220, based on the single-shot coverage length and the blade inspection image overlap rate, calculate the first step distance of the blade and the subsequent step length of the blade.

[0131] S1230, determine the blade single-step moving distance sequence set based on the first step distance of the blade, the subsequent step length of the blade and the blade inspection shooting frequency.

[0132] S1240, determine the total moving distance of the blade based on the blade length and the single-shot coverage length of the blade.

[0133] Specifically, in order to make the inspection image cover the blade completely, a blade inspection image overlap rate (such as 20%) is set. The blade single shooting coverage length is determined by calculation using the blade length, the blade inspection image overlap rate and the blade inspection shooting times, which represents the blade length that can be effectively covered by a single inspection image. The blade first step distance is obtained by dividing the blade single shooting coverage length by the integer 2. The blade subsequent step length is determined by calculation using the blade single shooting coverage length and the blade inspection image overlap rate. According to the blade inspection shooting times, the blade subsequent step length can be repeated for a specific number of times to ensure that the distribution of the shooting positions can meet the requirements of complete image coverage and the set overlap rate during the whole blade inspection process. In this way, the blade first step distance and the blade subsequent step length repeated for the specific number of times constitute a blade single step moving distance sequence set. The single step moving distance sequence set is used to guide the unmanned aerial vehicle to move from the actual blade center point to the blade tip area, and the distance of each single step movement is set. Based on the blade length and the blade single shooting coverage length, the blade total moving distance is determined by a corresponding calculation method. The blade total moving distance is used to set the distance from the blade tip to the actual blade center point after the single blade is inspected.

[0134] For example, the blade length is , the blade inspection image overlap rate is , and the blade inspection shooting times are . The following formula is used for calculation: The blade single shooting coverage length is calculated as

[0135] The blade single step moving distance sequence set is calculated as The blade first step distance (from the actual blade center point to the center of the first shooting point)

[0136] The blade subsequent step length (shooting point moving distance)

[0137] The blade single step moving distance sequence set is:

[0138] The blade total moving distance is calculated as

[0139] ​​​It should be noted that since the length of each blade of the wind turbine is consistent, the blade single-step moving distance sequence set and the blade total moving distance of each blade in the upwind inspection process and the blade single-step moving distance sequence set and the blade total moving distance of each blade in the downwind inspection process are consistent.

[0140] In the above embodiment, by determining the blade single-step moving distance sequence set and the blade total moving distance, the dynamic planning of the inspection moving distance is realized.

[0141] In some embodiments, referring to Figure 13 , the component safety distance includes the tower safety height, the component inspection image overlap rate includes the tower inspection image overlap rate, the component inspection shooting frequency includes the tower inspection shooting frequency, and based on the component safety distance, the component inspection image overlap rate and the component inspection shooting frequency, the determination of the component single-step moving distance sequence set and the component total moving distance can include the following steps: S1310, based on the tower safety height, the tower inspection image overlap rate and the tower inspection shooting frequency, the tower single-shot coverage length is determined.

[0142] S1320, based on the tower single-shot coverage length and the tower inspection image overlap rate, the tower first-step distance and the tower subsequent step length are determined.

[0143] S1330, based on the tower first-step distance, the tower subsequent step length and the tower inspection shooting frequency, the tower single-step moving distance sequence set is determined.

[0144] S1340, based on the tower safety height and the tower single-shot coverage length, the tower total moving distance is determined.

[0145] Specifically, in the process of tower inspection, in order to make the inspection image cover the tower completely, the tower inspection image overlap rate (such as 20%) is set. The tower single shot coverage length is determined by calculation using the tower safe height, the tower inspection image overlap rate and the tower inspection shot number, which represents the length of the tower that can be effectively covered by a single inspection image. The tower first step distance is obtained by dividing the tower single shot coverage length by the integer 2. The tower subsequent step length is determined by calculation using the tower single shot coverage length and the tower inspection image overlap rate. According to the tower inspection shot number, the tower subsequent step length can be repeated for a specific number of times to ensure that the distribution of the shot points can meet the requirements of complete image coverage and the set overlap rate in the whole tower inspection process. In this way, the tower first step distance and the tower subsequent step length repeated for a specific number of times constitute the blade single step moving distance sequence set. The single step moving distance sequence set is used to guide the UAV to move from the actual blade center point to the bottom area of the tower, and the distance of each single step moving is set. Based on the tower safe height and the tower single shot coverage length, the tower total moving distance is determined by the corresponding calculation method. The tower total moving distance is used to set the distance from the bottom of the tower to the actual blade center point after the tower is inspected.

[0146] For example, the height of the fan tower is , the number of tower inspection shots is , the lowest safe height of the tower is , and the inspection image overlap rate is . The tower safe height is determined based on the height of the fan tower and the number of tower inspection shots . The calculation is as follows: The tower single shot coverage length is calculated as follows:

[0147] The tower single step moving distance sequence set is calculated as follows: The tower first step distance (from the actual blade center point to the center of the first shot point)

[0148] The tower subsequent step length (moving distance between shot points)

[0149] The tower single step moving distance sequence set

[0150] The tower total moving distance is calculated as follows:

[0151] It should be noted that since the tower drum safety height is consistent, the tower drum single-step moving distance sequence set and the tower drum total moving distance of the upwind inspection process and the tower drum single-step moving distance sequence set and the tower drum total moving distance of the leeward inspection process are consistent.

[0152] In the above embodiment, by determining the tower drum single-step moving distance sequence set and the tower drum total moving distance, the dynamic planning of the inspection moving distance is realized.

[0153] In some embodiments, referring to Figure 14a , based on the component reference flight direction, the component single-step moving distance sequence set and the component total moving distance, the image acquisition is performed to obtain the component inspection image set, which can include the following steps: S1410, based on the third blade flight direction and the blade single-step moving distance sequence set, the unmanned aerial vehicle is controlled to fly from the actual blade center point and perform image acquisition to obtain a third blade inspection image set.

[0154] S1420, based on the blade total moving distance, the unmanned aerial vehicle is controlled to fly to the actual blade center point.

[0155] Specifically, based on the flight control system of the unmanned aerial vehicle, combined with the third blade flight direction and the blade single-step moving distance sequence set, the unmanned aerial vehicle is ensured to fly from the actual blade center point along the third blade flight direction for inspection by a control algorithm. During the inspection flight of the unmanned aerial vehicle along the third blade flight direction, the unmanned aerial vehicle gradually advances according to the blade single-step moving distance sequence set, and after each single-step moving distance flight, image data is acquired in real time to obtain the third blade inspection image of the corresponding position. After each flight and image acquisition, the inspection task of a single-step moving distance is completed, and finally a complete third blade inspection image set is formed. After the unmanned aerial vehicle completes the third blade inspection image acquisition, the unmanned aerial vehicle will adjust the flight direction and fly back the blade total moving distance in the opposite direction based on the flight control system and the control algorithm, so that the unmanned aerial vehicle returns to the actual blade center point for the next blade inspection image acquisition.

[0156] Exemplarily, the third blade flight direction is , the blade single-step moving distance sequence set is , and the blade total moving distance is . The unmanned aerial vehicle is controlled to fly from the actual blade center point to the direction, the first flight distance is , and the third blade inspection image acquisition is performed after the flight. Then continue to fly in the direction, the flight distance is each time, and the above steps are repeated After each flight, the third blade inspection image is collected. After the flight of the blade single-step moving distance sequence set is completed, the third blade inspection image set can be obtained. Finally, the UAV is controlled to fly in the opposite direction and return to the actual blade center point.

[0157] It should be noted that during the upwind inspection, the third blade inspection is performed once according to the above process, and during the leeward inspection, the third blade inspection is performed once according to the above process.

[0158] S1430, based on the fourth blade flight direction and the blade single-step moving distance sequence set, controlling the UAV to fly from the actual blade center point and collect images to obtain a fourth blade inspection image set.

[0159] S1440, based on the total blade moving distance, controlling the UAV to fly to the actual blade center point.

[0160] Specifically, based on the flight control system of the UAV, combined with the fourth blade flight direction and the blade single-step moving distance sequence set, the control algorithm ensures that the UAV starts from the actual blade center point and performs inspection flight along the fourth blade flight direction. During the inspection flight of the UAV along the fourth blade flight direction, the UAV gradually advances according to the blade single-step moving distance sequence set, and after each single-step moving distance flight is completed, image data is collected in real time to obtain the fourth blade inspection image of the corresponding position. After each flight and image collection, a single-step moving distance inspection task is completed, and finally a complete fourth blade inspection image set is formed. After the UAV completes the fourth blade inspection image collection, the UAV will adjust the flight direction and fly in the opposite direction of the total blade moving distance based on the flight control system and the control algorithm, so that the UAV returns to the actual blade center point for the next blade inspection image collection.

[0161] S1450, based on the fifth blade flight direction and the blade single-step moving distance sequence set, controlling the UAV to fly from the actual blade center point and collect images to obtain a fifth blade inspection image set.

[0162] S1460, based on the total blade moving distance, controlling the UAV to fly to the actual blade center point.

[0163] Specifically, based on the flight control system of the unmanned aerial vehicle, combined with the fifth blade flight direction and the blade single-step moving distance sequence set, the control algorithm ensures that the unmanned aerial vehicle starts from the actual blade center point and flies along the fifth blade flight direction for inspection. In the process of the unmanned aerial vehicle flying along the fifth blade flight direction for inspection, the unmanned aerial vehicle gradually advances according to the blade single-step moving distance sequence set, and after completing a single-step moving distance flight each time, real-time image data is collected to obtain the fifth blade inspection image at the corresponding position. After each flight and image collection, the inspection task of a single-step moving distance is completed, and finally a complete set of fifth blade inspection images is formed. After the unmanned aerial vehicle completes the fifth blade inspection image collection, the unmanned aerial vehicle will adjust the flight direction and fly the total moving distance of the blade in reverse based on the flight control system and the control algorithm, so that the unmanned aerial vehicle returns to the actual blade center point for the next blade inspection image collection.

[0164] Exemplarily, please refer to Figure 14b , Figure 14b to inspect the images of the wind turbine blades and the tower.

[0165] In the above implementation, by inspecting each blade of the wind turbine, the collection of each blade inspection image is realized, so as to realize the operation and maintenance detection of each blade subsequently.

[0166] In some embodiments, please refer to Figure 15 , based on the component reference flight direction, the component single-step moving distance sequence set and the component total moving distance for image collection to obtain the component inspection image set, which can include the following steps: S1510, based on the tower flight direction and the tower single-step moving distance sequence set, controlling the unmanned aerial vehicle to fly from the actual blade center point and collect images to obtain the tower inspection image set.

[0167] S1520, based on the total moving distance of the tower, controlling the unmanned aerial vehicle to fly to the actual blade center point.

[0168] Specifically, based on the flight control system of the unmanned aerial vehicle, combined with the tower flight direction and the tower single-step moving distance sequence set, the control algorithm ensures that the unmanned aerial vehicle starts from the actual blade center point and flies along the tower flight direction for inspection. In the process of the unmanned aerial vehicle flying along the tower flight direction for inspection, the unmanned aerial vehicle gradually advances according to the tower single-step moving distance sequence set, and after completing a single-step moving distance flight each time, real-time image data is collected to obtain the tower inspection image at the corresponding position. After each flight and image collection, the inspection task of a single-step moving distance is completed, and finally a complete set of tower inspection images is formed. After the unmanned aerial vehicle completes the tower inspection image collection, the unmanned aerial vehicle will adjust the flight direction and fly the total moving distance of the tower in reverse based on the flight control system and the control algorithm, so that the unmanned aerial vehicle returns to the actual blade center point.

[0169] Exemplarily, please refer to Figure 14b , Figure 14b to inspect the image of the wind turbine blade and the tower. It should be noted that during the inspection of the windward surface, one tower inspection is performed according to the above process, and during the inspection of the leeward surface, one tower inspection is performed according to the above process.

[0170] In the above embodiment, by inspecting the tower of the wind turbine, the collection of the tower inspection image is realized, so as to realize the operation and maintenance detection of the tower subsequently.

[0171] In some embodiments, please refer to Figure 16 , the wind turbine swept surface image includes a wind turbine leeward surface image, based on the wind wheel axial flight direction, the horizontal safety distance and the blade safety height, the unmanned aerial vehicle is controlled to fly to the wind turbine blade center point and collect the wind turbine swept surface image, which can include the following steps: S1610, based on the leeward surface flight direction, the unmanned aerial vehicle is controlled to fly horizontally for a horizontal safety distance, and then fly downward for a blade safety height, so that the unmanned aerial vehicle flies to the wind turbine blade center point.

[0172] S1620, based on the leeward surface flight direction, the unmanned aerial vehicle is controlled to fly horizontally for a horizontal safety distance, and then fly downward for a blade safety height, so that the unmanned aerial vehicle flies to the wind turbine blade center point.

[0173] Wherein, the wind turbine leeward surface image corresponds to a leeward surface image center point.

[0174] Specifically, the length of the blade The distance between the safety distance of the wind turbine The sum of the distances is the blade safety height. In order to ensure the safety of the unmanned aerial vehicle during the inspection process, the unmanned aerial vehicle is controlled to fly horizontally along the leeward surface flight direction for a horizontal safety distance by using the control algorithm, so as to ensure that the unmanned aerial vehicle can effectively avoid the wind turbine blade during the flight process, and can be maintained within a suitable range for detection. The unmanned aerial vehicle can effectively avoid the wind turbine blade during the flight process, and can be maintained within a suitable range for detection. After completing the horizontal flight to reach the horizontal safety distance, the unmanned aerial vehicle is controlled to fly downward for a blade safety height, so that the unmanned aerial vehicle flies to the wind turbine blade center point. After reaching the wind turbine blade center point, the unmanned aerial vehicle needs to adjust the heading direction of the unmanned aerial vehicle based on the leeward surface flight direction, so as to ensure that the heading direction is aligned with the windward surface of the wind turbine, and then image collection is performed to obtain the wind turbine leeward surface image. The center point of the leeward surface image is determined by using the wind turbine leeward surface image. The horizontal safety distance is an adaptive inspection parameter.

[0175] In the above embodiment, the horizontal flight safety distance of the unmanned aerial vehicle is controlled based on the flight direction of the leeward side, and then the unmanned aerial vehicle flies to the safety height of the blade to fly to the center point of the blade, the heading of the unmanned aerial vehicle is adjusted based on the flight direction of the leeward side, and the image of the leeward side of the wind turbine is collected, thereby providing a data basis for subsequent blade and tower inspection.

[0176] In some embodiments, referring to Figure 17 , the second target area contour includes a nacelle segmentation contour, and based on the second target area contour, determining the centroid coordinates of the wind wheel center structure can include the following steps: S1710, based on the nacelle segmentation contour, the nacelle centroid coordinates are determined by calculation.

[0177] Specifically, in the inspection process of the leeward side, the image recognition algorithm is used to recognize the image of the leeward side of the wind turbine, and the segmentation contour recognized in the image can be determined. In the case where the image of the leeward side of the wind turbine includes the segmentation contour of three blades, the segmentation contour of the tower and the segmentation contour of the nacelle, the nacelle centroid coordinates are determined by calculating the segmentation contour of the nacelle through the centroid calculation formula. If it is determined that the image of the leeward side of the wind turbine does not include at least one of the segmentation contour of three blades, the segmentation contour of the tower and the segmentation contour of the nacelle, the unmanned aerial vehicle is controlled to return.

[0178] Based on the second target area contour, the centroid coordinates of the wind wheel center structure and the center point of the swept surface image, the actual error distance of the swept surface can include the following steps: S1720, based on the nacelle centroid coordinates and the center point of the leeward side image, the leeward side image error distance is determined by calculation.

[0179] S1730, based on the area of the nacelle segmentation contour and the actual area of the nacelle, the leeward side proportion coefficient is determined.

[0180] S1740, based on the leeward side image error distance and the leeward side proportion coefficient, the actual error distance of the leeward side is determined.

[0181] Specifically, the distance between the nacelle centroid coordinates and the center point of the leeward side image is calculated based on the geometric projection relationship, the distance value between the nacelle and the image center point is obtained, and the distance is taken as the leeward side image error distance. The area of the nacelle segmentation contour is calculated by the contour area calculation formula. Next, the calculated area of the nacelle segmentation contour is compared with the actual area of the nacelle, and the area proportion of the two is calculated to determine the leeward side proportion coefficient. Finally, the leeward side actual error distance is determined by calculating the leeward side image error distance and the leeward side proportion coefficient. The actual area of the nacelle is an adaptive inspection parameter.

[0182] Exemplarily, the center point of the leeward side image is , the centroid coordinate of the nacelle is , the area of the nacelle segmentation contour is , the actual area of the nacelle is The actual error distance of the leeward surface is calculated by the following formula:

[0183]

[0184]

[0185]

[0186]

[0187] wherein, is the width of the leeward surface image of the fan, is the height of the leeward surface image of the fan, is the error distance of the leeward surface image, is the proportion coefficient of the leeward surface, is the actual error distance of the leeward surface.

[0188] In the above embodiment, the centroid coordinate of the nacelle is determined based on the calculation of the nacelle segmentation contour, the error distance of the leeward surface image is determined based on the calculation of the centroid coordinate of the nacelle and the center point of the leeward surface image, the proportion coefficient of the leeward surface is determined based on the area of the nacelle segmentation contour and the actual area of the nacelle, and the actual error distance of the leeward surface is determined based on the error distance of the leeward surface image and the proportion coefficient of the leeward surface, thereby providing a data basis for subsequent correction of the windward surface of the unmanned aerial vehicle.

[0189] In some embodiments, referring to Figure 18 , the method can further include the following steps: S1810, in the case that the actual error distance of the leeward surface does not meet the preset condition of the swept surface, the flight direction of the leeward correction is determined based on the calculation of the centroid coordinate of the nacelle and the center point of the leeward surface image.

[0190] S1820, after controlling the unmanned aerial vehicle to fly the actual error distance of the leeward surface based on the flight direction of the leeward correction, the image of the corrected leeward surface is collected.

[0191] wherein the corrected leeward surface image corresponds to a center point of the leeward correction image.

[0192] ​Specifically, in the case that the actual error distance of the leeward surface does not meet the preset condition of the swept surface, it indicates that the position of the unmanned aerial vehicle is not in the preset central range of the leeward surface of the wind turbine, and there is a deviation phenomenon, so it is necessary to correct the position of the unmanned aerial vehicle in order to improve the accuracy and reliability of the overall inspection process, and avoid the problem of inaccurate inspection data or missing key parts due to position deviation. Taking the preset direction of the unmanned aerial vehicle as a reference, the angle between the current position of the unmanned aerial vehicle and the centroid coordinate of the cabin is calculated according to the geometric relationship between the centroid coordinate of the cabin and the center point of the leeward surface image, and the leeward correction flight direction is determined. It should be noted that clockwise is positive. Then, the control algorithm is used to control the unmanned aerial vehicle to fly along the leeward correction flight direction for the actual error distance of the leeward surface, and then reach the required position, and then perform image acquisition to obtain the corrected leeward surface image. The center point of the corrected leeward surface image is determined by using the corrected leeward surface image. The swept surface preset condition can be less than the center point correction threshold .

[0193] Exemplarily, based on the centroid coordinate of the cabin and the center point of the leeward surface image = ), the leeward correction flight direction is determined:

[0194] S1830, based on the corrected leeward surface image, the leeward correction target area contour is determined.

[0195] S1840, based on the leeward correction target area contour, the corrected cabin centroid coordinate and the leeward correction image center point, the corrected leeward actual error distance is determined.

[0196] Wherein, the leeward correction target area contour corresponds to the corrected cabin centroid coordinate; Specifically, the corrected windward surface image is identified by using an image recognition algorithm, and the leeward correction target area contour identified in the image can be determined. In the case that the leeward correction target area contour includes a segmented contour of three blades and a corrected cabin segmented contour, the corrected cabin centroid coordinate is determined by calculating the corrected cabin segmented contour by using a centroid calculation formula.

[0197] The distance between the corrected nacelle centroid coordinate and the center point of the leeward correction image is calculated based on the geometric projection relationship, the distance value between the nacelle and the image center point is obtained, and the distance is taken as the leeward surface correction image error distance. The area of the corrected nacelle segmentation contour is determined by calculating the corrected nacelle segmentation contour through the contour area calculation formula. Next, the area of the corrected nacelle segmentation contour obtained by calculation is compared with the actual area of the nacelle, and the area ratio conversion of the two is performed to determine the leeward surface correction proportion coefficient. Finally, the corrected leeward actual error distance is determined by calculating the leeward surface correction image error distance and the leeward surface correction proportion coefficient.

[0198] S1850, based on the corrected leeward actual error distance and the sweep surface preset condition, the corrected leeward actual error distance is compared with the sweep surface preset condition, and the iteration number is determined. If the corrected leeward actual error distance does not meet the sweep surface preset condition and the iteration number meets the leeward surface iteration preset number, the above process is iteratively executed until the corrected leeward actual error distance meets the sweep surface preset condition and the corrected leeward surface image is taken as the wind turbine leeward surface image.

[0199] Specifically, the corrected leeward actual error distance is compared with the sweep surface preset condition. If the corrected leeward actual error distance does not meet the sweep surface preset condition, the number of position corrections performed by the current unmanned aerial vehicle is determined to determine the iteration number. If the iteration number does not meet the leeward surface iteration preset number, the above process can be re-executed until the corrected leeward actual error distance meets the sweep surface preset condition, at which time the corrected leeward surface image is taken as the wind turbine leeward surface image. If the iteration number meets the leeward surface iteration preset number, the unmanned aerial vehicle is controlled to return. The leeward surface iteration preset number can be the maximum correction number in the single inspection stage .

[0200] In the above embodiments, the position of the unmanned aerial vehicle is corrected to achieve more accurate unmanned aerial vehicle flight movement and acquisition of the inspection image.

[0201] In some embodiments, the adjustment of the camera focal length multiple of the unmanned aerial vehicle can be realized according to the obtained image and the calculated parameters, so as to realize more accurate image acquisition.

[0202] The embodiments of the present application provide an unmanned aerial vehicle-based wind turbine inspection device 1900, please refer to Figure 19 The unmanned aerial vehicle-based wind turbine inspection device 1900 includes a wind turbine top image acquisition module 1910, a wind turbine sweep surface image acquisition module 1920, a sweep surface actual error distance determination module 1930, a component reference flight direction determination module 1940, a component movement data determination module 1950, and a component inspection image acquisition module 1960.

[0203] The wind turbine top image acquisition module 1910 is configured to acquire a wind turbine top image when the UAV flies to a cruising initial position and the power is in a preset range, where the cruising initial position includes coordinates and a safety height of the wind turbine; The wind turbine swept surface image acquisition module 1920 is configured to control the UAV to fly to a wind turbine blade center point of the wind turbine based on the wind turbine top image and acquire a wind turbine swept surface image. The swept surface actual error distance determination module 1930 is configured to determine a swept surface actual error distance based on the wind turbine swept surface image. The component reference flight direction determination module 1940 is configured to determine a component reference flight direction of the wind turbine when the swept surface actual error distance meets a swept surface preset condition. The component movement data determination module 1950 is configured to determine a component single-step movement distance sequence set and a component total movement distance based on a component safety distance, a component inspection image overlap rate, and a component inspection shooting frequency. The component inspection image acquisition module 1960 is configured to acquire images based on the component reference flight direction, the component single-step movement distance sequence set, and the component total movement distance to obtain a component inspection image set.

[0204] For specific descriptions of the wind turbine inspection device based on the UAV, refer to the descriptions of the wind turbine inspection method based on the UAV, which will not be repeated here.

[0205] In some embodiments, a computer device, which can be a terminal, can have an internal structure as shown in FIG. 19. Figure 20 The computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected through a system bus. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for running the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is configured to perform wired or wireless communication with external terminals. The wireless communication can be achieved through WIFI, a carrier network, NFC (Near Field Communication), or other technologies. The computer program is executed by the processor to implement a wind turbine inspection method based on a UAV. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer overlaid on the display screen, or a key, trackball, or touchpad arranged on the housing of the computer device. The input device can also be an external keyboard, touchpad, or mouse, etc.

[0206] Those skilled in the art can understand that Figure 20 The structure shown in the figure is only a block diagram of part of the structure related to the solution disclosed in the specification, and does not constitute a limitation on the computer device to which the solution disclosed in the specification is applied. Specifically, the computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0207] In some embodiments, a computer device is provided, including a memory and a processor, the memory storing a computer program, and the processor implementing the method steps in the above embodiments when executing the computer program.

[0208] An embodiment of the specification provides a computer readable storage medium, which stores a computer program, and the computer program implements the steps of the method in any of the above embodiments when executed by a processor.

[0209] An embodiment of the specification provides a computer program product, which includes instructions, and the instructions are executed by a processor of a computer device to enable the computer device to perform the steps of the method in any of the above embodiments.

[0210] It should be noted that the logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a list of executable instructions for implementing logic functions, and can be specifically embodied in any computer readable medium for use by or in conjunction with an instruction execution system, apparatus or device, such as a computer-based system, a system including a processor or other system that can fetch and execute instructions from an instruction execution system, apparatus or device. For the purpose of the specification, "computer readable medium" can be any device that can contain, store, communicate, propagate or transport programs for use by or in conjunction with an instruction execution system, apparatus or device, or in conjunction with these instruction execution systems, apparatus or devices. More specific examples (non-exhaustive list) of computer readable medium include the following: electrical connections having one or more wires (electronic devices), portable computer diskettes (magnetic devices), random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memories), fiber optic devices, and portable compact disk read-only memories (CDROMs). In addition, the computer readable medium can even be paper or other suitable medium on which the program can be printed, as the program can be electronically obtained, for example, by optical scanning of the paper or other medium, followed by editing, interpretation or necessary processing, and then stored in a computer memory if necessary.

Claims

1. A wind turbine inspection method based on drone, characterized in that: The method comprises: When the UAV reaches an initial cruising position and the battery level is within a preset range, an image of the top of the wind turbine is acquired, wherein the initial cruising position includes the coordinates and safety height of the wind turbine; Based on the wind turbine top image, controlling the drone to fly to the center point of the wind turbine blade of the wind turbine and collecting the wind turbine swept surface image; determining an actual error distance of the swept surface based on the wind turbine swept surface image; determining a reference flight direction of a component of the wind turbine when the actual error distance of the swept surface meets a preset condition of the swept surface; Based on the component safety distance, component inspection image overlap rate and component inspection shooting times, the component single-step movement distance sequence set and the component total movement distance are determined; Image acquisition is performed based on the component reference flight direction, the component single-step movement distance sequence set, and the component total movement distance to obtain a component inspection image set.

2. The method according to claim 1, characterized in that The wind turbine top image corresponds to a top image center point, and based on the wind turbine top image, controlling the drone to fly to the center point of the wind turbine blade of the wind turbine and collecting the wind turbine swept surface image includes: Performing recognition based on the wind turbine top image to determine a first target area outline; Perform calculation based on the first target area outline to determine the vertical foot coordinates of the blade; Determining a top actual error distance based on the first target area outline, the blade vertical foot coordinates, and the top image center point; When the actual top error distance meets the top preset condition, determining the axial flight direction of the wind rotor based on the vertical foot coordinate of the blade; Based on the axial flight direction of the wind rotor, the horizontal safety distance and the blade safety height, the UAV is controlled to fly to the center point of the wind turbine blade and collect the image of the wind turbine swept surface.

3. The method according to claim 1, characterized in that The wind turbine swept surface image corresponds to a swept surface image center point, and determining the actual swept surface error distance based on the wind turbine swept surface image includes: Performing recognition based on the wind turbine swept surface image to determine the contour of the second target area; Determining the centroid coordinates of the central structure of the wind rotor based on the outline of the second target area; The actual error distance of the swept surface is determined based on the second target area outline, the centroid coordinates of the wind wheel central structure and the center point of the swept surface image.

4. The method according to claim 2, characterized in that The first target area contour includes a first blade segmentation contour, a second blade segmentation contour, and a nacelle segmentation contour. The calculating based on the first target area contour to determine the blade vertical foot coordinates includes: Perform calculation based on the segmented contour of the first blade to determine the coordinates of the center of mass of the first blade; Perform calculation based on the segmented contour of the second blade to determine the coordinates of the center of mass of the second blade; Calculate based on the cabin segmentation outline to determine the coordinates of the cabin center of mass; The blade vertical foot coordinate is determined by performing calculation based on the first blade mass center coordinate, the second blade mass center coordinate, and the nacelle mass center coordinate.

5. The method according to claim 4, characterized in that The determining of the top actual error distance based on the first target area outline, the blade vertical foot coordinates, and the top image center point includes: Perform calculation based on the blade vertical foot coordinate and the center point of the top image to determine the top image error distance; determining a top scale factor based on the area of ​​the cabin segmentation outline and the actual area of ​​the cabin top; The top actual error distance is determined based on the top image error distance and the top scale factor.

6. The method according to claim 2, characterized in that The method further comprises: When the actual top error distance does not meet the top preset condition, a calculation is performed based on the vertical foot coordinates of the blade and the center point of the top image to determine the corrected flight direction; After controlling the UAV to fly the actual top error distance based on the corrected flight direction, collecting a corrected image of the wind turbine top, wherein the corrected image of the wind turbine top corresponds to a center point of the corrected image; Performing recognition based on the corrected wind turbine top image to determine a correction target area outline, wherein the correction target area outline corresponds to the corrected nacelle centroid coordinates; Perform calculation based on the correction target area contour to determine the correction blade vertical foot coordinate; Determining the actual error distance of the top after correction based on the correction target area contour, the correction blade vertical coordinates and the correction image center point; Based on the comparison between the corrected actual error distance of the top and the top preset condition, when the corrected actual error distance of the top does not meet the top preset condition and the number of iterations does not meet the preset number of iterations, the above process is iteratively executed until the corrected actual error distance of the top meets the top preset condition and the corrected cabin center of mass coordinates are used as the cabin center of mass coordinates, and the corrected blade vertical foot coordinates are used as the blade vertical foot coordinates.

7. The method according to claim 2 or 6, characterized in that The axial flight direction of the wind rotor includes a windward flight direction and a leeward flight direction. When the top actual error distance meets the top preset condition, determining the axial flight direction of the wind rotor based on the blade vertical foot coordinate includes: Determining the flight direction of the windward surface based on the coordinates of the center of mass of the nacelle and the vertical foot coordinates of the blade; The leeward flight direction is determined based on the blade vertical foot coordinates and the nacelle center of mass coordinates.

8. The method according to claim 7, characterized in that The wind turbine swept surface image includes a windward surface image of the wind turbine. The controlling the UAV to fly to the center point of the wind turbine blade and collect the wind turbine swept surface image based on the axial flight direction of the wind rotor, the horizontal safety distance, and the blade safety height includes: Based on the windward flight direction, the UAV is controlled to fly horizontally to the horizontal safety distance, and then to fly downward to the blade safety height, so that the UAV flies to the center point of the wind turbine blade; The nose orientation of the UAV is adjusted based on the windward flight direction and the windward image of the wind turbine is collected, wherein the windward image of the wind turbine corresponds to a center point of the windward image.

9. The method according to claim 8, characterized in that The second target area contour includes a hub segmentation contour. Determining the centroid coordinates of the central structure of the wind rotor based on the second target area contour includes: Perform calculation based on the hub segmentation profile to determine the hub center of mass coordinates; Determining an actual error distance of the swept surface based on the second target area contour, the centroid coordinates of the wind rotor central structure, and the center point of the swept surface image includes: Calculating based on the hub center of mass coordinates and the center point of the windward surface image to determine the windward surface image error distance; Determine the windward surface proportionality factor based on the area of ​​the hub segmentation profile and the actual hub area; An actual windward surface error distance is determined based on the windward surface image error distance and the windward surface proportional coefficient.

10. The method according to claim 9, characterized in that The method further comprises: When the actual error distance of the windward surface does not meet the preset conditions of the swept surface, a calculation is performed based on the coordinates of the hub center of mass and the center point of the windward surface image to determine the windward correction flight direction; After controlling the UAV to fly the actual windward error distance based on the windward correction flight direction, collecting a corrected windward surface image, wherein the corrected windward surface image corresponds to a center point of the windward correction image; Recognizing the corrected windward surface image to determine a windward correction target area contour, wherein the windward correction target area contour corresponds to a corrected hub mass center coordinate; Determining the corrected actual windward error distance based on the windward correction target area contour, the corrected hub mass center coordinates, and the windward correction image center point; Based on the comparison between the corrected actual windward error distance and the swept surface preset conditions, when the corrected actual windward error distance does not meet the swept surface preset conditions and the number of iterations meets the windward surface preset number of iterations, the above process is iteratively executed until the corrected actual windward error distance meets the swept surface preset conditions and the corrected windward surface image is used as the windward surface image of the wind turbine.

11. The method according to claim 3, characterized in that The second target area further includes a third blade segmentation contour, a fourth blade segmentation contour, a fifth blade segmentation contour, and a tower segmentation contour. When the actual error distance of the swept surface meets a preset condition of the swept surface, determining the reference flight direction of the components of the wind turbine includes: Perform calculation based on the segmented contour of the third blade to determine the coordinates of the centroid of the third blade; Perform calculation based on the segmented outline of the fourth blade to determine the centroid coordinates of the fourth blade; Perform calculation based on the segmented outline of the fifth blade to determine the coordinates of the centroid of the fifth blade; Calculate based on the tower segmentation outline to determine the tower center of mass coordinates; Determining a flight direction of the third blade based on the coordinates of the mass center of the third blade and the coordinates of the mass center of the wind rotor center structure; determining a flight direction of the fourth blade based on the coordinates of the centroid of the fourth blade and the centroid of the central structure of the wind rotor; determining a flight direction of the fifth blade based on the coordinates of the mass center of the fifth blade and the coordinates of the mass center of the wind rotor central structure; The tower flight direction is determined based on the tower center of mass coordinates and the wind wheel center structure center of mass coordinates.

12. The method according to claim 1, characterized in that The component safety distance includes the blade length, the component inspection image overlap rate includes the blade inspection image overlap rate, and the number of component inspection shots includes the number of blade inspection shots. Determining the component single-step movement distance sequence set and the component total movement distance based on the component safety distance, the component inspection image overlap rate, and the number of component inspection shots includes: Determining a single-shot coverage length of the blade based on the blade length, the blade inspection image overlap rate, and the number of blade inspection shots; Calculating based on the single shot coverage length and the blade inspection image overlap rate to determine the blade first step distance and the blade subsequent step length; Determining a set of single-step movement distance sequences of the blade based on the first-step distance of the blade, the subsequent step length of the blade, and the number of inspection shots of the blade; The total movement distance of the blade is determined based on the blade length and the single-shot coverage length of the blade.

13. The method according to claim 1, wherein The component safety distance includes the tower safety height, the component inspection image overlap rate includes the tower inspection image overlap rate, the component inspection shooting times includes the tower inspection shooting times, and determining the component single-step movement distance sequence set and the component total movement distance based on the component safety distance, the component inspection image overlap rate, and the component inspection shooting times includes: Determining a single-shot coverage length of the tower based on the tower safety height, the tower inspection image overlap rate, and the number of tower inspection shots; Calculating based on the tower single shot coverage length and the tower inspection image overlap rate to determine the tower first step distance and the tower subsequent step length; Determine a tower single-step movement distance sequence set based on the tower first step distance, the tower subsequent step length, and the number of tower inspection and shooting times; The total moving distance of the tower is determined based on the tower safety height and the tower single shot coverage length.

14. The method according to claim 12, characterized in that The image acquisition based on the component reference flight direction, the component single-step movement distance sequence set and the component total movement distance to obtain the component inspection image set includes: Based on the flight direction of the third blade and the single-step movement distance sequence set of the blade, the drone is controlled to fly from the actual blade center point and perform image acquisition to obtain a third blade inspection image set; Controlling the UAV to fly to the actual blade center point based on the total movement distance of the blade; Based on the flight direction of the fourth blade and the sequence set of single-step movement distances of the blade, the drone is controlled to fly from the actual center point of the blade and to collect images, thereby obtaining a fourth blade inspection image set; Controlling the UAV to fly to the actual blade center point based on the total movement distance of the blade; Based on the flight direction of the fifth blade and the sequence set of single-step movement distances of the blade, the drone is controlled to fly from the actual center point of the blade and to collect images, thereby obtaining a fifth blade inspection image set; The UAV is controlled to fly to the actual blade center point based on the total movement distance of the blade.

15. The method according to claim 13, characterized in that The image acquisition based on the component reference flight direction, the component single-step movement distance sequence set and the component total movement distance to obtain the component inspection image set includes: Based on the tower flight direction and the tower single-step movement distance sequence set, the drone is controlled to fly from the actual blade center point and perform image acquisition to obtain a tower inspection image set; The drone is controlled to fly to the actual blade center point based on the total moving distance of the tower.

16. The method according to claim 7, characterized in that The wind turbine swept surface image includes a wind turbine leeward surface image. The controlling the UAV to fly to the center point of the wind turbine blade and collect the wind turbine swept surface image based on the axial flight direction of the wind rotor, the horizontal safety distance, and the blade safety height includes: Based on the leeward flight direction, the UAV is controlled to fly horizontally for the horizontal safety distance, and then to fly downward for the blade safety height, so that the UAV flies to the center point of the wind turbine blade; The nose orientation of the UAV is adjusted based on the leeward flight direction and the leeward image of the wind turbine is collected, wherein the leeward image of the wind turbine corresponds to a leeward image center point.

17. The method according to claim 16, characterized in that The second target area outline includes the nacelle segmentation outline. Based on the second target area outline, determining the centroid coordinates of the central structure of the wind rotor includes: Perform calculation based on the cabin segmentation outline to determine the coordinates of the cabin center of mass; Determining an actual error distance of the swept surface based on the second target area contour, the centroid coordinates of the wind rotor central structure, and the center point of the swept surface image includes: Calculating based on the coordinates of the center of mass of the cabin and the center point of the leeward image to determine the leeward image error distance; Determine the leeward side proportional coefficient based on the area of ​​the cabin segmentation outline and the actual area of ​​the cabin; An actual leeward side error distance is determined based on the leeward side image error distance and the leeward side proportional coefficient.

18. The method according to claim 17, characterized in that The method further comprises: When the actual error distance of the leeward surface does not meet the preset conditions of the swept surface, a calculation is performed based on the coordinates of the center of mass of the cabin and the center point of the leeward surface image to determine the leeward corrected flight direction; After controlling the UAV to fly the actual leeward error distance based on the leeward corrected flight direction, collecting a corrected leeward image, wherein the corrected leeward image corresponds to a center point of the leeward corrected image; performing recognition based on the corrected leeward surface image to determine a leeward correction target area contour, wherein the leeward correction target area contour corresponds to a corrected cabin centroid coordinate; determining a corrected lee wind actual error distance based on the lee wind correction target area contour, the corrected cabin centroid coordinates, and the lee wind correction image center point; Based on the comparison between the corrected actual leewind error distance and the swept surface preset conditions, when the corrected actual leewind error distance does not meet the swept surface preset conditions and the number of iterations meets the leeward surface iteration preset number, the above process is iteratively executed until the corrected actual leewind error distance meets the swept surface preset conditions and the corrected leeward surface image is used as the leeward surface image of the wind turbine.

19. A wind turbine inspection device based on a drone, characterized in that: The device comprises: a wind turbine top image acquisition module, configured to acquire an image of the wind turbine top when the UAV reaches an initial cruising position and the battery level is within a preset range, wherein the initial cruising position includes the coordinates and safety height of the wind turbine; A wind turbine swept surface image acquisition module is used to control the UAV to fly to the center point of the wind turbine blade of the wind turbine and acquire a wind turbine swept surface image based on the wind turbine top image; A swept surface actual error distance determination module, configured to determine a swept surface actual error distance based on the wind turbine swept surface image; a component reference flight direction determination module, configured to determine a component reference flight direction of the wind turbine when the actual error distance of the swept surface meets a preset condition of the swept surface; A component movement data determination module is used to determine a component single-step movement distance sequence set and a component total movement distance based on a component safety distance, a component inspection image overlap rate, and a component inspection shooting number; The component inspection image acquisition module is used to acquire images based on the component reference flight direction, the component single-step movement distance sequence set and the component total movement distance to obtain a component inspection image set.

20. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 18 are implemented.

21. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 18 are implemented.

Images