Power equipment defect detection method based on unmanned aerial vehicle inspection

Abstract

The invention relates to the technical field of power equipment inspection and detection, and relates to a power equipment defect detection method based on unmanned aerial vehicle inspection. The method determines the radial detection distance of the unmanned aerial vehicle through the size data of the fan blades of the wind driven generator, plans the inspection route in combination with the inspection flight speed of the unmanned aerial vehicle, and judges whether an abnormal region exists according to the collected fan blade inspection image and determines the boundary contour of the abnormal region. And analyzing the annular flight speed of the unmanned aerial vehicle at the directional inspection position by combining the current fan blade real-time rotating speed, realizing directional refined inspection of the fan blade defect area, ensuring the fan blade surface image acquisition quality, screening the surface image with qualified definition, identifying the defect area feature data in the surface image, and judging the health state of the fan blade. When the healthy states of all the fan blades are normal states, the overall balance stability of the wind generating set is analyzed, comprehensive control over overall balance stability analysis of the wind generating set is achieved, and the overall operation life of the wind generating set is prolonged.

Description

Technical Field

[0001] This invention relates to the field of power equipment inspection and testing technology, and specifically to a method for detecting defects in power equipment based on unmanned aerial vehicle (UAV) inspection. Background Technology

[0002] As the core component for capturing wind energy, wind turbine blades are subjected to alternating loads, environmental erosion, and fatigue stress over long periods, making them prone to defects such as cracks, corrosion, and delamination. Failure to detect and address these defects in a timely manner can lead to serious equipment failures or even safety accidents. Therefore, achieving efficient and accurate blade defect detection is of great significance for ensuring the safe operation of wind turbine generators.

[0003] With the rapid development of drone technology, drone-based wind turbine blade inspection technology has gradually become a research hotspot in the industry. In existing technology, Chinese Patent Publication No. CN120747780A discloses a method and device for detecting wind turbine blade defects based on drone dynamic inspection. This method, based on blade motion parameters, drone flight state parameters, and shooting time parameters, performs deblurring and stitching processing on the initial blade images acquired during drone dynamic inspection to obtain a second image; it then uses the SIFT feature matching algorithm to extract features from the second image to obtain a third image; and finally, it employs a dual-channel convolutional neural network to perform defect detection on the third image, achieving image sharpening and defect detection of rotating blades.

[0004] The existing technology has the following problems: 1. The existing technology only predicts motion vectors through blade motion parameters and UAV flight parameters to deal with image blurring. It does not combine blade size data to determine the radial detection distance of the UAV, nor does it plan an appropriate inspection route according to the blade rotation speed. It only uses a general inspection path to collect images, which has the problem of insufficient adaptability of the inspection route. This results in incomplete image collection and missed detection of some blade areas during UAV inspection. At the same time, it is easy to cause collision risk due to too close distance to the blade, which affects the safety and integrity of the inspection.

[0005] 2. Existing technologies only perform preprocessing such as deblurring and alignment on the acquired images before directly detecting defects. They do not first identify abnormal areas in the images and conduct targeted and refined inspections. They rely solely on a single image acquisition for defect judgment, which results in insufficient accuracy in defect detection. This leads to the inability to effectively identify small and hidden defects on the fan blade surface, making it easy to miss or misjudge, and failing to detect early defects in the fan blade in a timely manner.

[0006] 3. Existing technologies only focus on blade defect detection and do not analyze the overall balance and stability of the wind turbine generator set. They only judge the equipment operation status based on the defect status of a single blade, which has the problem of a single detection and analysis dimension. This makes it impossible to identify potential risks such as unit centroid shift and imbalance caused by local blade defects, which can easily lead to increased unit vibration, accelerated component wear, and reduced overall service life of the wind turbine generator set. Summary of the Invention

[0007] This invention aims to overcome the shortcomings of existing technologies and provide a method for detecting defects in power equipment based on drone inspection, so as to realize intelligent and accurate detection of defects in wind turbine blades and ensure the safe and stable operation of wind turbine units.

[0008] The technical solution adopted by the present invention to solve its technical problem is: a method for detecting defects in power equipment based on unmanned aerial vehicle (UAV) inspection, including: determining the radial detection distance between the UAV and the blade based on the size data of the wind turbine blade, obtaining the UAV inspection flight speed by combining the current real-time rotation speed of the blade, and planning the corresponding inspection route of the UAV.

[0009] Collect images of fan blades from a drone flying along the inspection route, determine if there are any abnormal areas in the images, and identify the boundary contours of the abnormal areas.

[0010] The location of the UAV for directional inspection is determined based on the boundary contour of the abnormal region. The circular flight speed of the UAV at the directional inspection location is analyzed in combination with the real-time rotation speed of the current fan blades, and surface image sequences covering different directional inspection locations are collected.

[0011] The surface image sequence is analyzed for sharpness, and surface images with acceptable sharpness are selected. The feature data of defect areas in the surface images are identified to determine the health status of the fan blades.

[0012] When all blades are in normal health condition, the overall balance and stability of the wind turbine generator set are analyzed based on the blade assembly parameters.

[0013] Compared with the prior art, the present invention has the following beneficial effects: (1) The present invention determines the radial detection distance of the UAV by the size data of the wind turbine blade, and calculates the inspection flight speed and direction deflection angle of the UAV according to the real-time rotation speed of the blade, and plans the inspection route of the UAV, realizing the adaptive dynamic route planning of the UAV in the state of blade rotation, ensuring that the UAV fully covers the blade area during inspection, avoiding missed inspections, and ensuring that the UAV and the blade maintain a safe distance, eliminating the risk of collision, and improving the safety of UAV inspection and the integrity of image acquisition.

[0014] (2) This invention collects images of fan blades by flying along the inspection route of the UAV, determines whether there are abnormal areas in the fan blade inspection images and determines their boundary contours, and analyzes the circular flight speed of the UAV at the directional inspection position in combination with the real-time rotation speed of the fan blade, thereby realizing directional and refined inspection of the defective areas of the fan blade, so that the UAV can remain relatively stationary with the fan blade surface through circular flight, ensuring the quality of the image acquisition of the fan blade surface.

[0015] (3) This invention performs clarity analysis on surface image sequences, selects surface images with qualified clarity, realizes quantitative evaluation of image clarity, provides data support for defect feature extraction and identification, identifies defect area feature data in surface images, determines the health status of fan blades, realizes evaluation of the impact of defect feature data on fan blade structural performance, effectively distinguishes between surface defects and structural defects, and improves the accuracy of defect judgment and the rationality of maintenance decisions.

[0016] (4) This invention analyzes the overall balance and stability of the wind turbine generator set based on the blade assembly parameters when all blades are in normal health condition, thereby achieving comprehensive control over the overall balance and stability analysis of the unit, timely identifying potential risks of unit imbalance caused by blade defects, avoiding problems such as increased vibration during unit operation and accelerated wear of components, and extending the overall service life of the wind turbine generator set. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram showing the flow of steps in the method of the present invention.

[0019] Figure 2 This is a schematic diagram illustrating the planning steps of the inspection route of the UAV in this invention.

[0020] Figure 3 This is a schematic diagram illustrating the overall balance and stability analysis steps of the wind turbine generator set in this invention. Detailed Implementation

[0021] Various exemplary embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It should be noted that, unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps set forth in these embodiments do not limit the scope of the invention. Furthermore, it should be understood that, for ease of description, the dimensions of the various parts shown in the drawings are not drawn to actual scale.

[0022] The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the invention or its application or use. Techniques, methods, and apparatus known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and apparatus should be considered part of the specification.

[0023] In all examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values.

[0024] Please see Figure 1 As shown, the present invention provides a method for detecting defects in power equipment based on unmanned aerial vehicle (UAV) inspection, including: S1, planning the UAV inspection route based on the UAV's radial detection distance and inspection flight speed.

[0025] Considering that wind turbine blades are in a continuous rotation state, general inspection routes are prone to problems such as missed inspections of blade areas and collisions between drones and blades. Furthermore, the mismatch between the inspection flight speed and the blade rotation speed can lead to blurry images. Therefore, it is necessary to determine a safe and suitable radial inspection distance based on the blade size data, and then plan an adaptive dynamic inspection route based on the real-time blade rotation speed to ensure comprehensive inspection coverage and flight safety, laying the foundation for subsequent image acquisition.

[0026] Based on this, the present invention determines the radial detection distance between the UAV and the blade based on the wind turbine blade size data, and obtains the UAV inspection flight speed by combining the current real-time rotation speed of the blade, and plans the corresponding UAV inspection route.

[0027] like Figure 2 As shown, the specific implementation of the present invention includes: S11, extracting the blade boundary contour from the wind turbine blade size data, and determining the radial detection distance between the UAV and the blade based on the reference shooting range diameter corresponding to different shooting distances of the UAV. The determination method is as follows: First, filter the maximum contour width of the blade boundary in the blade boundary contour, retrieve the reference shooting range diameter corresponding to different shooting distances of the UAV in the UAV equipment parameter library, and match the shooting distance corresponding to the maximum contour width from the reference shooting range diameter corresponding to different shooting distances of the UAV.

[0028] Then, if the shooting distance corresponding to the maximum profile width is greater than the safe shooting distance of the wind turbine blades, then the shooting distance is used as the radial detection distance between the drone and the blades.

[0029] Conversely, the safe shooting distance of the wind turbine blades is used as the radial detection distance between the drone and the blades, eliminating the risk of collision between the drone and the blades.

[0030] Preferably, in one embodiment of the present invention, the safe shooting distance for wind turbine blades is the maximum radial sway during the rotation of the blades in the historical operation of the wind turbine generator set.

[0031] S12. Based on the reference shooting range of the UAV at the corresponding radial detection distance, the fan blade boundary contour is divided into different boundary contour regions according to the reference shooting range. The different boundary contour regions are connected to each other without overlap, and the combination of different boundary contour regions can completely cover the entire fan blade.

[0032] S13. Establish a three-dimensional spatial coordinate system with the center of the wind turbine hub as the origin, and determine the timing sequence of the center position coordinates of different boundary contour regions based on the current real-time rotation speed of the blades.

[0033] S14. Obtain the unit time displacement corresponding to the center position of the same boundary contour area and the unit time interval displacement corresponding to the center position of adjacent boundary contour areas, and calculate the UAV inspection direction deflection angle and UAV inspection flight speed.

[0034] Preferably, in a specific embodiment of the present invention, for example, the center position of a certain boundary contour region is denoted as position A, and the center position of the adjacent boundary contour region corresponding to the boundary contour region is denoted as position B. Then, the unit time displacement corresponding to the center position of the same boundary contour region is the distance between adjacent unit time in the coordinate time sequence of position A, denoted as a; the unit time interval displacement corresponding to the center positions of adjacent boundary contour regions is the distance between the current unit time coordinate in the coordinate time sequence of position A and the next unit time coordinate in the coordinate time sequence of position B, denoted as b.

[0035] The coordinate distance between the centers of adjacent boundary contour regions within the same unit of time is obtained and denoted as c. The formula for calculating the UAV inspection direction deflection angle is then given. The calculation formula is existing technology and will not be elaborated here.

[0036] The method for calculating the flight speed of drone inspection is as follows: .

[0037] In the formula, For drone inspection flight speed, This is the ratio of the coordinate distance between the centers of adjacent boundary contour regions in the same unit of time to the unit of time itself. This is the current real-time rotational speed of the fan blades, which is the rotational speed at the position of the fan blade tip.

[0038] S15. Combining the radial detection distance between the UAV and the fan blades, establish a sequence of inspection path nodes according to the direction of the fan blade rotation plane, and connect them sequentially to form the UAV inspection route.

[0039] In one specific embodiment, the inspection path node sequence can be a coordinate node at position A corresponding to unit time 1, a coordinate node at position B corresponding to unit time 2, and a coordinate node at position C corresponding to unit time 3, etc., where unit time 2 is the next unit time after unit time 1, and position C is the center position of the adjacent boundary contour region of the boundary contour region corresponding to position B.

[0040] This invention determines the radial detection distance of a UAV by using wind turbine blade size data, and calculates the UAV's inspection flight speed and directional deflection angle based on the real-time rotation speed of the blades. It then plans the UAV's inspection route, enabling adaptive dynamic route planning for the UAV while the blades are rotating. This ensures that the UAV fully covers the blade area during inspection, avoids missed inspections, maintains a safe distance between the UAV and the blades, eliminates the risk of collision, and improves the safety of UAV inspection and the integrity of image acquisition.

[0041] S2. Judgment of abnormal areas and determination of boundary contours.

[0042] Considering that directly performing defect detection on UAV inspection images would increase computational load and easily overlook small abnormal areas, it is necessary to first stitch and integrate the inspection images, and quickly identify abnormal areas and determine their boundary contours by comparing them with standard defect-free images, thereby achieving pre-screening for defect detection and improving the targeting and efficiency of subsequent detection.

[0043] Based on this, the present invention acquires images of fan blades inspected by a drone flying along an inspection route, determines whether there are abnormal areas in the fan blade inspection images, and determines the boundary contours of the abnormal areas.

[0044] The specific implementation of the present invention includes: S21, stitching and integrating the fan blade inspection image set collected by the UAV along the inspection route to obtain the integrated overall image of the fan blade. The specific operation process is as follows: S211, controlling the UAV to fly at a constant speed along the planned inspection route, starting the high-definition camera equipment on the UAV, continuously collecting inspection images of the boundary contour areas of each fan blade to form a fan blade inspection image set.

[0045] S212. The image stitching method is used to perform feature matching and stitching integration on the images in the fan blade inspection image set, eliminating the overlapping areas and stitching gaps between images, and obtaining a complete overall image of the fan blade to restore the surface morphology of the fan blade.

[0046] It should be noted that image stitching methods are existing technologies well-known to those skilled in the art, and will not be described in detail here.

[0047] S22. Compare the grayscale values ​​of the overall image of the fan blade with the standard defect-free image of the fan blade retrieved from the wind turbine database. Calculate the grayscale deviation value of the corresponding pixel points between the overall image of the fan blade and the standard defect-free image. If the grayscale deviation value of a certain pixel point exceeds the preset grayscale deviation allowable range, mark the pixel point as a grayscale anomaly.

[0048] S23. Connect all gray-level abnormal points to adjacent regions. If there is a gray-level abnormal connected region, it is determined that there is an abnormal region in the fan blade inspection image. Then, edge detection is performed on the abnormal region, and the boundary contour of the abnormal region is fitted.

[0049] Preferably, in one embodiment of the present invention, considering that the closer the grayscale deviation value is to 0, the higher the grayscale consistency between the fan blade detection image and the standard defect-free image, and the greater the possibility of no abnormal area, a symmetrical preset grayscale deviation allowable range centered on 0 is set, for example [-5, 5], with grayscale values ​​ranging from 0 to 255. When the grayscale deviation value of the detected image pixel is within this range, it is determined to be a normal grayscale difference and is not marked as a grayscale abnormal point; otherwise, it is determined to be a grayscale abnormality and is marked as a grayscale abnormal point.

[0050] In other examples, implementers can also adjust the preset grayscale deviation allowable range themselves, and appropriately expand or shrink the interval threshold, but should not deviate too much from 0, so as to avoid affecting the recognition accuracy of abnormal areas due to improper threshold setting.

[0051] S3. Determine the circular flight speed of the UAV at the directional inspection location and collect surface image sequences.

[0052] Considering that the fan blades are rotating, it is difficult to obtain clear and comprehensive surface images of abnormal areas with a single shot. Furthermore, ordinary flight methods are prone to image blurring due to relative motion. Therefore, it is necessary to determine the directional inspection position for abnormal areas and calculate the circular flight speed of the drone based on the fan blade rotation speed. This will keep the drone relatively stationary with respect to the fan blade surface, enabling refined inspection of abnormal areas and improving image acquisition quality.

[0053] Based on this, the specific implementation of the present invention includes: S31, determining the directional inspection position of the UAV based on the boundary contour of the abnormal region. The determination method is as follows: S311, based on the boundary contour of the abnormal region, combined with the reference shooting range diameter corresponding to different shooting distances of the UAV, the directional inspection radial distance corresponding to the UAV is determined similarly to the method for determining the radial detection distance between the UAV and the fan blades.

[0054] S312. Extract the geometric center coordinates of the boundary contour of the abnormal area. Using these geometric center coordinates as a reference, and combining them with the radial distance of the UAV's directional inspection, determine the reference position for the UAV's directional inspection.

[0055] S313. Extend a preset distance from the reference position to the edge of the boundary contour of the abnormal area to determine multiple auxiliary inspection positions, and combine the reference position and the multiple auxiliary inspection positions to form the directional inspection position of the UAV.

[0056] S32. Analyze the UAV's circular flight speed at the directional inspection position based on the current real-time rotation speed of the blades. The specific analysis method is as follows: S321. Divide the blades into multiple segments radially according to different rotation radii. Based on the current real-time rotation speed of the blades and the rotation radius of each segment, obtain the surface tangential linear velocity of each segment, and extract the surface tangential linear velocity at the geometric center of the boundary contour.

[0057] It should be noted that the surface tangential linear velocity of each section is obtained as follows: based on the current real-time rotation speed of the fan blade and the rotation radius corresponding to the position of the fan blade tip, the fan blade angular velocity is obtained by ratio calculation, and the fan blade angular velocity is multiplied by the rotation radius of each section to obtain the surface tangential linear velocity of each section.

[0058] S322. Based on the radius of the UAV's circular inspection trajectory at the directional inspection location, calculate the reference circular flight speed required for the UAV to maintain relatively stationary observation. The specific calculation process is as follows: First, obtain the horizontal projection distance of the UAV's directional inspection location relative to the center of the wind turbine hub, and use this horizontal projection distance as the radius of the UAV's circular inspection trajectory.

[0059] Then, the ratio of the surface tangential linear velocity at the geometric center of the boundary profile to the radius of the UAV's circular inspection trajectory is analyzed to obtain the angular velocity required for the UAV to maintain relatively stationary observation.

[0060] Finally, the required angular velocity is compared with the allowable range of safe flight angular velocity for UAVs in the UAV equipment parameter library to determine the reference circular flight angular velocity of the UAV. This reference circular flight angular velocity is then multiplied by the radius of the UAV's circular inspection trajectory to obtain the vector magnitude of the reference circular flight velocity. The tangent direction of the circular inspection trajectory is taken as the vector direction of the reference circular flight velocity, and the reference circular flight velocity is constructed by combining the vector magnitude.

[0061] As an example, considering that excessively high angular velocities of the UAV can easily lead to flight loss of control, while excessively low angular velocities will prevent the realization of relative stationary observation with respect to the fan blades, a safe range of flight angular velocities is set with the minimum stable angular velocity corresponding to the UAV flight control system as the lower limit and the maximum controllable angular velocity corresponding to the flight control system as the upper limit.

[0062] It should be noted that if the required angular velocity is within the allowable range of safe flight angular velocity for the drone, then this angular velocity shall be used as the reference circular flight angular velocity for the drone.

[0063] If the required angular velocity exceeds the upper limit of the allowable range for safe flight angular velocity of the UAV, adjust the spatial coordinates of the UAV's directional inspection position, increase the radius of the circular inspection trajectory, and recalculate the angular velocity until it is within the allowable range.

[0064] If the required angular velocity is lower than the lower limit of the allowable range for safe flight angular velocity of the UAV, adjust the spatial coordinates of the UAV's directional inspection position, reduce the radius of the circular inspection trajectory, and recalculate the angular velocity until it is within the allowable range.

[0065] S323. Based on the surface tangential linear velocity at the geometric center of the boundary contour and the reference circular flight velocity, a vector synthesis analysis is performed to obtain the circular flight velocity of the UAV at the directional inspection position.

[0066] S33. Control the UAV to fly at a circular flight speed and collect surface image sequences covering different directional inspection locations.

[0067] This invention acquires images of fan blades by a drone flying along an inspection route, determines whether there are abnormal areas in the images and identifies their boundary contours, and analyzes the drone's circular flight speed at the directional inspection position by combining the real-time rotation speed of the fan blades. This enables directional and refined inspection of defective areas of the fan blades, and ensures that the drone remains relatively stationary with the fan blade surface through circular flight, thus guaranteeing the quality of image acquisition of the fan blade surface.

[0068] S4. Select surface images with acceptable clarity and determine the health status of the fan blades.

[0069] Considering that some images in the surface image sequence are blurred due to shooting shake and lighting environment interference, such images will affect the accuracy of defect feature identification. Moreover, different types and degrees of defects have different effects on the performance of the fan blade structure. Therefore, it is necessary to first perform clear quantitative analysis on the images and screen qualified images, then extract defect feature data and combine it with the fan blade performance degradation law to determine the health status, so as to improve the accuracy of defect detection and the rationality of maintenance decisions.

[0070] Based on this, the specific implementation of the present invention includes: S41, performing sharpness analysis on the surface image sequence and selecting surface images with qualified sharpness. The specific selection method is as follows: S411, performing grayscale processing on each frame of the surface image sequence to obtain the gradient magnitude of each pixel in the image, and establishing a gradient magnitude distribution matrix.

[0071] S412. Statistical feature extraction is performed on the gradient magnitude distribution matrix to obtain statistical feature indicators of gradient magnitude distribution, including the mean and variance of gradient magnitude and the proportion of pixels with gradient magnitude exceeding the mean.

[0072] S413. Based on the set of clear surface images of the fan blades inspected by the UAV, a preset clarity index threshold is set. The preset clarity index threshold is the smallest gradient amplitude distribution statistical feature index in the set of clear surface images of the fan blades collected by the UAV under different angles and relative motion states. Surface images whose gradient amplitude distribution statistical feature index is greater than the corresponding preset clarity index threshold are selected as surface images with qualified clarity.

[0073] It should be noted that this invention achieves quantitative screening of image clarity through three statistical features: the mean, variance, and pixel percentage of gradient amplitude. This eliminates blurry images caused by shooting shake and lighting interference, providing high-quality image data for subsequent defect feature extraction. This solves the problem of low defect detection accuracy and missed detection of minor defects caused by poor image quality.

[0074] S42. Identify the feature data of defect areas in the surface image to determine the health status of the fan blades. The specific determination method is as follows: S421. Perform image recognition on the surface image with acceptable clarity, segment the defect areas in the surface image, extract the contour size, texture features and grayscale features of the defect areas, and identify the type of defect areas and the corresponding feature dataset.

[0075] It should be noted that defect region type identification can be achieved by matching the texture and grayscale features of the defect region with the texture and grayscale features corresponding to various preset fan blade defect images. Cosine similarity calculation can be used to obtain the similarity score, which is already a known technique and will not be elaborated upon further. The fan blade defect region type with the highest similarity score is then selected.

[0076] Various fan blade defects include, but are not limited to, crack defects, corrosion defects, and coating peeling defects. The feature dataset for crack defects includes the crack length to depth ratio, the feature dataset for corrosion defects includes the corrosion area, and the feature dataset for coating peeling defects includes the peeling thickness and the coating peeling area.

[0077] S422. Based on the material type and service life of the wind turbine blades, retrieve the characteristic threshold values ​​of the performance degradation of the corresponding blade samples under various defects from the blade structure performance degradation database.

[0078] S423. If any feature data in the feature dataset corresponding to the defective area exceeds the corresponding feature threshold value, the fan blade health status is determined to be a defective state; otherwise, the fan blade health status is determined to be a normal state.

[0079] It should be noted that the blade structure performance degradation database is used to store structural performance degradation data of wind turbine blades of different material types and different service years under long-term operating conditions. This database was established by conducting multiple rounds of structural strength performance tests and defect verification on blade samples of the same model, under the same operating conditions and with the same service time, and includes the correspondence between various defects and structural bearing capacity.

[0080] In the specific implementation process, firstly, based on the material type and service life of the wind turbine blades, corresponding historical sample data are selected from the blade structural performance degradation database. Then, the characteristic critical parameters corresponding to the gradual degradation of the structural performance from the normal state to the critical failure state under the action of different defects such as crack defects, corrosion defects, and coating peeling defects are retrieved. The critical parameter is set as the characteristic threshold value of the corresponding defect. This characteristic threshold value is used to characterize the upper limit of defect development allowed by the blade under the premise of ensuring safe and stable operation, and provides a quantitative judgment basis for the subsequent determination of the blade health status.

[0081] This invention performs clarity analysis on a sequence of surface images, selects surface images with acceptable clarity, and achieves quantitative evaluation of image clarity. This provides data support for defect feature extraction and identification, identifies defect area feature data in surface images, determines the health status of the fan blade, and evaluates the impact of defect feature data on the structural performance of the fan blade. It effectively distinguishes between surface defects and structural defects, improving the accuracy of defect judgment and the rationality of maintenance decisions.

[0082] S5. Analyze the overall balance and stability of the wind turbine generator set.

[0083] Considering that the operational stability of wind turbine generators depends on the mass balance of each blade, although a small local defect in a single blade may not reach the defect state, the superposition of defects in multiple blades may cause the unit's center of gravity to shift and become unbalanced, leading to problems such as increased vibration and accelerated component wear. Current detection only focuses on the defect state of a single blade and lacks overall unit balance analysis. It is necessary to analyze the overall balance stability of the unit from the perspective of mass eccentricity based on the blade assembly parameters to achieve comprehensive safety control of wind turbine generators.

[0084] Based on this, when all blades are in normal health condition, the present invention analyzes the overall balance and stability of the wind turbine generator set based on the blade assembly parameters of the wind turbine generator set.

[0085] like Figure 3 As shown, the specific implementation of the present invention includes: S51, extracting the contour dimensions from the feature data of the defect areas corresponding to all blades of the wind turbine, and calculating the defect mass loss of each blade in combination with the corresponding density of the blade material.

[0086] The defect mass loss is the product of the estimated defect volume in the profile dimensions and the corresponding density of the blade material. If there are multiple defect areas in the same blade, the mass loss of each defect area is summed to obtain the total defect mass loss of the blade.

[0087] S52. Based on the blade rotation axis in the blade assembly parameters of the wind turbine generator set, determine the vertical distance from the center of the blade defect area to the blade rotation axis, multiply it with the defect mass loss to obtain the mass eccentricity of the defect area, where the mass eccentricity reflects the degree of mass eccentricity of a single blade due to the defect.

[0088] S53. Based on the installation angle of each fan blade in the fan blade assembly parameters, the installation angle is the angle between the fan blade and the positive X-axis direction. The radial component of the mass eccentricity moment of each fan blade defect area along the X-axis direction and the tangential component along the Y-axis direction of the unit coordinate system are transformed by coordinate rotation. The radial and tangential components of all fan blades after transformation are summed to obtain the radial offset and tangential offset of the unit's center of mass.

[0089] It should be noted that the coordinate system of the unit is established with the center of the wind turbine hub as the origin, the central axis of the tower as the Z-axis, and the prevailing wind direction as the positive X-axis.

[0090] S54. Perform a vector synthesis operation on the radial offset of the unit's center of mass and the tangential offset of the unit's center of mass to obtain the overall center of mass offset of the unit.

[0091] S55. Compare the overall center-of-gravity offset of the unit with the preset stability threshold to determine the overall balance stability of the wind turbine generator set. The specific determination method is as follows: if the overall center-of-gravity offset of the unit is less than or equal to the preset stability threshold, the overall balance stability of the wind turbine generator set is deemed qualified, and the unit can operate safely and stably.

[0092] If the overall centroid offset of the unit exceeds the preset stability threshold, the overall balance stability of the unit is deemed unqualified, and there is an operational risk caused by the centroid offset.

[0093] As an example, the preset stability threshold can be the maximum value of the overall centroid offset of the wind turbine generator in each historical overall imbalance event within a set historical period. Implementers can also set the historical period to determine the preset stability threshold themselves.

[0094] This invention analyzes the overall balance and stability of a wind turbine generator set based on blade assembly parameters when all blades are in normal health conditions. This enables comprehensive control over the overall balance and stability of the generator set, timely identification of potential risks of generator set imbalance caused by blade defects, avoidance of problems such as increased vibration during generator set operation and accelerated component wear, and extension of the overall service life of the wind turbine generator set.

[0095] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0096] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0097] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0098] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0099] Finally, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for detecting defects in power equipment based on unmanned aerial vehicle (UAV) inspection, characterized in that, include: Based on the wind turbine blade size data, the radial detection distance between the UAV and the blade is determined. Combined with the real-time rotation speed of the blade, the inspection flight speed of the UAV is obtained, and the corresponding inspection route of the UAV is planned. Collect images of fan blades by a drone flying along the inspection route, determine whether there are abnormal areas in the fan blade inspection images, and determine the boundary contours of the abnormal areas. The location of the UAV directional inspection is determined based on the boundary contour of the abnormal region. The circular flight speed of the UAV at the directional inspection location is analyzed in combination with the real-time rotation speed of the fan blades. Surface image sequences covering different directional inspection locations are collected. Analyze the clarity of the surface image sequence, select surface images with acceptable clarity, identify the feature data of defect areas in the surface images, and determine the health status of the fan blades. When all blades are in normal health condition, the overall balance and stability of the wind turbine generator set are analyzed based on the blade assembly parameters.

2. The method for detecting defects in power equipment based on unmanned aerial vehicle (UAV) inspection according to claim 1, characterized in that: The steps for planning the inspection route of the UAV are as follows: Extract the blade boundary contour from the wind turbine blade size data, and determine the radial detection distance between the UAV and the blade based on the reference shooting range diameter corresponding to different shooting distances of the UAV. Based on the reference shooting range of the UAV at the corresponding radial detection distance, the fan blade boundary contour is divided into different boundary contour regions; A three-dimensional spatial coordinate system is established with the center of the wind turbine hub as the origin. Based on the real-time rotational speed of the current blades, the time sequence of the center position coordinates of different boundary contour regions is determined. Obtain the unit time displacement corresponding to the center position of the same boundary contour region and the unit time interval displacement corresponding to the center position of adjacent boundary contour regions, and calculate the drone inspection direction deflection angle and drone inspection flight speed. By combining the radial detection distance between the UAV and the fan blades, a sequence of inspection path nodes is established according to the rotation plane direction of the fan blades, and these nodes are connected sequentially to form the UAV inspection route.

3. The method for detecting defects in power equipment based on unmanned aerial vehicle (UAV) inspection according to claim 2, characterized in that: The method for determining the radial detection distance between the UAV and the fan blades is as follows: Filter the maximum profile width of the fan blade boundary in the fan blade boundary profile, and match the shooting distance corresponding to the maximum profile width from the reference shooting range diameter corresponding to different shooting distances of the drone; If the maximum profile width corresponds to a shooting distance greater than the safe shooting distance of the wind turbine blades, then that shooting distance is used as the radial detection distance between the drone and the blades. Conversely, the safe shooting distance of the wind turbine blades is used as the radial detection distance between the drone and the blades.

4. The method for detecting defects in power equipment based on unmanned aerial vehicle (UAV) inspection according to claim 1, characterized in that: The boundary contour of the abnormal region is determined as follows: The image set of fan blade inspection collected by the UAV along the inspection route is stitched together to obtain the overall image of the fan blade. Compare the grayscale values ​​of the whole image of the fan blade with the standard defect-free image of the fan blade, calculate the grayscale deviation value of the corresponding pixel in the whole image of the fan blade and the standard defect-free image, and mark the pixel as a grayscale abnormal point if the grayscale deviation value of a certain pixel exceeds the preset grayscale deviation allowable range. All gray-level anomalies are connected to adjacent regions. If a gray-level anomaly connected region exists, it is determined that there is an anomaly region in the fan blade inspection image. Edge detection is performed on the anomaly region, and the boundary contour of the anomaly region is fitted.

5. The method for detecting defects in power equipment based on unmanned aerial vehicle (UAV) inspection according to claim 1, characterized in that: The method for determining the directional inspection location of the drone is as follows: Based on the boundary contour of the abnormal area, and combined with the reference shooting range diameter corresponding to different shooting distances of the UAV, the radial distance of the UAV for directional inspection is determined. Extract the geometric center coordinates of the boundary contour of the abnormal area. Using these geometric center coordinates as a reference, and combining them with the radial distance of the UAV's directional inspection, determine the reference position for the UAV's directional inspection. Extend a preset distance from the reference position to the edge of the boundary contour of the abnormal area to determine multiple auxiliary inspection positions, and combine the reference position and the multiple auxiliary inspection positions to form the directional inspection position of the UAV.

6. A method for detecting defects in power equipment based on unmanned aerial vehicle (UAV) inspection according to claim 5, characterized in that: The method for analyzing the circular flight speed of the UAV at the directional inspection location is as follows: The fan blades are divided into multiple segments radially according to different rotation radii. Based on the real-time rotation speed of the fan blades and the rotation radius of each segment, the surface tangential linear velocity of each segment is obtained, and the surface tangential linear velocity at the geometric center of the boundary contour is extracted. Based on the radius of the circular inspection trajectory of the UAV at the directional inspection location, calculate the benchmark circular flight speed required for the UAV to maintain relatively stationary observation. By performing vector synthesis analysis on the surface tangential linear velocity at the geometric center of the boundary contour and the reference circular flight velocity, the circular flight velocity of the UAV at the directional inspection position is obtained.

7. A method for detecting defects in power equipment based on unmanned aerial vehicle (UAV) inspection according to claim 6, characterized in that: The calculation method for the baseline circular flight speed required for the UAV to maintain relatively stationary observation is as follows: Obtain the horizontal projection distance of the UAV's directional inspection position relative to the center of the wind turbine hub, and use this horizontal projection distance as the radius of the UAV's circular inspection trajectory. By analyzing the ratio of the surface tangential linear velocity at the geometric center of the boundary profile to the radius of the UAV's circular inspection trajectory, the angular velocity required for the UAV to maintain relatively stationary observation is obtained. The required angular velocity is compared with the allowable range of safe flight angular velocity for the UAV to determine the reference circular flight angular velocity. This reference circular flight angular velocity is then multiplied by the radius of the UAV's circular inspection trajectory to obtain the vector magnitude of the reference circular flight velocity. The reference circular flight speed is determined by taking the tangent direction of the circular inspection trajectory as the reference direction of the circular flight speed and combining it with the vector magnitude.

8. The method for detecting defects in power equipment based on unmanned aerial vehicle (UAV) inspection according to claim 1, characterized in that: The method for selecting surface images with acceptable clarity is as follows: Grayscale processing is performed on each frame of the surface image sequence to obtain the gradient magnitude of each pixel in the image, and a gradient magnitude distribution matrix is ​​established. Statistical features are extracted from the gradient magnitude distribution matrix to obtain statistical feature indicators of gradient magnitude distribution, including the mean, variance, and percentage of pixels with gradient magnitudes exceeding the mean. Based on the set of clear surface images corresponding to the fan blades inspected by UAVs, a preset clarity index threshold is set, and surface images in which the gradient amplitude distribution statistical feature index is greater than the corresponding preset clarity index threshold are selected as surface images with qualified clarity.

9. A method for detecting defects in power equipment based on unmanned aerial vehicle (UAV) inspection according to claim 8, characterized in that: The method for determining the health status of the fan blades is as follows: Image recognition is performed on surface images with acceptable clarity to segment defect areas in the surface images, extract the contour size, texture features and grayscale features of the defect areas, and identify the defect area type and corresponding feature dataset. Based on the material type and service life of wind turbine blades, characteristic threshold values ​​of performance degradation of corresponding blade samples under various defects are retrieved from the blade structure performance degradation database. If any feature data in the feature dataset corresponding to the defective region exceeds the corresponding feature threshold value, the fan blade health status is determined to be defective; otherwise, the fan blade health status is determined to be normal.

10. A method for detecting defects in power equipment based on unmanned aerial vehicle (UAV) inspection according to claim 9, characterized in that: The analysis of the overall balance stability of the wind turbine generator set includes the following details: Extract the contour dimensions from the feature data of the defect areas of all blades of the wind turbine, and calculate the defect mass loss of each blade by combining the corresponding density of the blade material. Based on the blade rotation axis in the blade assembly parameters of the wind turbine generator set, the vertical distance from the center of the blade defect area to the blade rotation axis is determined, and the distance is multiplied with the defect mass loss to obtain the mass eccentricity of the defect area. Based on the installation angle of each blade in the blade assembly parameters, the radial and tangential components of the mass eccentricity of each blade defect area are subjected to coordinate rotation transformation. The radial and tangential components of all blades after transformation are summed to obtain the radial and tangential offset of the unit's center of mass. The radial offset of the unit's center of mass and the tangential offset of the unit's center of mass are vectorized to obtain the overall center of mass offset of the unit. The overall balance and stability of the wind turbine generator set are determined by comparing the overall centroid offset of the unit with a preset stability threshold.

Images