Method for inspecting state of wind farm grid-connected test equipment by fusing unmanned aerial vehicle vision

Abstract

The present application relates to the field of image processing, in particular to a kind of wind farm involved network test equipment state inspection method of fusing unmanned aerial vehicle vision.The present application utilizes unmanned aerial vehicle to collect equipment box image, carries out pre-processing to obtain gray image in view of the problems of high risk, low efficiency and strong subjectivity in the prior art manual inspection;Suspected equipment box is located by edge detection and connected domain analysis;Abnormal pixel points are detected based on pixel gray difference, and suspected damage area is obtained;The first probability parameter of shape regularity and the second probability parameter of texture uniformity are calculated;Damage probability is obtained by combining the two, to judge whether the equipment is damaged.The present application realizes automatic, non-contact inspection, improves efficiency and accuracy, reduces manual risk and cost, and provides reliable technical support for wind farm equipment maintenance.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of image processing, in particular to a wind farm grid-connected test equipment state inspection method fusing unmanned aerial vehicle vision. BACKGROUND

[0002] Before the power generated by the wind farm is connected to the power grid, it must be strictly detected by the grid-connected test equipment. Only when it is confirmed that all indicators meet the safety standards, the grid-connected operation can be completed. This link is a key barrier to ensure the safe and stable operation of the power grid. Therefore, whether the grid-connected test equipment can maintain normal operation state is directly related to whether the power generated by the wind farm can be smoothly connected to the power grid. In order to ensure that the grid-connected test equipment of the wind farm is always in good operating condition, it is particularly important to carry out systematic and standardized daily inspection work on these equipment.

[0003] These grid-connected test equipment are widely distributed in various areas of the wind farm, and their installation positions and environments vary greatly. When conducting manual inspection, for those equipment installed in low places or indoors, the inspection work is relatively simple and can be quickly completed because it is easy to access and observe. However, for those equipment installed in high places and scattered, the inspection personnel need to use special climbing tools to reach the inspection position. Not only does the whole process take time and effort, greatly increasing the intensity and cost of the inspection work, but the judgment result is too dependent on the subjective judgment of the workers. Moreover, when working at high altitudes, there are many safety risks such as falling and slipping, which pose a threat to the personal safety of the inspection personnel.

[0004] Based on the many problems of manual inspection when facing high and scattered equipment, the present application proposes a wind farm grid-connected test equipment state inspection method fusing unmanned aerial vehicle vision. Through the unmanned aerial vehicle, the equipment can be quickly reached at high altitude, and the image of the equipment can be collected and the defects can be detected and identified. This way not only can significantly shorten the time of daily inspection and improve the efficiency of inspection work, but also can avoid the inspection personnel working directly at high altitudes, fundamentally reducing the safety risks of manual inspection, and providing more reliable protection for the stable operation of the grid-connected test equipment of the wind farm. SUMMARY

[0005] The present application provides a wind farm grid-connected test equipment state inspection method fusing unmanned aerial vehicle vision to solve the existing problems: the existing grid-connected test equipment of the wind farm is mostly directly detected by personnel, which has high safety risks, high work intensity and cost, and the detection result is more subjective and not accurate enough.

[0006] The wind farm grid-connected test equipment state inspection method fusing unmanned aerial vehicle vision of the present application adopts the following technical scheme:

[0007] The application provides a wind farm grid-connected test equipment state inspection method fusing unmanned aerial vehicle vision.

[0008] An image of the equipment box is collected by using the unmanned aerial vehicle, the image of the equipment box is preprocessed, and a gray-scale image is obtained.

[0009] Edge detection is performed on the gray-scale image, a plurality of connected domains are obtained according to the edge detection result, and a suspected equipment box connected domain is determined according to the ratio of the contour perimeter to the area of the connected domain.

[0010] Abnormal pixel points are obtained according to the difference between the gray-scale values of the non-edge pixel points in the suspected equipment box connected domain and the pixel points above and to the right thereof, and a suspected damage area connected domain is obtained according to the position information of the abnormal pixel points.

[0011] Taking the geometric center of each suspected damage area connected domain as the origin, straight lines passing through the origin are drawn in K different directions, two line segments are formed by the intersection of each straight line and the edge of the connected domain, the average value of the length ratio of the two line segments in all directions is calculated as the probability that each suspected damage area connected domain is a regular shape, and a first probability parameter of each suspected damage area connected domain is obtained according to the difference between the average value and 1.

[0012] Taking the geometric center of each suspected damage area connected domain as the center, circles are drawn with a fixed pixel interval E, the radius is gradually increased until the circumference exceeds the boundary of the connected domain, and all circles that do not exceed the boundary are retained, the second probability parameter of each suspected damage area connected domain is obtained by calculating the difference between the average gray-scale values of the pixel points covered by adjacent two circles.

[0013] The first probability parameter and the second probability parameter of each suspected damage area connected domain are calculated, the damage probability of each suspected damage area connected domain is obtained, and whether the wind farm grid-connected test equipment box is damaged is determined according to the damage probability of each suspected damage area connected domain.

[0014] Further, the method for collecting the image of the equipment box by using the unmanned aerial vehicle, preprocessing the image of the equipment box and obtaining the gray-scale image comprises the following specific method:

[0015] An image of the wind farm grid-connected test equipment box at the same position is photographed by using the unmanned aerial vehicle, the dispersion degree of the gray-scale value of each image pixel is calculated, and the image with the largest dispersion degree is selected as a test image, and the test image is subjected to gray-scale processing to obtain a gray-scale image.

[0016] Further, the method for collecting the image of the equipment box by using the unmanned aerial vehicle, preprocessing the image of the equipment box and obtaining the gray-scale image comprises the following specific method:

[0017] The Canny operator is used to detect edges of the image to obtain a binary image with clear edges, obtain p connected domains with closed edges, and calculate the regularity degree of the mth connected domain by the following specific method:

[0018]

[0019] In the formula, The regularity degree of the mth connected domain is represented by The perimeter of the mth connected domain is represented by The number of pixel points in the mth connected domain is represented by

[0020] The threshold value is set The absolute value of the difference is determined If the absolute value of the difference is less than or equal to The mth connected domain is recorded as a suspected equipment cabinet connected domain.

[0021] Further, the abnormal pixel points are obtained according to the difference between the gray value of the non-edge pixel point in the suspected equipment cabinet connected domain and the gray value of the pixel point above and to the right thereof, and the suspected damage area connected domain is obtained according to the position information of the abnormal pixel points, including the following specific method:

[0022] Starting from the top-left pixel point of the suspected equipment cabinet connected domain, the absolute values of the differences between the gray values of the current pixel and the pixel points to the right and below thereof are calculated for each non-connected domain edge pixel point, and the absolute value of the difference between the two absolute values is calculated as the gray relative change intensity of each non-connected domain edge pixel point. The mean value of the gray relative change intensities of all non-connected domain edge pixel points is taken as the mutation judgment threshold value. The pixel points with a gray relative change intensity exceeding the mutation judgment threshold value are recorded as abnormal pixel points. The abnormal pixel region capable of forming a closed region or constituting a closed region together with the connected domain edge, the pixel set capable of self-closing or connecting with the original edge of the suspected equipment cabinet connected domain to form a closed region, and the region surrounded by the closed boundary are taken as the suspected damage area connected domain.

[0023] Further, the geometric center of each suspected damage area connected domain is taken as the origin, a straight line is drawn through the origin in K different directions, and the straight line intersects the connected domain edge to form two line segments. The average value of the length ratios of the two line segments in all directions is taken as the probability that each suspected damage area connected domain is a regular shape, and the first probability parameter of each suspected damage area connected domain is obtained according to the difference between the average value and 1, including the following specific method:

[0024] The geometric center of each suspected damage area connected domain is obtained, a rectangular coordinate system is constructed with the geometric center as the origin, a horizontal axis is made through the origin, and a vertical axis is made through the origin, starting from the horizontal direction, a straight line is made through the origin every K degrees, and the two points of intersection with the edge of the suspected damage area are calculated, the distances of the two points to the origin are calculated, and the average value of the ratio of the two distances is calculated as follows:

[0025]

[0026] In the formula, The probability that the nth suspected damage area connected domain in the mth suspected equipment cabinet connected domain is a regular shape is represented by Pm(n), the coordinate system origin coordinate of the nth suspected damage area connected domain in the mth suspected equipment cabinet connected domain is represented by O, and K represents the angular interval. The straight line through the origin of the nth suspected damage area connected domain in the mth suspected equipment cabinet connected domain is represented by Lm(n), and the two intersection points of the straight line with the nth suspected damage area connected domain in the mth suspected equipment cabinet connected domain are represented by A and B. The distance from the origin to A is less than or equal to K, and the distance from the origin to B is greater than K. The horizontal coordinate and vertical coordinate values of A are represented by X1 and Y1, respectively. The horizontal coordinate and vertical coordinate values of B are represented by X2 and Y2, respectively. The horizontal coordinate and vertical coordinate values of A are represented by X1 and Y1, respectively. The horizontal coordinate and vertical coordinate values of B are represented by X2 and Y2, respectively. The horizontal coordinate and vertical coordinate values of A are represented by X1 and Y1, respectively. The horizontal coordinate and vertical coordinate values of B are represented by X2 and Y2, respectively. The horizontal coordinate and vertical coordinate values of A are represented by X1 and Y1, respectively. The horizontal coordinate and vertical coordinate values of B are represented by X2 and Y2, respectively. The horizontal coordinate and vertical coordinate values of A are represented by X1 and Y1, respectively. The horizontal coordinate and vertical coordinate values of B are represented by X2 and Y2, respectively. The first probability parameter of the nth suspected damage area connected domain in the mth suspected equipment cabinet connected domain is represented by 1-

[0027] Further, the geometric center of each suspected damage area connected domain is taken as the center, and a circle is made with a fixed pixel interval E and a gradually increasing radius, until the circumference exceeds the boundary of the connected domain, and all circles that do not exceed the boundary are retained.

[0028] The geometric center of the nth suspected damage area connected domain in the mth suspected equipment cabinet connected domain is taken as the center, and a circle is made with a fixed pixel interval E and a gradually increasing radius, until the circumference exceeds the boundary of the connected domain, and all circles that do not exceed the boundary are retained.

[0029] The second probability parameter of the nth suspected damage area connected domain in the mth suspected equipment cabinet connected domain is calculated as follows:​​

[0030]

[0031] wherein, represents the difference between the average gray value of the pixel point where the fth circle in the nth suspected damage area connected domain in the mth suspected equipment cabinet connected domain and the pixel point where the (f-1)th circle, represents the number of pixel points where the fth circle in the nth suspected damage area connected domain in the mth suspected equipment cabinet connected domain, represents the gray value of the ith pixel point of the fth circle in the nth suspected damage area connected domain in the mth suspected equipment cabinet connected domain, represents the number of pixel points where the (f-1)th circle in the nth suspected damage area connected domain in the mth suspected equipment cabinet connected domain, represents the gray value of the jth pixel point of the (f-1)th circle in the nth suspected damage area connected domain in the mth suspected equipment cabinet connected domain, represents the gray value of the lth pixel point of the (f-1)th circle in the nth suspected damage area connected domain in the mth suspected equipment cabinet connected domain, represents the second probability parameter of the nth suspected damage area connected domain in the mth suspected equipment cabinet connected domain.

[0032] Further, the first probability parameter and the second probability parameter of each suspected damage area connected domain are calculated, the damage probability of each suspected damage area connected domain is obtained, and whether the equipment cabinet of the wind power plant involved in the test is damaged is determined according to the damage probability of each suspected damage area connected domain, and the specific method comprises:

[0033]

[0034] wherein, represents the first probability parameter of the nth suspected damage area connected domain in the mth suspected equipment cabinet connected domain, represents the second probability parameter of the nth suspected damage area connected domain in the mth suspected equipment cabinet connected domain, represents the probability of damage of the nth suspected damage area connected domain in the mth suspected equipment cabinet connected domain, and a threshold value Q is set. When the damage probability of any damage connected domain in any suspected equipment cabinet connected domain in the image is greater than or equal to Q, the equipment cabinet is damaged, and relevant personnel are reminded to maintain.

[0035] ​​The second aspect of the application provides a wind farm network test equipment state inspection system integrating unmanned aerial vehicle vision, which comprises an image acquisition and processing module, an equipment detection module, a suspected damage area detection module, a first probability parameter calculation module, a second probability calculation module, and an equipment damage judgment module.

[0036] The image acquisition and processing module is used for collecting equipment box images by using an unmanned aerial vehicle, pre-processing the equipment box images, and obtaining a gray image.

[0037] The equipment detection module is used for obtaining abnormal pixel points according to the gray value difference between non-edge pixel points and pixel points above and to the right of the non-edge pixel points in the suspected equipment box connected domain, and obtaining a suspected damage area connected domain according to the position information of the abnormal pixel points.

[0038] The suspected damage area detection module is used for obtaining abnormal pixel points according to the gray value difference between non-edge pixel points and pixel points above and to the right of the non-edge pixel points in the suspected equipment box connected domain, and obtaining a suspected damage area connected domain according to the position information of the abnormal pixel points.

[0039] The first probability parameter calculation module is used for taking the geometric center of each suspected damage area connected domain as the origin, making straight lines passing through the origin in K different directions, forming two line segments by the intersection of each straight line and the edge of the connected domain, calculating the average value of the length ratio of the two line segments in all directions as the probability that each suspected damage area connected domain is a regular shape, and obtaining a first probability parameter of each suspected damage area connected domain according to the difference between the average value and 1.

[0040] The second probability parameter calculation module is used for taking the geometric center of each suspected damage area connected domain as the center of a circle, making circles with a fixed pixel interval E, and increasing the radius step by step until the circumference exceeds the boundary of the connected domain, retaining all circles that do not exceed the boundary, and obtaining a second probability parameter of each suspected damage area connected domain by calculating the difference between the average gray values of the pixel points covered by adjacent two circles.

[0041] The equipment damage judgment module is used for calculating the first probability parameter and the second probability parameter of each suspected damage area connected domain, obtaining the damage probability of each suspected damage area connected domain, and judging whether the wind farm network test equipment box is damaged according to the damage probability of each suspected damage area connected domain.

[0042] The third aspect of the application provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to realize the steps of the above-mentioned wind farm network test equipment state inspection method integrating unmanned aerial vehicle vision.

[0043] In a fourth aspect, the application provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method for inspecting the state of wind farm grid-connected test equipment by fusing unmanned aerial vehicle vision when executing the computer program.

[0044] The technical scheme of the application has the beneficial effects that:

[0045] The unmanned aerial vehicle vision technology is used to realize automatic inspection, thereby avoiding the risk of manual high-altitude operation, improving the inspection efficiency and accuracy, and ensuring the objectivity and reliability of damage detection through the multi-step image processing algorithm;

[0046] The image with the largest discrete degree is selected as the test image, so as to ensure the image definition and provide high-quality input for subsequent processing, thereby improving the overall detection accuracy;

[0047] The connected domain rule degree is used to screen the suspected equipment box, so as to effectively exclude natural background interference, accurately locate the target area, and reduce false detection;

[0048] The abnormal pixel points are adaptively detected through the relative change intensity of gray scale, so as to adapt to different light conditions, accurately identify the damage area, and reduce the probability of misjudgment;

[0049] Based on the multi-direction shape symmetry evaluation, the region rule degree is quantified, so as to effectively distinguish the real damage from the regular components and improve the damage identification specificity;

[0050] Through the radial gray scale change analysis, the texture anomaly is captured, the detection capability for microscopic damage is enhanced, and the detection comprehensiveness is improved;

[0051] The damage probability is calculated based on the shape and texture features, multi-dimensional judgment is realized, the false positive rate is significantly reduced, and the scientific nature of the maintenance decision is ensured. BRIEF DESCRIPTION OF DRAWINGS

[0052] In order to more clearly illustrate the technical scheme in the embodiments of the application or the prior art, the drawings needed to be used in the embodiment or prior art description will be briefly introduced. Obviously, the drawings in the following description only some embodiments of the application, and for those skilled in the art, other drawings can be obtained without creative labor on the basis of these drawings.

[0053] Figure 1 The flowchart of the steps of the method for inspecting the state of wind farm grid-connected test equipment by fusing unmanned aerial vehicle vision;

[0054] Figure 2 The structural block diagram of the system for inspecting the state of wind farm grid-connected test equipment by fusing unmanned aerial vehicle vision. DETAILED DESCRIPTION

[0055] To further clarify the technical means and effects taken by the present application to achieve the predetermined object of the application, the specific implementation, structure, features and effects of a wind farm grid-connected test equipment state inspection method integrating unmanned aerial vehicle vision according to the present application are described in detail as follows. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0056] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs.

[0057] The specific scheme of the wind farm grid-connected test equipment state inspection method integrating unmanned aerial vehicle vision provided by the present application is described in detail below with reference to the accompanying drawings.

[0058] Please refer to Figure 1 which shows the first object of the present application, a step flow chart of a wind farm grid-connected test equipment state inspection method integrating unmanned aerial vehicle vision, which comprises the following steps:

[0059] Step S001: Collecting equipment box images using an unmanned aerial vehicle, pre-processing the equipment box images, and obtaining a gray-scale image.

[0060] Traditional wind farm grid-connected test equipment inspection mainly relies on manual on-site inspection, which is low in efficiency, high in cost and prone to missed inspection. The present application uses unmanned aerial vehicle vision technology to achieve automatic and non-contact inspection. Through image preprocessing, edge detection and connected component analysis, the equipment box is preliminarily located; then based on pixel gray difference, abnormal areas are identified, and shape regularity and gray change probability parameters are introduced to quantitatively evaluate damage risk, solving the problem of strong subjectivity and insufficient precision of traditional methods.

[0061] Specifically, the equipment box images are collected using an unmanned aerial vehicle, the equipment box images are pre-processed to obtain a gray-scale image, and the specific method is as follows:

[0062] S wind farm grid-connected test equipment box images at the same position are taken by the unmanned aerial vehicle, the dispersion degree of the pixel gray value of each image is calculated, and the image with the largest dispersion degree is selected as the test image. The test image is processed to obtain a gray-scale image.

[0063] It should be noted that the image collection range of the unmanned aerial vehicle is wide, the data real-time performance is strong, the computer vision algorithm is combined to automatically identify the suspected damage area, the inspection efficiency and accuracy are greatly improved. Through the double probability parameter fusion judgment, the false alarm rate is reduced, the reliability of the damage detection is ensured, data support is provided for the wind power plant maintenance decision, the downtime loss is reduced, the number of photographed images S=10 in the embodiment, and the number of S is not specifically limited by the application.

[0064] Step S002: edge detection is performed on the gray image, a plurality of connected domains are obtained according to the edge detection result, and a suspected equipment box connected domain is determined according to the ratio of the contour perimeter to the area of the connected domain.

[0065] It should be noted that, in order to avoid the inherent defects of high-altitude operation risk, low detection efficiency and strong subjectivity of manual detection, based on the equipment image collected in step one, image recognition technology is introduced in this step to automatically analyze the obtained image data, and efficient and objective box state evaluation is realized.

[0066] It should be further noted that, due to the body shaking and automatic focusing delay of the unmanned aerial vehicle during flight hovering, the image obtained by single shooting is prone to be blurred, which leads to the failure of the subsequent damage recognition algorithm. Therefore, in this step, the clearest image is selected from the photographed images for subsequent recognition.

[0067] Specifically, edge detection is performed on the gray image, a plurality of connected domains are obtained according to the edge detection result, and a suspected equipment box connected domain is determined according to the ratio of the contour perimeter to the area of the connected domain, and the specific method is as follows:

[0068] The Canny operator is used to perform edge detection on the image to obtain a binary image containing clear edges, obtain p connected domains with closed edges, and calculate the regularity degree of the mth connected domain. The specific method is as follows:

[0069]

[0070] In the formula, the regularity degree of the mth connected domain is represented, the perimeter of the mth connected domain is represented, the number of pixel points in the mth connected domain is represented;

[0071] The threshold value is set , the absolute value of the difference between and 1 is judged, if the absolute value of the difference is less than or equal to , the mth connected domain is recorded as a suspected equipment box connected domain.

[0072] It should be noted that the connected domain rule degree essentially measures the deviation of the shape from the standard circle, and the closer the value is to 1, the closer the contour is to a circle. Since the wind farm equipment box is usually a regular rectangle, its value will be stably greater than 1 and within a certain range, while the contours of natural objects such as leaves and clouds in the background are extremely irregular, which can cause value to be much greater than 1 or fluctuate sharply. Therefore, by setting a threshold to filter out connected domains with a difference in absolute value less than or equal to , the regular man-made box can be efficiently and accurately separated from the complex background. This not only significantly reduces the computational load of subsequent image processing, but more importantly, effectively eliminates the interference of a large number of naturally irregular contours, laying a reliable foundation for subsequent accurate identification of box damage and improving the robustness and accuracy of the entire inspection system. The present application does not specifically limit the threshold , and in the present embodiment, the value of is 0.5.

[0073] Step S003: Obtain abnormal pixel points according to the difference in gray value between the non-edge pixel points in the suspected equipment box connected domain and the pixel points above and to the right thereof, and obtain a suspected damage area connected domain according to the position information of the abnormal pixel points.

[0074] It should be noted that since the damaged area usually does not have the uniform reflection characteristics of the normal equipment cabinet surface, changes in its shape and structure often result in a significant drop in local gray value. Based on this feature, by detecting areas in the image where the gray value changes sharply, the suspected damage area is initially determined as a candidate target for subsequent identification and analysis.

[0075] It should be further noted that conventional edge detection or global threshold segmentation methods face challenges when processing surface damage of equipment boxes: on the one hand, the structural edges of the box itself can be misdetected as damage; on the other hand, due to uneven lighting, the same damage may exhibit different gray values in different areas, making fixed thresholds less adaptable. To solve this problem, the present method discards the method of only focusing on the absolute gray value of a single pixel, and analyzes the relative gray change pattern of each pixel and its neighborhood. A normal, uniformly textured box surface should have smooth and continuous gray changes; when the surface has cracks, depressions, or damage such as stains, the gray value of the point will be discordant and conspicuous with the right and lower adjacent pixel points. By calculating the difference between these two-way gray differences, the difference between the damage point and the normal area in local texture consistency can be amplified, thereby more accurately locking the abnormal pixel.

[0076] Specifically, the abnormal pixel points are obtained according to the difference between the gray value of a non-edge pixel point in the suspected device box connected domain and the gray value of the pixel point right and below the non-edge pixel point, and the suspected damage area connected domain is obtained according to the position information of the abnormal pixel points, and the specific method is as follows:

[0077] Starting from the upper left corner pixel point of the suspected device box connected domain, the absolute values of the difference between the gray values of the current pixel and the pixel right to the current pixel and the pixel below the current pixel are calculated for each non-connected domain edge pixel point, and the absolute value of the difference between the two absolute values is calculated as the gray relative change intensity of each non-connected domain edge pixel point. The mean value of the gray relative change intensity of all non-connected domain edge pixel points is taken as the mutation judgment threshold, and the pixel points whose gray relative change intensity exceeds the mutation judgment threshold are recorded as abnormal pixel points. The abnormal pixel region capable of forming a closed region or constituting a closed region together with the connected domain edge, the pixel set capable of self-closing or connecting with the original edge of the suspected device box connected domain to form a closed region, and the region surrounded by the closed boundary are taken as the suspected damage area connected domain.

[0078] It should be noted that by calculating the "gray relative change intensity" composite index, the interference caused by single-direction light shadow or smooth gradient is effectively suppressed. Only when the gray value of the pixel point in the right and lower directions is discontinuous, the pixel point is determined to be abnormal. The specificity of damage identification is significantly improved, and the probability of misjudging the normal structure edge as damage is reduced. The mean value of the gray relative change intensity of all non-edge pixel points is taken as the adaptive mutation judgment threshold, so that the method can intelligently adapt to images taken under different times and different lighting conditions without manually setting a fixed threshold, thereby enhancing the robustness and practicality of the algorithm.

[0079] Step S004: Taking the geometric center of the connected domain of each suspected damage area as the origin, straight lines passing through the origin are drawn in K different directions, each straight line intersects the connected domain edge to form two line segments; the average value of the length ratio of the two line segments in all directions is calculated as the probability that each suspected damage area connected domain is a regular shape, and the first probability parameter of each suspected damage area connected domain is obtained according to the difference between the average value and 1.

[0080] It should be noted that the device usually has a regular-shaped box component, and the damaged part is relatively irregular in shape. Based on this, the shape regularity is judged by calculating the distance from each edge point to the geometric center of the image based on the shape division in the previous step, and the probability that it is a damage area is further judged.

[0081] It is further needed to be explained that the aforementioned step preliminarily locates the gray abnormal area, but it may contain normal gray changes caused by regular structures inherent to the equipment box (such as trademarks, indicator lights, etc.), and if it is directly determined that all of them are damages, it will lead to an increase in false positive rate. In order to distinguish between real damage and normal structure, this step introduces a discrimination mechanism based on geometric shape. The morphology of real damage (such as cracks, peeling) is usually irregular and asymmetric; while the inherent parts on the box often have relatively regular and symmetric geometric shapes. Traditional methods such as calculating rectangularity or aspect ratio can only describe the shape from a limited dimension, and are not sensitive enough to capture complex irregularities. Therefore, this method innovatively uses a multi-directional symmetry evaluation strategy based on the length ratio of radial line segments. By emitting rays from the geometric center of the region in all directions and calculating the length ratio of two line segments on each ray, the symmetry degree of the shape in each direction can be quantified. A perfect central symmetric figure (such as a circle, a square), the line segment ratio in any direction should be 1. By calculating the average value of the ratio in all directions, the probability that the entire region approaches a central symmetric shape can be obtained, thereby effectively distinguishing between regular components and irregular damage.

[0082] Specifically, taking the geometric center of each suspected damage region as the origin, straight lines passing through the origin are drawn in K different directions, each straight line intersects the edge of the connected domain to form two line segments; the average value of the length ratio of the two line segments in all directions is calculated as the probability that each suspected damage region connected domain is a regular shape, and the first probability parameter of each suspected damage region connected domain is obtained according to the difference between the average value and 1, including the specific method:

[0083] The geometric center of each suspected damage region connected domain is obtained, and a rectangular coordinate system is constructed with the geometric center as the origin, the horizontal direction as the horizontal axis and the vertical direction as the vertical axis. Starting from the horizontal direction, a straight line is drawn through the origin every K degrees to intersect the edge of the suspected damage region at two points, the distances of the two points to the origin are calculated respectively, and the average value of the ratio of the two distances is calculated as follows:

[0084]

[0085] In the formula, Pm,n represents the probability that the nth suspected damage region connected domain in the mth suspected equipment box connected domain is a regular shape, Pm,n represents the probability that the nth suspected damage region connected domain in the mth suspected equipment box connected domain is a regular shape, and respectively represent the two intersection points of the straight line passing through the origin and the nth suspected damage region connected domain in the mth suspected equipment box connected domain, wherein distance from the origin is less than or equal to distance from the origin, and respectively represent the numerical values of the abscissa and ordinate of and respectively represent the numerical values of the abscissa and ordinate of and respectively represent the numerical values of the abscissa and ordinate of .

[0086] It should be noted that by multi-directional intensive sampling, comprehensive and fine measurement of irregular shape is realized, the limitations of single direction evaluation are overcome, and various irregular damages have strong capturing ability. The probability of regular shape has clear mathematical meaning: the closer the value is to 1, the more regular the region is, and the first probability parameter directly quantifies the irregularity of the shape. This parameter is positively correlated with the damage probability. The method is not sensitive to the position and size of the damage region. No matter where the damage is located, its irregularity can be effectively detected by this method, which has strong robustness and universality. The invention provides a stable and reliable shape feature basis for subsequent final probability fusion judgment. The value of K is not limited in the invention. The smaller the value of K is, the higher the accuracy is, but the larger the calculation amount is. In the embodiment, the value of K is 1.

[0087] Step S005: Taking the connected domain geometric center of each suspected damage region as the center, a circle is made with a fixed pixel interval E and the radius is gradually increased until the circumference exceeds the connected domain boundary. All circles that do not exceed the boundary are retained. By calculating the difference between the average gray values of the pixels covered by two adjacent circles, the second probability parameter of each suspected damage region connected domain is obtained.

[0088] It should be noted that the above steps calculate the regularity probability of the region shape to determine the probability of the damaged region, but the shape characteristics of some damaged regions are very close to the standard shape, which reduces the discrimination of the method. In order to improve the recognition accuracy, the texture features of the region are introduced in this step to further calculate the probability of being a damaged region. The perfect box surface usually has uniform coating and light reflection characteristics, and the gray distribution in the local region is gently changed. Due to the material loss, dirt adhesion or surface roughness change, the micro-texture structure of the damaged region is destroyed, which further shows the dramatic fluctuation or step change of the radial gray distribution.

[0089] Specifically, with the geometric center of each suspected damage area as the center, a circle is made with a fixed pixel interval E and gradually increasing radius until the circumference exceeds the boundary of the connected domain, and all circles that do not exceed the boundary are retained; by calculating the difference between the average gray values of the pixels covered by two adjacent circles, the second probability parameter of each suspected damage area connected domain is obtained, and the specific method includes:

[0090] With the geometric center of the nth suspected damage area connected domain in the mth suspected equipment box connected domain as the center, a circle is made with a fixed pixel interval E and gradually increasing radius until the circumference exceeds the boundary of the connected domain, and all circles that do not exceed the boundary are retained, and all circles are traversed from small to large, and the second probability parameter of the nth suspected damage area connected domain in the mth suspected equipment box connected domain is calculated as follows:

[0091] ;

[0092] ;

[0093] In the formula, represents the difference between the average gray values of the pixels in the fth circle in the nth suspected damage area connected domain in the mth suspected equipment box connected domain and the pixels in the (f-1)th circle, represents the number of pixels in the fth circle in the nth suspected damage area connected domain in the mth suspected equipment box connected domain, represents the gray value of the ith pixel in the fth circle in the nth suspected damage area connected domain in the mth suspected equipment box connected domain, represents the number of pixels in the (f-1)th circle in the nth suspected damage area connected domain in the mth suspected equipment box connected domain, represents the gray value of the jth pixel in the (f-1)th circle in the nth suspected damage area connected domain in the mth suspected equipment box connected domain, represents the gray value of the lth pixel in the (f-1)th circle in the nth suspected damage area connected domain in the mth suspected equipment box connected domain, represents the second probability parameter of the nth suspected damage area connected domain in the mth suspected equipment box connected domain.

[0094] It should be noted that the perfect box surface usually has uniform coating and light reflection characteristics, and the gray scale distribution in the local area is gently changed; while the damaged area will cause the destruction of the micro-texture structure due to the lack of material, the attachment of dirt or the change of surface roughness, and then show the sharp fluctuation or step change of the radial gray scale distribution; the relative difference of the average gray scale of the adjacent circular rings is calculated, which is not sensitive to the overall brightness and contrast change of the image, and the second probability parameter is the average value of the absolute value of the relative difference between all adjacent circular rings, the greater the value, the more intense and unstable the gray gradient change in the region, and the more likely it belongs to the damaged area.

[0095] Step S006: Calculate the first probability parameter and the second probability parameter of each suspected damage area connected domain, obtain the damage probability of each suspected damage area connected domain, and judge whether the wind farm network test equipment box is damaged according to the damage probability of each suspected damage area connected domain.

[0096] It should be noted that the above steps extract damage features from two independent and complementary physical dimensions of shape irregularity and texture unevenness, but if either feature is used alone for judgment, it may lead to misjudgment, so this step needs to combine the first probability parameter and the second probability parameter to obtain a comprehensive index.

[0097] Specifically, the first probability parameter and the second probability parameter of each suspected damage area connected domain are calculated, the damage probability of each suspected damage area connected domain is obtained, and whether the wind farm network test equipment box is damaged is judged according to the damage probability of each suspected damage area connected domain, including the specific method:

[0098]

[0099] In the formula, The first probability parameter of the nth suspected damage area connected domain in the mth suspected equipment box connected domain is represented as Pm(n), and the second probability parameter of the nth suspected damage area connected domain in the mth suspected equipment box connected domain is represented as Qm(n). The first probability parameter of the nth suspected damage area connected domain in the mth suspected equipment box connected domain is represented as Pm(n), and the second probability parameter of the nth suspected damage area connected domain in the mth suspected equipment box connected domain is represented as Qm(n). The probability of damage of the nth suspected damage area connected domain in the mth suspected equipment box connected domain is represented as Pm(n), and the threshold Q is set, when the damage probability of any damage connected domain in any suspected equipment box connected domain in the image is greater than or equal to Q, the equipment box is damaged, prompting the relevant personnel to maintain.

[0100] It should be noted that: The multiplication relation makes the comprehensive probability very sensitive to the change of two characteristic parameters, and the increase of any parameter value directly leads to the increase of the comprehensive probability, embodies the logical AND relation that the damage probability is high when both shape and texture characteristics are abnormal, effectively amplifies the signal of the real damage, and after the second probability parameter is added by 1 to participate in the operation, the annihilation effect of the multiplication operation on the shape abnormal characteristic when the texture abnormality is not significant is skillfully avoided, the basic weight of the shape characteristic in decision is ensured, the complete automation and objectivity of the detection process are realized, the inspection efficiency is significantly improved, an instant and reliable scientific basis for preventive maintenance of the wind power plant equipment is provided, and finally the core target of reducing operation and maintenance risk and cost is achieved.

[0101] Please refer to Figure 2 which shows a structure block diagram of a wind power plant network-involved test equipment state inspection system fusing unmanned aerial vehicle vision, the system comprising the following modules:

[0102] An image acquisition and processing module is configured to acquire equipment box images by using an unmanned aerial vehicle, pre-process the equipment box images, and obtain grayscale images.

[0103] A device detection module is configured to obtain abnormal pixel points according to the difference between the grayscale values of non-edge pixel points and their right and lower pixel points in the suspected equipment box connected domain, and obtain suspected damage area connected domains according to the position information of the abnormal pixel points.

[0104] A suspected damage area detection module is configured to obtain abnormal pixel points according to the difference between the grayscale values of non-edge pixel points and their right and lower pixel points in the suspected equipment box connected domain, and obtain suspected damage area connected domains according to the position information of the abnormal pixel points.

[0105] A first probability parameter calculation module is configured to take the geometric center of each suspected damage area connected domain as the origin, draw straight lines passing through the origin along K different directions, form two line segments by the intersection of each straight line and the connected domain edge, calculate the average value of the length ratio of the two line segments in all directions as the probability that each suspected damage area connected domain is a regular shape, and obtain the first probability parameter of each suspected damage area connected domain according to the difference between the average value and 1.

[0106] A second probability parameter calculation module is configured to take the geometric center of each suspected damage area connected domain as the center of a circle, draw circles with a fixed pixel interval E, and increase the radius step by step until the circumference exceeds the connected domain boundary, and retain all circles that do not exceed the boundary; the second probability parameter of each suspected damage area connected domain is obtained by calculating the difference between the average grayscale values of the pixel points covered by adjacent two circles.

[0107] The device damage judgment module is configured to calculate a first probability parameter and a second probability parameter of each suspected damage area connected domain, obtain a damage probability of each suspected damage area connected domain, and determine whether the wind power plant network test device box is damaged according to the damage probability of each suspected damage area connected domain.

[0108] A third object of the embodiments of the present application is to provide a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned wind power plant network test device state inspection method combined with unmanned aerial vehicle vision.

[0109] A fourth object of the embodiments of the present application is to provide a computer readable storage medium, which stores a computer program, wherein the computer program is executable on a processor to implement the steps of the above-mentioned wind power plant network test device state inspection method combined with unmanned aerial vehicle vision.

[0110] The present application realizes automatic inspection through unmanned aerial vehicle vision technology, avoids the risk of manual high-altitude operation, improves the inspection efficiency and accuracy, and ensures the objectivity and reliability of damage detection through a multi-step image processing algorithm.

[0111] By selecting the image with the largest discrete degree as the test image, the image clarity is ensured, high-quality input is provided for subsequent processing, and the overall detection accuracy is improved.

[0112] The suspected device box is screened by using the connected domain rule degree, natural background interference is effectively excluded, the target area is accurately positioned, and false detection is reduced.

[0113] Abnormal pixel points are adaptively detected by the relative change intensity of gray scale, different light conditions are adapted, damage areas are accurately identified, and the probability of false judgment is reduced.

[0114] Based on multi-direction shape symmetry evaluation, the rule degree of the area is quantified, the real damage is effectively distinguished from the regular components, and the damage identification specificity is improved.

[0115] Through radial gray scale change analysis, texture abnormalities are captured, the detection ability of microscopic damage is enhanced, and the detection comprehensiveness is improved.

[0116] The damage probability is calculated by combining shape and texture features, multi-dimensional judgment is realized, the false positive rate is significantly reduced, and the scientific nature of the maintenance decision is ensured.

[0117] Those skilled in the art will appreciate that embodiments of the application can be devised for a method, a system, or a computer program product. Accordingly, the present application can be embodied in the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application can take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage devices, etc.) embodying computer readable program code.

[0118] The present application is described in reference to the flowchart and / or block diagrams of the method, apparatus (system) and computer program product according to embodiments of the application. It will be understood that each block of the flowchart and / or block diagrams, and combinations of blocks in the flowchart and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general purpose computer, special purpose computer, embedded processing device or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart and / or block diagram block or blocks. Figure 1 one or more functions specified in the flowchart and / or block diagram block or blocks. Figure 1 one or more functions specified in the flowchart and / or block diagram block or blocks.

[0119] These computer program instructions can also be stored in a computer- readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the function specified in the flowchart and / or block diagram block or blocks. Figure 1 one or more functions specified in the flowchart and / or block diagram block or blocks. Figure 1 one or more functions specified in the flowchart and / or block diagram block or blocks.

[0120] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart and / or block diagram block or blocks. Figure 1 one or more functions specified in the flowchart and / or block diagram block or blocks. Figure 1 one or more functions specified in the flowchart and / or block diagram block or blocks.

[0121] Finally, it should be noted that the above-mentioned embodiments are merely intended for describing the technical solutions of the present application, but not for limiting it. Although the present application is described in detail with reference to the above embodiments, those skilled in the field should understand that the technical solutions of the present application can be modified or replaced equivalently without departing from the spirit and scope of the present application, and any modification or equivalent replacement should be covered in the protection scope of the claims of the present application.

Claims

1. A method for condition inspection of wind farm grid-connected testing equipment integrating UAV vision, characterized in that, The method includes the following steps: Use drones to collect images of the equipment enclosure, preprocess the images to obtain grayscale images; Edge detection is performed on the grayscale image, and multiple connected components are obtained based on the edge detection results. The connected components of the suspected equipment enclosure are determined based on the ratio of the perimeter of the connected component outline to its area. Abnormal pixels are obtained based on the difference in grayscale values ​​between non-edge pixels and pixels directly to the right and directly below them in the suspected connected domain of the equipment enclosure; and the connected domain of the suspected damaged area is obtained based on the location information of the abnormal pixels. Taking the geometric center of the connected domain of each suspected damaged region as the origin, draw straight lines through the origin in K different directions. Each straight line intersects the edge of the connected domain to form two line segments. Calculate the average of the ratio of the lengths of these two line segments in all directions as the probability that the connected domain of each suspected damaged region is a regular shape, and obtain the first probability parameter of the connected domain of each suspected damaged region based on the difference between the average value and 1. Using the geometric center of the connected domain of each suspected damaged region as the center, circles are drawn with the radius gradually increased at a fixed pixel interval E until the circumference exceeds the boundary of the connected domain. All circles that do not exceed the boundary are retained. The second probability parameter of the connected domain of each suspected damaged region is obtained by calculating the difference in the average gray value of the pixels covered by two adjacent circles. The process of calculating the first and second probability parameters of the connected domain for each suspected damaged area to obtain the damage probability of each connected domain, and determining whether the wind farm grid-connected testing equipment enclosure is damaged based on the magnitude of the damage probability of each connected domain, includes the following specific methods: ； In the formula, Let represent the first probability parameter of the connected component of the nth suspected damaged area within the connected component of the mth suspected equipment enclosure. Let represent the second probability parameter of the connected domain of the nth suspected damaged area within the connected domain of the mth suspected equipment enclosure. Let Q represent the probability that the nth suspected damaged region in the connected region of the mth suspected equipment enclosure is damaged. A threshold Q is set. If the probability that any damaged region in any connected region of any suspected equipment enclosure in the image is damaged is greater than or equal to Q, then the equipment enclosure is damaged, and relevant personnel are reminded to carry out maintenance.

2. The method for status inspection of wind farm grid connection testing equipment integrating UAV vision as described in claim 1, characterized in that, The specific method for using a drone to collect images of the equipment enclosure, and preprocessing the images to obtain grayscale images, includes: Using a drone, S images of the wind farm grid connection test equipment enclosure at the same location were taken. The dispersion of the grayscale value of each pixel in the image was calculated, and the image with the largest dispersion was selected as the test image. The test image was then converted to grayscale to obtain a grayscale image.

3. The method for status inspection of wind farm grid connection testing equipment integrating UAV vision as described in claim 1, characterized in that, The method for performing edge detection on a grayscale image, obtaining multiple connected components based on the edge detection results, and determining suspected connected components of the equipment enclosure based on the ratio of the perimeter to the area of ​​the connected component outline includes the following specific steps: Edge detection is performed on the image using the Canny operator to obtain a binarized image with clear edges. The p connected components with closed edges are then identified, and the regularity of the m-th connected component is calculated using the following method: ； In the formula, This indicates the degree of regularity of the m-th connected component. This represents the perimeter of the m-th connected component. This represents the number of pixels in the m-th connected component; Set threshold ,judge The magnitude of the absolute value of the difference from 1, if the absolute value of the difference is less than or equal to Then the m-th connected component is denoted as the suspected device enclosure connected component.

4. The method for status inspection of wind farm grid connection testing equipment integrating UAV vision as described in claim 1, characterized in that, The method for obtaining abnormal pixels based on the difference in grayscale values ​​between non-edge pixels and pixels directly to their right and below in the suspected connected region of the equipment enclosure, and for obtaining the suspected damaged area connected region based on the location information of the abnormal pixels, includes the following specific methods: Starting from the top-left pixel of the suspected connected domain of the device enclosure, traverse the network. For each pixel at the edge of a non-connected domain, calculate the absolute value of the difference between the current pixel and the pixel to its right, and the pixel below it. Then calculate the absolute value of the difference between these two absolute values ​​as the relative grayscale change intensity of each pixel at the edge of a non-connected domain. Use the average of the relative grayscale change intensities of all pixels at the edges of non-connected domains as the mutation judgment threshold. Pixels whose relative grayscale change intensities exceed the mutation judgment threshold are recorded as abnormal pixels. Abnormal pixel regions that can form closed regions or together with the edges of connected domains form closed regions are considered as suspected damaged connected domains.

5. The method for status inspection of wind farm grid connection testing equipment integrating UAV vision according to claim 1, characterized in that, The method involves taking the geometric center of the connected domain of each suspected damaged region as the origin, drawing straight lines through the origin in K different directions, and forming two line segments by intersecting the edge of the connected domain in each direction. The average ratio of the lengths of these two line segments in all directions is calculated as the probability that the connected domain of each suspected damaged region has a regular shape. The first probability parameter of the connected domain of each suspected damaged region is obtained based on the difference between this average value and 1. The specific method includes: Process each suspected damaged region's connected components sequentially, obtaining the geometric center of each region's connected components. Construct a Cartesian coordinate system with the geometric center as the origin, drawing a horizontal axis through the origin and a vertical axis through the origin. Starting horizontally, draw a straight line every K degrees from the origin, intersecting the edge of the suspected damaged region at two points. Calculate the distances from these two points to the origin, and then calculate the ratio of these two distances. The average of these ratios is calculated using the following method: ； In the formula, Let represent the probability that the nth suspected damaged area within the connected component of the mth suspected equipment enclosure is a regular shape. This represents the coordinates of the origin of the coordinate system within the connected domain of the m-th suspected damaged area, where K represents the angular interval. and They represent points passing through the origin respectively. The straight line intersects the nth suspected damaged area in the connected domain of the mth suspected equipment enclosure at two points, where... The distance from the origin is less than or equal to Distance from the origin, and They represent The values ​​of the x and y coordinates, and They represent The values ​​of the x and y coordinates, and They represent Given the x and y coordinates, the first probability parameter of the connected region of the nth suspected damaged area within the connected region of the mth suspected equipment enclosure is 1- .

6. The method for status inspection of wind farm grid connection testing equipment integrating UAV vision according to claim 1, characterized in that, The method involves drawing circles with the geometric center of the connected domain of each suspected damaged region as the center, and gradually increasing the radius at fixed pixel intervals E until the circumference exceeds the boundary of the connected domain, retaining all circles that do not exceed the boundary; obtaining the second probability parameter of the connected domain of each suspected damaged region by calculating the difference in the average grayscale value of the pixels covered by two adjacent circles, including the following specific methods: Using the geometric center of the connected region of the nth suspected damaged area within the connected region of the mth suspected device enclosure as the center, draw circles with the radius increasing successively at fixed pixel intervals E until the circumference exceeds the boundary of the connected region. Retain all circles that do not exceed the boundary. Traverse all circles from smallest to largest. The second probability parameter of the connected region of the nth suspected damaged area within the connected region of the mth suspected device enclosure is calculated as follows: ； ； In the formula, This represents the difference in average grayscale value between the pixel containing the f-th circle and the pixel containing the (f-1)-th circle within the connected region of the m-th suspected damaged area. This represents the number of pixels containing the f-th circle in the connected region of the n-th suspected damaged area within the connected region of the m-th suspected device enclosure. This represents the grayscale value of the i-th pixel of the f-th circle within the connected region of the n-th suspected damaged area in the connected region of the m-th suspected device enclosure. This represents the number of pixels containing the (f-1)th circle in the connected region of the nth suspected damaged area within the connected region of the mth suspected device enclosure. This represents the grayscale value of the j-th pixel in the (f-1)-th circle of the (n-th)-th suspected damaged area within the connected region of the m-th suspected device enclosure. This represents the grayscale value of the l-th pixel of the (f-1)-th circle within the connected region of the n-th suspected damaged area in the connected region of the m-th suspected device enclosure. The second probability parameter represents the connected domain of the nth suspected damaged area within the connected domain of the mth suspected equipment enclosure.

7. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the method for inspecting the status of wind farm grid-connected testing equipment that integrates UAV vision as described in any one of claims 1 to 6.

8. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method for inspecting the status of wind farm grid-connected testing equipment that integrates UAV vision as described in any one of claims 1 to 6.

9. A status inspection system for wind farm grid connection testing equipment integrating UAV vision, characterized in that, The system includes the following modules: The image acquisition and processing module is used to acquire images of the equipment enclosure using a drone, preprocess the images of the equipment enclosure, and obtain grayscale images. The equipment detection module is used to obtain abnormal pixels based on the difference in grayscale values ​​between non-edge pixels and pixels directly to its right and directly below in the connected domain of a suspected equipment enclosure, and to obtain the connected domain of the suspected damaged area based on the location information of the abnormal pixels. The suspected damaged area detection module obtains abnormal pixels based on the difference in grayscale values ​​between non-edge pixels and pixels directly to its right and directly below in the connected domain of the suspected equipment enclosure, and obtains the connected domain of the suspected damaged area based on the location information of the abnormal pixels. The first probability parameter calculation module is used to draw straight lines through the origin along K different directions with the geometric center of the connected domain of each suspected damaged area as the origin. Each straight line intersects the edge of the connected domain to form two line segments. The average of the ratio of the lengths of these two line segments in all directions is calculated as the probability that the connected domain of each suspected damaged area is a regular shape. The first probability parameter of the connected domain of each suspected damaged area is obtained based on the difference between the average value and 1. The second probability parameter calculation module is used to draw circles with the geometric center of the connected domain of each suspected damaged area as the center, and the radius is increased successively at a fixed pixel interval E until the circumference exceeds the boundary of the connected domain, and all circles that do not exceed the boundary are retained; by calculating the difference in the average gray value of the pixels covered by two adjacent circles, the second probability parameter of the connected domain of each suspected damaged area is obtained. The equipment damage judgment module is used to calculate the first probability parameter and the second probability parameter of the connected domain of each suspected damaged area, obtain the damage probability of the connected domain of each suspected damaged area, and determine whether the wind farm grid-connected test equipment enclosure is damaged based on the magnitude of the damage probability of the connected domain of each suspected damaged area. The specific methods for calculating the first probability parameter and the second probability parameter of the connected domain of each suspected damaged area, obtaining the damage probability of the connected domain of each suspected damaged area, and determining whether the wind farm grid-connected test equipment enclosure is damaged based on the magnitude of the damage probability of the connected domain of each suspected damaged area are as follows: ； In the formula, Let represent the first probability parameter of the connected component of the nth suspected damaged area within the connected component of the mth suspected equipment enclosure. Let represent the second probability parameter of the connected domain of the nth suspected damaged area within the connected domain of the mth suspected equipment enclosure. Let Q represent the probability that the nth suspected damaged region in the connected region of the mth suspected equipment enclosure is damaged. A threshold Q is set. If the probability that any damaged region in any connected region of any suspected equipment enclosure in the image is damaged is greater than or equal to Q, then the equipment enclosure is damaged, and relevant personnel are reminded to carry out maintenance.

Images