A method and system for locating insulator discharge areas based on solar-blind ultraviolet imaging

Abstract

The present invention discloses a method and system for locating the discharge area of an insulator based on solar-blind ultraviolet imaging. The current method relies on manual identification of discharge photons in the insulator ultraviolet discharge detection video to determine whether the discharge exists and the location of the discharge, and roughly determine the amount of surface area of the equipment covered by the discharge. This method relies on the subjective judgment of the test personnel. At the same time, since ultraviolet discharge presents a fluctuating characteristic, the judgment of the personnel on the discharge coverage area is relatively error-free. The present invention is a method and system for locating the discharge area of an insulator using image frame superposition and ultraviolet image processing. The method obtains the boundary of the ultraviolet discharge area through image frame superposition, ultraviolet photon number filtering and polygonal approximation, and then obtains the ratio of the ultraviolet discharge area length to the insulator length, thereby avoiding the errors caused by the subjective judgment of the test personnel and the volatility of ultraviolet discharge, and can greatly improve the work efficiency of ultraviolet discharge data processing.

Description

Technical Field

[0001] The present invention relates to the technical field of live transmission line detection, and in particular to the identification and recognition of on-site insulator discharge defects, in particular to a method and system for locating an insulator discharge area based on solar-blind ultraviolet imaging. Background Art

[0002] To meet the electricity demand of the rapidly growing industrial and agricultural sectors, my country's power grid continues to expand in size and voltage levels, and the number of insulators in operation is increasing. Detecting and controlling insulator conditions is crucial for the safe operation of the power grid. UV imaging can detect different types of discharges on insulator surfaces, including corona and arc discharges. It offers high sensitivity, is unaffected by sunlight, can observe surface discharges outdoors in daylight, and has a long detection range. Internal defects and severe contamination on insulators can cause surface discharges. The location and length of the discharge area are crucial for describing the insulator's operating status, the severity of the defect, and determining subsequent maintenance strategies.

[0003] However, the analysis of ultraviolet images currently relies on manual work, requiring humans to determine whether discharge exists, the length of the discharge area, and then the degree of threat the discharge poses to operations. Due to the volatility of ultraviolet discharge, subjective judgments suffer from insufficient analysis efficiency and accuracy. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to overcome the defects of the above-mentioned prior art and provide a method for locating the insulator discharge area based on image frame superposition and ultraviolet image processing, so as to avoid the errors caused by the subjective judgment of the test personnel and the volatility of ultraviolet discharge, and greatly improve the work efficiency of ultraviolet discharge data processing.

[0005] To this end, a technical solution adopted by the present invention is as follows: a method for locating an insulator discharge area based on solar-blind ultraviolet imaging, comprising the steps of:

[0006] a) Collect equipment data and obtain the insulator structure height;

[0007] b) Conduct UV testing on insulators in operation and obtain UV imaging videos of insulators;

[0008] c) preprocessing the UV imaging video frame using the HSV color model to convert the RGB color space image into the HSV color space image;

[0009] d) extracting N consecutive or evenly spaced frames from the UV imaging video, superimposing the N frames to obtain a scatter plot of the cumulative UV photon number distribution of the N frames, i.e., the initial UV discharge area;

[0010] e) using a mean filter to smooth the UV discharge area on the UV photon distribution scatter plot, taking the average of the target pixel (i.e., the pixel where the photon is located) and the surrounding pixels, and then performing target pixel filling and isolated photon elimination processing;

[0011] f) using a polygonal approximation method to enclose the UV discharge area based on the result obtained in step e), so as to regularize the irregular area;

[0012] g) After obtaining the minimum circumscribed polygonal outline of each discharge region, calculate the contour area, contour perimeter, and maximum diagonal length of the corresponding discharge region, all of which are counted in pixels, and count the number of discharge photons within the contour area;

[0013] h) After labeling the insulator image samples, a target detection algorithm is used to establish a detection model for the insulator in the UV imaging video. Rotational target detection of the insulator is performed on the UV imaging video frame, and the position information of the insulator's circumscribed rectangle in the UV video frame is returned.

[0014] i) The length of the UV discharge area on the insulator surface is calculated by the length of the minimum circumscribed polygon of the discharge area along the insulator direction.

[0015] The present invention realizes the positioning of the insulator discharge area through image frame superposition and ultraviolet photon area image processing.

[0016] Furthermore, in step c), the HSV color model needs to set the value range of the dimension in HSV according to the photon color in the ultraviolet imaging palette, such as setting the range of the H dimension of red photons in the HSV space to 0-10 and 160-179, 255>S>100, 255>V>100.

[0017] Furthermore, in step e), the mean filtering is to select a (2n+1)×(2n+1) window in the scatter plot, where n≥1, and the average filter value of the point (x, y) is calculated as follows:

[0018]

[0019] Among them, (x, y) is the spatial coordinate of the scattered point in the scatter plot, and the coordinate unit is pixel; P(i, j) is the original number of ultraviolet photons at a certain spatial position, and p(x, y) is the number of photons at this spatial position after mean filtering.

[0020] Furthermore, in step e), the target pixel filling is to perform a morphological closing operation on the filtered and smoothed image to fill the small cracks, discontinuities and holes between photons in the discharge area. For the image X and the structural element S, the closing operation formula is:

[0021]

[0022] The structural element S is a 3×3 kernel matrix, and the anchor point is located at the center of the matrix.

[0023] Furthermore, in step e), the isolated photon elimination process is to eliminate isolated small dots and burrs through an opening operation, thereby further eliminating the noise of the image X. The formula for the opening operation is:

[0024]

[0025] The structural element S is a 3×3 kernel matrix, and the anchor point is located at the center of the matrix.

[0026] Furthermore, in step h), a detection model of insulators in the ultraviolet imaging video is established by the following steps:

[0027] h1) preprocessing the insulator image samples to form an image dataset that meets the standards;

[0028] h2) expanding the dataset of the insulator image samples processed in step h1); using a labeling tool to mark the insulator region in the insulator image with a rectangular frame, and generating a label text corresponding to the image based on the position information;

[0029] h3) Using the insulator image labeled in step h2) as a sample, a convolutional neural network is trained using a target detection algorithm, and an insulator detection model is obtained according to training adjustment parameters to obtain the insulator circumscribed rectangle.

[0030] Furthermore, in step i), the length ratio is calculated as follows:

[0031]

[0032] Among them, L is the total length of the insulator, l i is the length of a discharge area along the insulator direction, and P is the ratio of the length of the ultraviolet discharge area on the insulator surface.

[0033] Another technical solution adopted by the present invention is: an insulator discharge area positioning system based on solar-blind ultraviolet imaging, which includes:

[0034] Insulator height acquisition unit: collects equipment data and obtains the insulator structure height;

[0035] Insulator UV imaging video acquisition unit: performs UV testing on insulators in operation and acquires UV imaging videos of insulators;

[0036] Video frame preprocessing unit: Use the HSV color model to preprocess the UV imaging video frame and convert the RGB color space image into the HSV color space image;

[0037] Initial UV discharge area acquisition unit: extracts N consecutive or evenly spaced frames from the UV imaging video, superimposes the N frames, and obtains a scatter plot of the UV photon number distribution accumulated in the N frames, i.e., the initial UV discharge area;

[0038] Ultraviolet discharge area smoothing processing unit: Use mean filtering on the ultraviolet photon number distribution scatter plot to achieve smoothing of the ultraviolet discharge area, take the average value of the target pixel (i.e. the pixel where the photon is located) and the surrounding pixels, and then perform target pixel filling and isolated photon elimination processing;

[0039] Graphic regularization processing unit: Use polygonal approximation method to enclose the smoothed UV discharge area to regularize the irregular area;

[0040] Discharge photon number calculation unit: After finding the minimum circumscribed polygon outline of each discharge area, the contour area, contour perimeter and maximum diagonal length of the corresponding discharge area are calculated. The above parameters are counted in pixels, and the number of discharge photons in the contour area is counted;

[0041] Insulator detection model building unit: Labels insulator image samples, uses the target detection algorithm to build an insulator detection model in the UV imaging video, performs insulator rotation target detection on the UV imaging video frame, and returns the position information of the insulator circumscribed rectangle in the UV video frame;

[0042] Length ratio calculation unit: calculates the length ratio of the ultraviolet discharge area on the insulator surface through the length of the minimum circumscribed polygon of the discharge area along the insulator direction.

[0043] Compared with the existing methods, the present invention has the following beneficial technical effects: the present invention uses the superposition of image frames, ultraviolet photon number filtering, and polygonal approximation to obtain the boundary of the ultraviolet discharge area, and then obtains the ratio of the ultraviolet discharge area length to the insulator length, avoiding the errors caused by the subjective judgment of the test personnel and the volatility of ultraviolet discharge, and can greatly improve the work efficiency of ultraviolet discharge data processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0045] Figure 1 is a flow chart of the method of the present invention;

[0046] Figure 2 This is a typical frame image of an insulator ultraviolet imaging video in a specific embodiment of the present invention;

[0047] Figure 3 This is a result diagram of converting an insulator discharge image frame into an HSV color space in a specific embodiment of the present invention;

[0048] Figure 4 This is a scatter diagram of photon distribution of N frames of cumulative discharge images of an insulator in a specific embodiment of the present invention;

[0049] Figure 5 This is a diagram showing the result of morphological filtering processing of an insulator discharge image in a specific embodiment of the present invention;

[0050] Figure 6 A contour diagram of the discharge area is shown, in which a polygon circumscribing the discharge area of the insulator is positioned in a specific embodiment of the present invention;

[0051] Figure 7 A diagram of an insulator identification area and a discharge area in a specific embodiment of the present invention;

[0052] Figure 8 It is a structural block diagram of the system of the present invention. DETAILED DESCRIPTION

[0053] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0054] Example 1

[0055] This embodiment provides a method for locating the discharge area of an insulator. Figure 1 As shown, the steps are as follows:

[0056] a) Collect equipment data and obtain the insulator structure height;

[0057] b) Conduct UV testing on insulators in operation and obtain UV imaging videos of insulators;

[0058] c) preprocessing the UV imaging video frame using the HSV color model to convert the RGB color space image into the HSV color space image;

[0059] d) extracting N consecutive or evenly spaced frames from the UV imaging video, for example, N=50, and superimposing the N frames to obtain a scatter plot of the cumulative UV photon number distribution of the N frames, i.e., the initial UV discharge area;

[0060] e) using a mean filter to smooth the UV discharge area on the UV photon distribution scatter plot, taking the average of the target pixel (i.e., the pixel where the photon is located) and the surrounding pixels, and then performing target pixel filling and isolated photon elimination processing;

[0061] f) using a polygonal approximation method to enclose the UV discharge area based on the result obtained in step e), so as to regularize the irregular area;

[0062] g) After obtaining the minimum circumscribed polygonal outline of each discharge region, calculate the contour area, contour perimeter, and maximum diagonal length of the corresponding discharge region, all of which are counted in pixels, and count the number of discharge photons within the contour area;

[0063] h) After labeling the insulator image samples, the S2ANet algorithm is used to establish an insulator detection model in the UV imaging video. The insulator rotation target detection is performed on the UV imaging video frame, and the position information of the insulator circumscribed rectangle in the UV video frame is returned.

[0064] i) The length of the UV discharge area on the insulator surface is calculated by the length of the minimum circumscribed polygon of the discharge area along the insulator direction.

[0065] Specifically, in step c), the HSV color model needs to set the value range in HSV according to the photon color in the ultraviolet imaging palette, such as setting the H dimension range of red photons in the HSV space to 0-10 and 160-179, 255>S>100, and 255>V>100.

[0066] Specifically, in step e), the mean filtering is to select a (2n+1)×(2n+1) window in the scatter plot, where n≥1 and is generally 1-3. The average filter value of the point (x, y) is calculated as follows:

[0067]

[0068] Among them, (x, y) is the spatial coordinate of the scattered point in the scatter plot, and the coordinate unit is pixel; P(i, j) is the original number of ultraviolet photons at a certain spatial position, and p(x, y) is the number of photons at this spatial position after mean filtering.

[0069] Specifically, in step e), the target pixel filling is to perform a morphological closing operation on the filtered and smoothed image to fill the small cracks, discontinuities and holes between photons in the discharge area. For the image X and the structural element S, the closing operation formula is:

[0070]

[0071] The structural element S is a 3×3 kernel matrix, and the anchor point is located at the center of the matrix.

[0072] Specifically, in step e), the isolated photon elimination process is to eliminate isolated small dots and burrs through an opening operation to further eliminate image noise. The formula for the opening operation is:

[0073]

[0074] The structural element S is a 3×3 kernel matrix, and the anchor point is located at the center of the matrix.

[0075] Specifically, the detection model of insulators in the ultraviolet imaging video is established by the following steps:

[0076] h1) Perform data preprocessing on the insulator image samples to form an image dataset that meets the standards. Image preprocessing includes removing redundant and mixed image samples;

[0077] h2) expanding the dataset of the insulator image samples processed in step h1); using a labeling tool to mark the insulator region in the insulator image with a rectangular frame, and generating a label text corresponding to the image based on the position information;

[0078] h3) Using the insulator image labeled in step h2) as a sample, a convolutional neural network is trained using the S2ANet algorithm, and an insulator detection model is obtained according to the training adjustment parameters to obtain the insulator circumscribed rectangle.

[0079] Specifically, in step i), the discharge length ratio is calculated as follows:

[0080]

[0081] Where L is the total length of the insulator, l i is the length of a discharge area along the insulator direction, and P is the ratio of the length of the ultraviolet discharge area on the insulator surface.

[0082] Application Examples

[0083] Taking a 220kV overhead line insulator as an example, the method of the present invention is used to locate the discharge defect of the insulator and calculate the length ratio. The specific steps are as follows:

[0084] 1) Collect equipment data and obtain the insulator structure height as 2240mm.

[0085] 2) Carry out UV test on the insulator to obtain the UV discharge imaging video of the insulator, one of which is shown in the attached picture. Figure 2 .

[0086] 3) Use the HSV color model to preprocess the UV discharge imaging video frame and convert the RGB color space image into the HSV color space image. Figure 2 The corresponding conversion results are shown in the attached Figure 3 .

[0087] 4) Extract 200 consecutive frames from the UV imaging video, superimpose the 200 frames of images, and obtain a scatter plot of the UV photon number distribution of the 200 frames, that is, the initial UV discharge area is as shown in the attached figure. Figure 4 .

[0088] 5) The mean filter is used to smooth the ultraviolet discharge area of the ultraviolet photon distribution scatter plot. The target pixel (i.e., the pixel where the photon is located) and the surrounding pixels are averaged, and then the target pixel is filled and isolated photons are eliminated.

[0089] The mean filter is to select a (2n+1)×(2n+1) window in the scatter plot, with n being 2. The formula for calculating the average filter value of point (x, y) is as follows:

[0090]

[0091] (x,y) is the spatial coordinate of the scatter plot, and the coordinate unit is pixel. P(i,j) is the original number of ultraviolet photons at a certain spatial position, and p(x,y) is the number of photons at that spatial position after mean filtering.

[0092] The target filling is to perform a morphological closing operation on the filtered and smoothed image to fill the small cracks, discontinuities, and holes between photons in the discharge area. For the image X and the structural element S, the closing operation formula is:

[0093]

[0094] The process of eliminating isolated photons is to eliminate isolated small dots and burrs through opening operation, thereby further eliminating the noise of the image. The formula for opening operation is:

[0095]

[0096] The structural element S is a 3×3 kernel matrix, and the anchor point is located at the center of the matrix. The filtering and morphological processing results are shown in the attached figure. Figure 5 .

[0097] 6) For the result of step 5), the discharge area is surrounded by the polygonal approximation method to regularize the irregular area and obtain the location of the discharge area, as shown in the attached figure. Figure 6 .

[0098] 7) After obtaining the minimum circumscribed polygonal contour of each discharge area, the contour area of the corresponding discharge area is calculated to be 916 pixels, the contour perimeter is 132 pixels, the maximum diagonal length is 42 pixels, and the number of discharge photons in the contour area is counted to be 745.

[0099] 8) After marking the insulator samples, the S2ANet algorithm is used to establish the insulator detection model in the UV imaging video. The rotating target of the insulator is detected on the UV imaging video frame, and the position information of the insulator circumscribed rectangle in the UV video frame is returned, as shown in the attached figure. Figure 7 .

[0100] 9) The length of the minimum circumscribed polygon of the discharge area along the insulator direction is calculated to be 7.03% of the length of the ultraviolet discharge area on the insulator surface.

[0101] Example 2

[0102] This embodiment provides an insulator discharge area positioning system based on solar-blind ultraviolet imaging, such as Figure 8 As shown, it includes:

[0103] Insulator height acquisition unit: collects equipment data and obtains the insulator structure height;

[0104] Insulator UV imaging video acquisition unit: performs UV testing on insulators in operation and acquires UV imaging videos of insulators;

[0105] Video frame preprocessing unit: Use the HSV color model to preprocess the UV imaging video frame and convert the RGB color space image into the HSV color space image;

[0106] Initial UV discharge area acquisition unit: extracts N consecutive or evenly spaced frames from the UV imaging video, superimposes the N frames, and obtains a scatter plot of the UV photon number distribution accumulated in the N frames, i.e., the initial UV discharge area;

[0107] Ultraviolet discharge area smoothing processing unit: Use mean filtering on the ultraviolet photon number distribution scatter plot to achieve smoothing of the ultraviolet discharge area, take the average value of the target pixel (i.e. the pixel where the photon is located) and the surrounding pixels, and then perform target pixel filling and isolated photon elimination processing;

[0108] Graphic regularization processing unit: Use polygonal approximation method to enclose the smoothed UV discharge area to regularize the irregular area;

[0109] Discharge photon number calculation unit: After finding the minimum circumscribed polygon outline of each discharge area, the contour area, contour perimeter and maximum diagonal length of the corresponding discharge area are calculated. The above parameters are counted in pixels, and the number of discharge photons in the contour area is counted;

[0110] Insulator detection model building unit: After annotating insulator image samples, the S2ANet algorithm is used to build an insulator detection model in the UV imaging video. The insulator's rotation target is detected in the UV imaging video frame, and the position information of the insulator's circumscribed rectangle in the UV video frame is returned.

[0111] Length ratio calculation unit: calculates the length ratio of the ultraviolet discharge area on the insulator surface through the length of the minimum circumscribed polygon of the discharge area along the insulator direction.

[0112] Specifically, in the video frame preprocessing unit, the HSV color model needs to set the value range in HSV according to the photon color in the ultraviolet imaging palette, such as setting the H dimension range of red photons in the HSV space to 0-10 and 160-179, 255>S>100, 255>V>100.

[0113] Specifically, in the ultraviolet discharge area smoothing processing unit, the mean filtering is to select a (2n+1)×(2n+1) window in the scatter plot, n≥1, and the average filtering value of the point (x, y) is calculated as follows:

[0114]

[0115] Where (x, y) is the spatial coordinate of the scattered point in the scatter plot, and the coordinate unit is pixel; P(i, j) is the original number of ultraviolet photons at a certain spatial position, and p(x, y) is the number of photons at this spatial position after mean filtering;

[0116] The target pixel filling is to perform a morphological closing operation on the filtered and smoothed image to fill the small cracks, discontinuities and holes between photons in the discharge area. For the image X and the structural element S, the closing operation formula is:

[0117]

[0118] Among them, the structural element S is a 3×3 kernel matrix, and the anchor point is located at the center of the matrix;

[0119] The isolated photon elimination process is to eliminate isolated small dots and burrs through an opening operation, thereby further eliminating the noise of the image X. The formula for the opening operation is:

[0120]

[0121] The structural element S is a 3×3 kernel matrix, and the anchor point is located at the center of the matrix.

[0122] Specifically, in the insulator detection model building unit, the insulator detection model in the ultraviolet imaging video is established by the following steps:

[0123] h1) preprocessing the insulator image samples to form an image dataset that meets the standards;

[0124] h2) expanding the dataset of the insulator image samples processed in step h1); using a labeling tool to mark the insulator region in the insulator image with a rectangular frame, and generating a label text corresponding to the image based on the position information;

[0125] h3) Using the insulator image labeled in step h2) as a sample, a convolutional neural network is trained using the S2ANet algorithm, and an insulator detection model is obtained according to the training adjustment parameters to obtain the insulator circumscribed rectangle.

[0126] Specifically, in the length ratio calculation unit, the length ratio is calculated as follows:

[0127]

[0128] Among them, L is the total length of the insulator, l i is the length of a discharge area along the insulator string direction, and P is the ratio of the length of the ultraviolet discharge area on the insulator surface.

[0129] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only intended to help understand the method and core concept of the present invention. At the same time, those skilled in the art will find that the specific implementation methods and application scopes may vary based on the concept of the present invention. In summary, the contents of this specification should not be construed as limiting the present invention.

Claims

1. A method for locating the discharge area of an insulator based on solar-blind ultraviolet imaging, characterized in that: Including steps: a) Collect equipment data and obtain the insulator structure height; b) Conduct UV testing on insulators in operation and obtain UV imaging videos of insulators; c) preprocessing the UV imaging video frames using the HSV color model to convert the RGB color space image into the HSV color space image, where the HSV color thresholds corresponding to red photons are set to 0-10 and 160-179 in the H dimension, with S>100 and V>100; d) extracting N consecutive or evenly spaced frames from the UV imaging video, superimposing the N frames to obtain a scatter plot of the cumulative UV photon number distribution of the N frames, i.e., the initial UV discharge area; e) using a mean filter to smooth the UV discharge area on the UV photon distribution scatter plot, taking the average of the target pixel (i.e., the pixel where the photon is located) and the surrounding pixels, and then performing target pixel filling and isolated photon elimination processing; f) using a polygonal approximation method to enclose the UV discharge area based on the result obtained in step e), so as to regularize the irregular area; g) after obtaining the minimum circumscribed polygonal outline of each UV discharge region, calculating the outline area, outline perimeter, and maximum diagonal length of the corresponding UV discharge region, with the above parameters all counted in pixels, and counting the number of discharge photons within the outline region; h) After labeling the insulator image samples, the S2ANet target detection algorithm is used to establish a detection model for insulators in the UV imaging video. The insulator rotation target detection is performed on the UV imaging video frame, and the position information of the insulator circumscribed rectangle in the UV video frame is returned. i) The length ratio of the ultraviolet discharge area on the insulator surface is calculated by the length of the minimum circumscribed polygon of the discharge area along the insulator direction; the length ratio is calculated according to the following formula: Among them, L is the total length of the insulator, l i is the length of a discharge area along the insulator direction, and P is the ratio of the length of the ultraviolet discharge area on the insulator surface.

2. The method for locating the discharge area of an insulator according to claim 1, characterized in that: In step e), the mean filtering is to select a (2n+1)×(2n+1) window in the scatter plot, where n≥1, and the average filter value of the point (x, y) is calculated as follows: Among them, (x, y) is the spatial coordinate of the scattered point in the scatter plot, and the coordinate unit is pixel; P(i, j) is the original number of ultraviolet photons at a certain spatial position, and p(x, y) is the number of photons at this spatial position after mean filtering.

3. The method for locating the discharge area of an insulator according to claim 1, characterized in that: In step e), the target pixel filling is to perform a morphological closing operation on the filtered and smoothed image to fill the small cracks, discontinuities and holes between photons in the discharge area. For the image X and the structural element S, the closing operation formula is: The structural element S is a 3×3 kernel matrix, and the anchor point is located at the center of the matrix.

4. The method for locating the discharge area of an insulator according to claim 1, characterized in that: In step e), the isolated photon elimination process is to eliminate isolated small dots and burrs through an opening operation, thereby further eliminating the noise of the image X. The formula for the opening operation is: The structural element S is a 3×3 kernel matrix, and the anchor point is located at the center of the matrix.

5. The method for locating the discharge area of an insulator according to claim 1, characterized in that: In step h), a detection model of insulators in the ultraviolet imaging video is established by the following steps: h1) preprocessing the insulator image samples to form an image dataset that meets the standards; h2) expanding the dataset of the insulator image samples processed in step h1); using a labeling tool to mark the insulator region in the insulator image with a rectangular frame, and generating a label text corresponding to the image based on the position information; h3) Using the insulator image labeled in step h2) as a sample, a convolutional neural network is trained using a target detection algorithm, and an insulator detection model is obtained according to training adjustment parameters to obtain the insulator circumscribed rectangle.

6. An insulator discharge area positioning system based on solar-blind ultraviolet imaging, characterized in that: include: Insulator height acquisition unit: collects equipment data and obtains the insulator structure height; Insulator UV imaging video acquisition unit: performs UV testing on insulators in operation and acquires UV imaging videos of insulators; Video frame preprocessing unit: preprocesses UV imaging video frames using the HSV color model, converting RGB color space images into HSV color space images, where the HSV color thresholds corresponding to red photons are set to 0-10 and 160-179 in the H dimension, S>100, and V>100; Initial UV discharge area acquisition unit: extracts N consecutive or evenly spaced frames from the UV imaging video, superimposes the N frames, and obtains a scatter plot of the UV photon number distribution accumulated in the N frames, i.e., the initial UV discharge area; Ultraviolet discharge area smoothing processing unit: Use mean filtering on the ultraviolet photon number distribution scatter plot to achieve smoothing of the ultraviolet discharge area, take the average value of the target pixel (i.e. the pixel where the photon is located) and the surrounding pixels, and then perform target pixel filling and isolated photon elimination processing; Graphic regularization processing unit: Use polygonal approximation method to enclose the smoothed UV discharge area to regularize the irregular area; Discharge photon number calculation unit: After finding the minimum circumscribed polygon outline of each discharge area, the contour area, contour perimeter and maximum diagonal length of the corresponding discharge area are calculated. The above parameters are counted in pixels, and the number of discharge photons in the contour area is counted; Insulator detection model building unit: After annotating insulator image samples, the S2ANet target detection algorithm is used to build an insulator detection model in the UV imaging video. The unit then performs rotational target detection on the UV imaging video frame and returns the position information of the insulator's circumscribed rectangle in the UV video frame. Length ratio calculation unit: calculates the length ratio of the ultraviolet discharge area on the insulator surface through the length of the minimum circumscribed polygon of the discharge area along the insulator direction; In the length ratio calculation unit, the length ratio is calculated as follows: Among them, L is the total length of the insulator, l i is the length of a discharge area along the insulator direction, and P is the ratio of the length of the ultraviolet discharge area on the insulator surface.

7. The insulator discharge area positioning system according to claim 6, characterized in that: In the ultraviolet discharge area smoothing processing unit, the mean filtering is to select a (2n+1)×(2n+1) window in the scatter plot, n≥1, and the average filtering value calculation formula of the point (x, y) is as follows: Where (x, y) is the spatial coordinate of the scattered point in the scatter plot, and the coordinate unit is pixel; P(i, j) is the original number of ultraviolet photons at a certain spatial position, and p(x, y) is the number of photons at this spatial position after mean filtering; The target pixel filling is to perform a morphological closing operation on the filtered and smoothed image to fill the small cracks, discontinuities and holes between photons in the discharge area. For the image X and the structural element S, the closing operation formula is: Among them, the structural element S is a 3×3 kernel matrix, and the anchor point is located at the center of the matrix; The isolated photon elimination process is to eliminate isolated small dots and burrs through an opening operation, thereby further eliminating the noise of the image X. The formula for the opening operation is: The structural element S is a 3×3 kernel matrix, and the anchor point is located at the center of the matrix.

8. The insulator discharge area positioning system according to claim 6, characterized in that: In the insulator detection model building unit, the insulator detection model in the ultraviolet imaging video is established by the following steps: h1) preprocessing the insulator image samples to form an image dataset that meets the standards; h2) expanding the dataset of the insulator image samples processed in step h1); using a labeling tool to mark the insulator region in the insulator image with a rectangular frame, and generating a label text corresponding to the image based on the position information; h3) Using the insulator image labeled in step h2) as a sample, a convolutional neural network is trained using the S2ANet algorithm, and an insulator detection model is obtained according to the training adjustment parameters to obtain the insulator circumscribed rectangle.

Images