Discharge defect detection method and system based on ultraviolet imager

Abstract

The invention relates to a discharge defect detection method and system based on an ultraviolet imager, and belongs to the field of electric power detection. The method comprises the steps that incident light is divided into two paths, the two paths of incident light enter a visible imaging module and an ultraviolet imaging module respectively, and a visible light image and a solar blind ultraviolet image are obtained; fusing the visible light image and the solar blind ultraviolet image through an improved image fusion method to obtain a discharge defect image; a corrected photon number and a corrected light spot area are obtained through the discharge defect image, a discharge defect detection model is constructed through the corrected photon number and the corrected light spot area, discharge defect detection information is output, and the discharge defect detection information comprises discharge low-risk information, discharge medium-risk information and discharge high-risk information; and when the discharge risk information or the discharge high risk information is output, obtaining a discharge defect position through laser ranging, and feeding back a discharge defect elimination instruction to the terminal. According to the invention, rapid, simple and accurate discharge defect detection is realized, and the transverse space occupied by each module is reduced.

Description

Technical Field

[0001] The present invention belongs to the technical field of electric power detection, and in particular relates to a discharge defect detection method and system based on an ultraviolet imager. Background Art

[0002] During the construction and operation of power lines, transmission lines often produce defects such as burrs, deformation, and broken strands due to human or natural factors, which cause local field strength distortion under high voltage, thereby causing corona discharge. These corona discharges not only cause a large amount of power loss, but also lead to audible noise, electromagnetic interference, and even seriously threaten the safe operation of power lines. Therefore, it is crucial to detect corona discharge caused by defects in transmission lines in a timely and effective manner. Corona discharge detection is a fault detection method for high-voltage equipment. It mainly analyzes the changes in characteristic signals such as electricity, light, sound, and heat generated during the discharge process to accurately locate corona discharge faults. In recent years, ultraviolet imaging technology has been widely used in corona discharge detection. Compared with other detection methods, this technology has the advantages of high sensitivity, strong anti-interference ability, and simple operation, and can intuitively and effectively locate the discharge point. However, at present, ultraviolet imaging technology still has problems such as large equipment size, low detection efficiency, low accuracy, and a relatively single basis for the classification of discharge defects. Summary of the invention

[0003] In order to solve the above problems existing in the prior art, the present invention provides a discharge defect detection method and system based on an ultraviolet imager.

[0004] The purpose of the present invention can be achieved through the following technical solutions:

[0005] A discharge defect detection method based on an ultraviolet imager, the implementation of the discharge defect detection method comprising the following steps:

[0006] The ultraviolet imager divides the incident light into two paths through a beam splitter, and the two paths enter the visible imaging module and the ultraviolet imaging module respectively to obtain a visible light image and a day-blind ultraviolet image, wherein the visible imaging module and the ultraviolet imaging module are arranged vertically;

[0007] The discharge defect image is obtained by fusing the visible light image and the solar-blind ultraviolet image through an improved image fusion method;

[0008] Acquire a corrected photon number and a corrected spot area through the discharge defect image, construct a discharge defect detection model through the corrected photon number and the corrected spot area, and output discharge defect detection information, wherein the discharge defect detection information includes discharge low risk information, discharge medium risk information, and discharge high risk information;

[0009] When the medium-risk discharge information or the high-risk discharge information is output, the discharge defect position is acquired by laser ranging, and a discharge defect elimination instruction is fed back to the terminal.

[0010] Preferably, the visible imaging module includes an optical lens and a visible light charge coupled device, and the ultraviolet imaging module includes an ultraviolet imaging objective lens, a multi-layer solar blind filter, and an ultraviolet microchannel plate.

[0011] Preferably, the step of fusing the visible light image and the solar-blind ultraviolet image by an improved image fusion method to obtain a discharge defect image comprises:

[0012] Obtaining a reconstructed adjustment image by weighted averaging the visible light image and the solar-blind ultraviolet image;

[0013] The discharge defect image is obtained by fusing the reconstructed adjustment image through a dual-branch generator structure.

[0014] Preferably, the dual-branch generator structure includes a dense residual network and an attention subnet, the dense residual network consists of an encoder, a converter, and a decoder, the encoder consists of three convolutional layers, the converter consists of three densely connected blocks, and the decoder reconstructs low-level features through deconvolution; the attention subnet consists of three dual attention groups, two convolution blocks, and residual edges.

[0015] Preferably, the obtaining of the corrected photon number and the corrected spot area through the discharge defect image, constructing a discharge defect detection model through the corrected photon number and the corrected spot area, and outputting discharge defect detection information comprises:

[0016] Preprocessing the discharge defect image to obtain a spot binary discharge defect image and filtering out redundant information from the spot binary discharge defect image, wherein the preprocessing includes graying and threshold segmentation;

[0017] The number of photons is directly output by the UV imager, and the spot area is obtained by the spot binary discharge defect image;

[0018] Processing the photon number by the 10m photon number normalization principle to obtain the corrected photon number, processing the spot area by the corrected spot area parameter to obtain the corrected spot area, wherein the corrected spot area parameter includes a detection distance parameter and an instrument gain parameter;

[0019] The discharge defect detection model is constructed according to the correction photon number and the correction spot area.

[0020] Preferably, the threshold segmentation includes: obtaining the total number of pixels and grayscale of the discharge defect image, introducing an initial grayscale threshold to divide the discharge defect image into a target and a background, calculating the probability of target occurrence and the probability of background occurrence, calculating the target average grayscale and the background average grayscale according to the target occurrence and the background occurrence probability, and calculating the improved inter-class variance by assigning weights, and the calculation formula is ε 2 =ωP0(t)Q0(t) 2 +P1(t)Q1(t) 2 , where ε 2 To improve the inter-class variance, P0(t) is the probability of target occurrence, P1(t) is the probability of background occurrence, t is the initial grayscale threshold, Q0(t) is the target average grayscale, Q1(t) is the background average grayscale, ω is the weight, and its value is P0(t). When the improved inter-class variance is the largest, the corresponding initial grayscale threshold is the optimal segmentation threshold. The discharge defect image is segmented according to the optimal segmentation threshold to obtain the spot binary discharge defect image.

[0021] The redundant information filtering is achieved by using an opening operation and a closing operation based on a structural element as a basic operator. The opening operation is to first erode the image and then expand the image, and the closing operation is to first expand the image and then erode the image.

[0022] Preferably, the expression of the discharge defect detection model is: Among them, S z To correct the spot area, N 10m To correct the number of photons, when the discharge defect detection model takes a value of 0, the discharge low risk information is output; when the discharge defect detection model takes a value of 1, the discharge medium risk information is output; and when the discharge defect detection model takes a value of 2, the discharge high risk information is output.

[0023] A discharge defect detection system based on an ultraviolet imager, used to execute the above-mentioned discharge defect detection method, comprising an image acquisition module, a model building module, and an instruction output module;

[0024] The image acquisition module is used for the ultraviolet imager to split the incident light into two paths through a beam splitter, respectively entering the visible imaging module and the ultraviolet imaging module to acquire a visible light image and a day-blind ultraviolet image, wherein the visible imaging module and the ultraviolet imaging module are arranged vertically; the discharge defect image is obtained by fusing the visible light image and the day-blind ultraviolet image through an improved image fusion method;

[0025] The model building module is used to obtain the corrected photon number and the corrected spot area through the discharge defect image, build a discharge defect detection model through the corrected photon number and the corrected spot area, and output discharge defect detection information, wherein the discharge defect detection information includes discharge low risk information, discharge medium risk information, and discharge high risk information;

[0026] The instruction output module is used to obtain the discharge defect position through laser ranging when outputting the discharge medium risk information or the discharge high risk information, and to feed back the discharge defect elimination instruction to the terminal.

[0027] An electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the above-mentioned discharge defect detection method is implemented when the processor executes the program.

[0028] A storage medium comprising computer executable instructions, wherein the computer executable instructions are used to perform the above-mentioned discharge defect detection method when executed by a computer processor.

[0029] The beneficial effects of the present invention are:

[0030] (1) By vertically arranging the visible imaging module and the ultraviolet imaging module, the device is small and light, can be held in one hand, and is easy to carry and operate on site;

[0031] (2) The discharge defect image is obtained by fusing the visible light image and the day-blind ultraviolet image through an improved image fusion method, thereby avoiding problems such as feature loss, loss of visible light texture and contour features, halo artifacts, low image resolution, and blurred details that may occur during the image fusion process;

[0032] (3) The discharge defect level is comprehensively judged by correcting the number of photons and the area of ​​the corrected spot, and the result is more accurate;

[0033] (4) The problem of insufficient sensitivity of traditional UV imagers in detecting weak UV rays is solved by using multi-layer solar-blind filters and UV microchannel plates, ensuring that weak discharge phenomena such as corona and arc can be accurately captured. At the same time, its good out-of-band suppression performance and anti-sunlight interference ability enable the equipment to perform accurate detection all day and night, avoiding the influence of sunlight interference on the detection results. The UV microchannel plate can detect very weak UV signals, has high sensitivity, and has a short response time, while being able to suppress noise; the multi-layer solar-blind filter can ensure that UV light in a specific band passes through with high transmittance, effectively suppressing light signals outside the passband, and improving the selectivity and signal-to-noise ratio of the system;

[0034] (5) The reconstructed adjustment image is obtained by weighted averaging the visible light image and the day-blind ultraviolet image, and the discharge defect image is obtained by fusing the reconstructed adjustment image through the dual-branch generator structure, which can extract higher-level features, so that shallow features can be directly combined with deep features to prevent the loss of shallow features, and can convert image feature vectors from the source domain to the target domain, and can reconstruct low-level features to improve the accuracy of target recognition;

[0035] (6) The inter-class variance is improved by assigning weights to the calculation and using adaptive weighting to better process image details. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to facilitate understanding by those skilled in the art, the present invention is further described below with reference to the accompanying drawings.

[0037] Figure 1 The present invention is a flow chart of the steps of a discharge defect detection method based on an ultraviolet imager. DETAILED DESCRIPTION

[0038] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation methods, structures, features and effects of the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments.

[0039] The working principle and use process of the present invention:

[0040] See also Figure 1 , a discharge defect detection method based on ultraviolet imager, comprising:

[0041] S1: The ultraviolet imager divides the incident light into two paths through a beam splitter, and the two paths enter the visible imaging module and the ultraviolet imaging module respectively to obtain a visible light image and a solar-blind ultraviolet image, wherein the visible imaging module and the ultraviolet imaging module are arranged vertically;

[0042] S2: fusing the visible light image and the solar-blind ultraviolet image by an improved image fusion method to obtain a discharge defect image;

[0043] S3: obtaining a corrected photon number and a corrected spot area through the discharge defect image, constructing a discharge defect detection model through the corrected photon number and the corrected spot area, and outputting discharge defect detection information, wherein the discharge defect detection information includes discharge low risk information, discharge medium risk information, and discharge high risk information;

[0044] S4: When the medium-risk discharge information or the high-risk discharge information is output, the discharge defect position is obtained by laser ranging, and a discharge defect elimination instruction is fed back to the terminal.

[0045] In this embodiment, the visible imaging module includes an optical lens and a visible light charge coupled device, and the ultraviolet imaging module includes an ultraviolet imaging objective lens, a multilayer solar blind filter, and an ultraviolet microchannel plate. The visible light charge coupled device can convert the received visible light signal into an electrical signal, thereby realizing the imaging function to obtain the visible light image. The multilayer solar blind filter is used to make incident light of different wavelengths reflect and transmit multiple times to form an interference pattern. By adjusting the refractive index and thickness, ultraviolet light of specific wavelengths is mutually reinforced during the interference process and passes through the multilayer solar blind filter, while light of other wavelengths is mutually offset and suppressed. The ultraviolet microchannel plate is used to emit photoelectrons when ultraviolet light reaches the photocathode. The photoelectrons are accelerated into the channel of the ultraviolet microchannel plate under the action of the electric field and multiplied step by step to form a large number of secondary electrons. The secondary electrons are accelerated and bombarded onto the fluorescent screen, causing the fluorescent material to emit light, thereby forming an enhanced solar blind ultraviolet image on the fluorescent screen. The problem of insufficient sensitivity of traditional UV imagers in weak UV detection is solved by using multi-layer solar-blind filters and UV microchannel plates, ensuring that weak discharge phenomena such as corona and arc can be accurately captured. At the same time, its good out-of-band suppression performance and anti-sunlight interference ability enable the equipment to perform accurate detection all day and night, avoiding the influence of sunlight interference on the detection results. The UV microchannel plate can detect very weak UV signals, has high sensitivity, and has a short response time, while being able to suppress noise; the multi-layer solar-blind filter can ensure that UV light in a specific band passes through with high transmittance, effectively suppressing light signals outside the passband, and improving the selectivity and signal-to-noise ratio of the system.

[0046] In this embodiment, the discharge defect image is obtained by fusing the visible light image and the day-blind ultraviolet image through an improved image fusion method, which can be specifically implemented through the following steps:

[0047] S201: Obtain a reconstructed adjustment image by weighted averaging the visible light image and the day-blind ultraviolet image, wherein the weighted average is mathematically described as F(x, y) = αV(x, y) + (1-α)I(x, y), wherein F(x, y) is the reconstructed adjustment image, V(x, y) is the visible light image, I(x, y) is the day-blind ultraviolet image, and α is a weight coefficient, which ranges from (0 to 1) and is 0.6 in this embodiment;

[0048] S202: The discharge defect image is obtained by fusing the reconstructed adjustment image through a dual-branch generator structure, so as to avoid problems such as feature loss, loss of texture and contour features of visible light, halo artifacts, low image resolution, blurred details, etc. that may occur during the image fusion process.

[0049] In this embodiment, the discharge defect image is obtained by fusing the reconstructed adjustment image through a dual-branch generator structure, and higher-level features are effectively extracted, so that shallow features can be directly combined with deep features to prevent shallow feature loss, and the image feature vector can be converted from the source domain to the target domain, and low-level features can be reconstructed to improve the target recognition accuracy. Specifically, it can be implemented by the following steps:

[0050] The dual-branch generator structure includes a dense residual network and an attention subnet. The dense residual network consists of an encoder, a converter, and a decoder. The encoder consists of three convolutional layers and can extract higher-level features. The converter consists of three densely connected blocks, so that shallow features can be directly combined with deep features to prevent the loss of shallow features, and can convert image feature vectors from the source domain to the target domain. The decoder reconstructs low-level features through deconvolution; the attention subnet consists of three dual attention groups, two convolutional blocks, and residual edges.

[0051] In this embodiment, the corrected photon number and the corrected spot area are obtained through the discharge defect image, a discharge defect detection model is constructed through the corrected photon number and the corrected spot area, and discharge defect detection information is output, which can be specifically implemented through the following steps:

[0052] S301: pre-processing the discharge defect image to obtain a spot binary discharge defect image and filtering out redundant information from the spot binary discharge defect image, wherein the pre-processing includes graying and threshold segmentation;

[0053] S302: directly outputting the number of photons through the ultraviolet imager, and obtaining the spot area through the spot binary discharge defect image. The expression of the spot binary discharge defect image is: Among them, g(x,y) is the binary discharge defect image of the light spot, and the calculation formula of the light spot area is: Among them, S is the spot area, M×N is the size of the spot binary discharge defect image;

[0054] S303: Processing the photon number by the 10m photon number normalization principle to obtain the corrected photon number, processing the spot area by the correction spot area parameter to obtain the corrected spot area, the correction spot area parameter includes a detection distance parameter and an instrument gain parameter, the detection distance parameter is 5 meters, the instrument gain parameter is 60%, and the mathematical description of the 10m photon number normalization principle is: Among them, N 10m To correct the number of photons, N is the number of photons, and L is the detection distance in meters;

[0055] S304: constructing the discharge defect detection model according to the corrected photon number and the corrected spot area. The expression of the discharge defect detection model is: Among them, S Z To correct the spot area, N 10m To correct the number of photons, when the discharge defect detection model takes a value of 0, the discharge low risk information is output; when the discharge defect detection model takes a value of 1, the discharge medium risk information is output; when the discharge defect detection model takes a value of 2, the discharge high risk information is output. When both levels are met at the same time, the higher level is prioritized. For example, when the correction spot area is measured to be greater than 30,000 and the correction photon number is less than 8,000 at the same time, the discharge high risk information is output.

[0056] In this embodiment, the discharge defect image is preprocessed to obtain a spot binary discharge defect image and redundant information is filtered out of the spot binary discharge defect image, which can be specifically implemented by the following steps:

[0057] The threshold segmentation includes: obtaining the total number of pixels and grayscale of the discharge defect image, introducing an initial grayscale threshold to divide the discharge defect image into a target and a background, and calculating the probability of the target appearing and the probability of the background appearing. The calculation formula is: Among them, P0(t) is the probability of target appearance, P1(t) is the probability of background appearance, t is the initial gray level threshold, N i is the number of pixels when the gray level is i, N is the total number of pixels, k is the highest gray level, and the target average gray level and the background average gray level are calculated according to the target occurrence probability and the background occurrence probability. The calculation formula is: Among them, Q0(t) is the target average gray level, Q1(t) is the background average gray level;

[0058] By assigning weights to improve the inter-class variance, adaptive weighting is used to better process image details. The calculation formula is ε 2 =ωP o (t)Q o (t) 2 +P1(t)Q1(t) 2 , where ε 2 is to improve the inter-class variance, ω is the weight, and its value is P0(t). When the improved inter-class variance is maximized, the corresponding initial gray level threshold is the optimal segmentation threshold. The discharge defect image is segmented according to the optimal segmentation threshold to obtain the spot binary discharge defect image.

[0059] The redundant information filtering is achieved by using an open operation and a closed operation based on a structural element as a basic operator. The structural element is a small image area set according to actual needs. In this embodiment, a circular shape similar to a corona discharge light spot is selected. The open operation is image erosion first and then image expansion. The closed operation is image expansion first and then image erosion. The mathematical description of image erosion is: Among them, A is the binary discharge defect image of the light spot, B is the structural element, (x, y) is the pixel point in the binary discharge defect image of the light spot, For the erosion operation, the mathematical description of the image dilation is in, is the dilation operation, Is an empty set.

[0060] A discharge defect detection system based on an ultraviolet imager includes an image acquisition module, a model building module, and an instruction output module;

[0061] The image acquisition module is used for the ultraviolet imager to split the incident light into two paths through a beam splitter, respectively entering the visible imaging module and the ultraviolet imaging module to acquire a visible light image and a day-blind ultraviolet image, wherein the visible imaging module and the ultraviolet imaging module are arranged vertically; the discharge defect image is obtained by fusing the visible light image and the day-blind ultraviolet image through an improved image fusion method;

[0062] The model building module is used to obtain the corrected photon number and the corrected spot area through the discharge defect image, build a discharge defect detection model through the corrected photon number and the corrected spot area, and output discharge defect detection information, wherein the discharge defect detection information includes discharge low risk information, discharge medium risk information, and discharge high risk information;

[0063] The instruction output module is used to obtain the discharge defect position through laser ranging when outputting the discharge medium risk information or the discharge high risk information, and to feed back the discharge defect elimination instruction to the terminal.

[0064] The computer storage medium of the embodiment of the present invention may adopt any combination of one or more computer-readable media. The computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples (non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, device or device.

[0065] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, which carry computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0066] The program code included on the computer readable medium can be transmitted with any appropriate medium, including but not limited to wireless, electric wire, optical cable, RF, etc., or any suitable combination of the above. The computer program code for performing the operation of the present invention can be written in one or more programming languages ​​or their combinations, and the programming language includes object-oriented programming languages-such as Java, Smalltalk, C++, and also includes conventional procedural programming languages-such as "C" language or similar programming languages. The program code can be executed completely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer and partially on the remote computer, or completely on the remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (for example, using an Internet service provider to connect through the Internet).

[0067] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technical personnel in this field can make some changes or modify the technical contents disclosed above into equivalent embodiments without departing from the scope of the technical solution of the present invention. However, any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present invention without departing from the content of the technical solution of the present invention still fall within the scope of the technical solution of the present invention.

Claims

1. A discharge defect detection method based on ultraviolet imager, characterized in that: The implementation of the discharge defect detection method comprises the following steps: The ultraviolet imager divides the incident light into two paths through a beam splitter, and the two paths enter the visible imaging module and the ultraviolet imaging module respectively to obtain a visible light image and a day-blind ultraviolet image, wherein the visible imaging module and the ultraviolet imaging module are arranged vertically; The discharge defect image is obtained by fusing the visible light image and the solar-blind ultraviolet image through an improved image fusion method; Acquire a corrected photon number and a corrected spot area through the discharge defect image, construct a discharge defect detection model through the corrected photon number and the corrected spot area, and output discharge defect detection information, wherein the discharge defect detection information includes discharge low risk information, discharge medium risk information, and discharge high risk information; When the medium-risk discharge information or the high-risk discharge information is output, the discharge defect position is acquired by laser ranging, and a discharge defect elimination instruction is fed back to the terminal.

2. The discharge defect detection method according to claim 1, characterized in that: The visible imaging module includes an optical lens and a visible light charge coupled device, and the ultraviolet imaging module includes an ultraviolet imaging objective lens, a multi-layer solar blind filter, and an ultraviolet microchannel plate.

3. The discharge defect detection method according to claim 1, characterized in that: The step of fusing the visible light image and the solar-blind ultraviolet image by an improved image fusion method to obtain a discharge defect image comprises: Obtaining a reconstructed adjustment image by weighted averaging the visible light image and the solar-blind ultraviolet image; The discharge defect image is obtained by fusing the reconstructed adjustment image through a dual-branch generator structure.

4. The discharge defect detection method according to claim 3, characterized in that: The dual-branch generator structure includes a dense residual network and an attention subnet. The dense residual network consists of an encoder, a converter, and a decoder. The encoder consists of three convolutional layers, the converter consists of three densely connected blocks, and the decoder reconstructs low-level features through deconvolution. The attention subnet consists of three dual attention groups, two convolutional blocks, and residual edges.

5. The discharge defect detection method according to claim 1, characterized in that: The step of acquiring the corrected photon number and the corrected spot area through the discharge defect image, constructing a discharge defect detection model through the corrected photon number and the corrected spot area, and outputting discharge defect detection information comprises: Preprocessing the discharge defect image to obtain a spot binary discharge defect image and filtering out redundant information from the spot binary discharge defect image, wherein the preprocessing includes graying and threshold segmentation; The number of photons is directly output by the UV imager, and the spot area is obtained by the spot binary discharge defect image; Processing the photon number by the 10m photon number normalization principle to obtain the corrected photon number, processing the spot area by the corrected spot area parameter to obtain the corrected spot area, wherein the corrected spot area parameter includes a detection distance parameter and an instrument gain parameter; The discharge defect detection model is constructed according to the correction photon number and the correction spot area.

6. The discharge defect detection method according to claim 5, characterized in that: The threshold segmentation includes: obtaining the total number of pixels and grayscale of the discharge defect image, introducing an initial grayscale threshold to divide the discharge defect image into a target and a background, calculating the target appearance probability and the background appearance probability, calculating the target average grayscale and the background average grayscale according to the target appearance probability and the background appearance probability, and calculating the improved inter-class variance by assigning weights, and the calculation formula is ε2=ωP0(t)Q0(t) 2 +P1(t)Q1(t) 2 , where ε 2 To improve the inter-class variance, P0(t) is the probability of target occurrence, P1(t) is the probability of background occurrence, t is the initial grayscale threshold, Q0(t) is the target average grayscale, Q1(t) is the background average grayscale, ω is the weight, and its value is P0(t). When the improved inter-class variance is the largest, the corresponding initial grayscale threshold is the optimal segmentation threshold. The discharge defect image is segmented according to the optimal segmentation threshold to obtain the spot binary discharge defect image. The redundant information filtering is achieved by using an opening operation and a closing operation based on a structural element as a basic operator. The opening operation is to first erode the image and then expand the image, and the closing operation is to first expand the image and then erode the image.

7. The discharge defect detection method according to claim 5, characterized in that: The expression of the discharge defect detection model is: Among them, S Z To correct the spot area, N 10m To correct the number of photons, when the discharge defect detection model takes a value of 0, the discharge low risk information is output; when the discharge defect detection model takes a value of 1, the discharge medium risk information is output; and when the discharge defect detection model takes a value of 2, the discharge high risk information is output.

8. A discharge defect detection system based on ultraviolet imager, characterized in that: The system is applied to the discharge defect detection method as described in any one of claims 1 to 7, comprising an image acquisition module, a model building module, and an instruction output module; The image acquisition module is used for the ultraviolet imager to split the incident light into two paths through a beam splitter, respectively entering the visible imaging module and the ultraviolet imaging module to acquire a visible light image and a day-blind ultraviolet image, wherein the visible imaging module and the ultraviolet imaging module are arranged vertically; the discharge defect image is obtained by fusing the visible light image and the day-blind ultraviolet image through an improved image fusion method; The model building module is used to obtain the corrected photon number and the corrected spot area through the discharge defect image, build a discharge defect detection model through the corrected photon number and the corrected spot area, and output discharge defect detection information, wherein the discharge defect detection information includes discharge low risk information, discharge medium risk information, and discharge high risk information; The instruction output module is used to obtain the discharge defect position through laser ranging when outputting the discharge medium risk information or the discharge high risk information, and to feed back the discharge defect elimination instruction to the terminal.

9. An electronic 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 program, the discharge defect detection method as described in any one of claims 1-7 is implemented.

10. A storage medium containing computer executable instructions, characterized in that: The computer executable instructions are used to perform the discharge defect detection method as claimed in any one of claims 1 to 7 when executed by a computer processor.

Images