Deterioration identification method and system based on temperature difference adjustment of infrared thermal image of insulator string

Abstract

The invention discloses a degradation identification method and system based on temperature difference adjustment of infrared thermal images of insulator strings. The method comprises the following steps: obtaining an infrared thermal image of an insulator string; extracting the area where the insulator string is located in the image and dividing the area where each insulator element is located; obtaining the highest and lowest values ​​of brightness in the area where each insulator element is located; calculating the temperature corresponding to the highest and lowest values ​​of brightness in the area where the insulator element is located and obtaining the optimal temperature measurement range in combination with the current temperature, adjusting the insulator string image, and obtaining the optimal insulator string image; calculating the abnormal area in the area where the insulator string is located, performing adaptive connected domain analysis on the abnormal area, optimizing the connected domain analysis through conditional random field CRF, merging adjacent abnormal areas, and obtaining a complete degraded area; the invention greatly reduces the time required for staff to inspect a large number of insulator strings.

Description

Technical Field

[0001] The present invention relates to the technical field of deteriorated insulator identification, and in particular to a deterioration identification method and system based on temperature difference adjustment of infrared thermal images of insulator strings. Background Art

[0002] An insulator string refers to an assembly of two or more insulator elements combined together to flexibly suspend the conductor. An insulator string is a protective device with fixing and operation requirements, used to suspend the conductor and insulate the conductor from the pole tower and the ground; when inspecting outdoor insulator strings, each time a valid infrared thermal image is obtained, the temperature difference needs to be manually adjusted so that the brightness on the insulator string can clearly distinguish the temperature difference, thereby finding the deteriorated insulators in the darker area. However, when inspecting a large number of insulator strings, the time required is greatly increased, resulting in a decrease in the efficiency of identifying deteriorated insulators. Summary of the invention

[0003] In view of the deficiencies in the prior art, the present invention provides a degradation identification method and system based on temperature difference adjustment of infrared thermal images of insulator strings, which aims to solve the problems in the background technology.

[0004] To achieve the above object, the present invention provides the following technical solution: a degradation identification method based on temperature difference adjustment of infrared thermal images of insulator strings, comprising the following steps:

[0005] Step S1: acquiring an insulator string image; the insulator string image is an infrared thermal image;

[0006] Step S2: Scanning the insulator string image to obtain a pixel matrix of the insulator string image;

[0007] Step S3: Filling the pixel points in the pixel matrix whose pixel value deviation is greater than a preset value, and filtering the insulator string image according to the filled pixel matrix;

[0008] Step S4: using the Sobel operator to obtain edge features of the insulator string image after filtering;

[0009] Step S5: obtaining the real-time insulator string shape, comparing the real-time insulator string shape with the edge features obtained in step S4, and obtaining the region where the insulator string is located in the insulator string image after filtering;

[0010] Step S6: dividing the area where the insulator string is located according to the real-time insulator element appearance, and obtaining the area where each insulator element is located in the insulator string image after filtering;

[0011] Step S7: performing mean value processing on the brightness of the pixels in the area where each insulator element is located, and obtaining the highest value and the lowest value of the brightness in the area where each insulator element is located according to the mean value of the brightness of the pixels in the area where each insulator element is located;

[0012] Step S8: Calculate the temperature corresponding to the highest and lowest brightness values ​​in the area where the insulator element is located;

[0013] Step S9: Calculate the temperature difference in the area where the insulator element is located according to the temperatures corresponding to the highest and lowest brightness values ​​in the area where the insulator element is located, obtain the optimal temperature measurement range in combination with the current temperature, and adjust the insulator string image to the optimal temperature measurement range through the optimal temperature measurement range to obtain the optimal insulator string image;

[0014] Step S10: Obtain the area where the insulator string is located in the optimal insulator string image, calculate the abnormal area in the area where the insulator string is located through a multi-head attention mechanism and a multi-scale convolutional neural network, perform adaptive connected domain analysis on the abnormal area, and optimize the connected domain analysis through the conditional random field CRF, merge adjacent abnormal areas, obtain a complete degraded area, and highlight it.

[0015] Furthermore, the specific process of step S10 is as follows:

[0016] Obtain the area where the insulator string is located in the optimal insulator string image, and calculate the area where the insulator elements with abnormal temperature distribution are located in the area where the insulator string is located through a multi-head attention mechanism;

[0017] The area where the insulator element with abnormal temperature distribution is located is combined with its own temperature gradient information to identify the part with significant temperature change, that is, the abnormal area;

[0018] The multi-scale convolutional neural network MSCNN is used to process the area where the insulator string is located in the optimal insulator string image, and the parts with different abnormal temperatures in the area where the insulator string is located are extracted, which is expressed as:

[0019] ;

[0020] In the formula, Represents the parts with different abnormal temperatures extracted by the multi-scale convolutional network MSCNN, It is Convolution operations at different scales, The position of the part with different abnormal temperature The temperature value at Represents the total number of scales used by the multi-scale convolutional network MSCNN;

[0021] By calculating the temperature gradient of each pixel in the parts with different abnormal temperatures, the area with abnormal temperature change, i.e., the abnormal area, is determined; the temperature gradient calculation formula is:

[0022] ;

[0023] In the formula, The position of the part showing different abnormal temperatures The temperature gradient at It means to find partial derivatives;

[0024] Perform adaptive connected domain analysis on abnormal areas, optimize the connected domain analysis through conditional random field (CRF), merge adjacent abnormal areas, obtain complete degraded areas, and highlight them;

[0025] The connected domain analysis is optimized through the conditional random field CRF, which is expressed as:

[0026] ;

[0027] In the formula, is the objective function of the conditional random field, Indicates abnormal area; Labels indicating degraded areas; and represents any two pixel positions in the abnormal region; express The local characteristics of express and The joint features between .

[0028] Furthermore, the specific process of step S5 is as follows:

[0029] Step S5.1: obtaining the real-time insulator string outline, scanning the real-time insulator string outline, and obtaining the real-time insulator string outline parameters;

[0030] Step S5.2: Analyze the edge features obtained in step S4, obtain feature parameters of the edge features, and compare the feature parameters of the edge features with the real-time insulator string shape parameters;

[0031] Step S5.3: Filter out the area where the difference between the feature parameters of the edge features in the filtered insulator string image and the real-time insulator string shape parameters is the lowest, which is the area where the insulator string is located in the filtered insulator string image.

[0032] Furthermore, the specific process of step S6 is as follows:

[0033] Step S6.1: obtaining real-time insulator component shape parameters;

[0034] Step S6.2: Divide the area where the insulator string is located according to the real-time insulator element shape parameters to obtain the number of insulator elements in the area where the insulator string is located, and mark the insulator elements from top to bottom to obtain the area where each insulator element is located in the insulator string image after filtering;

[0035] Among them, the area where the insulator string is located is divided according to the real-time insulator element shape parameters as follows:

[0036] ;

[0037] In the formula, The real-time insulator component shape parameters and the first The similarity of the features; The real-time insulator component shape parameters The first feature characteristic index value; The first The first feature characteristic index value; is the total number of characteristic index values.

[0038] Furthermore, the specific process of step S7 is as follows:

[0039] Step S7.1: Obtain the brightness value of the pixel points in the area where each insulator element is located;

[0040] Step S7.2: performing mean value processing on the brightness values ​​of the pixels in the area where each insulator element is located, respectively, to obtain the brightness mean value of the pixels in the area where each insulator element is located;

[0041] Step S7.3: Obtain the highest and lowest values ​​of brightness in the area where each insulator element is located according to the average brightness of the pixel points in the area where each insulator element is located, expressed as:

[0042] ;

[0043] ;

[0044] In the formula, is the number of insulator elements; Indicates the average brightness of the pixels in the area where the insulator element is located; Indicates the preset value of the number of insulator elements.

[0045] Furthermore, the specific process of step S8 is as follows:

[0046] Step S8.1: obtaining the temperature acquisition range of the initially acquired insulator string image;

[0047] Step S8.2: Based on the temperature acquisition range of the initially acquired insulator string image and the brightness range, the temperature corresponding to the brightness value within the brightness range is obtained, which is expressed as:

[0048] ;

[0049] In the formula, and They respectively represent the highest temperature and the lowest temperature within the temperature acquisition range of the initially acquired insulator string image; represents the temperature adjustment factor;

[0050] Step S8.3: Match the highest and lowest brightness values ​​in the region where the insulator element is located with the temperature corresponding to the brightness value, and obtain the temperature corresponding to the highest and lowest brightness values ​​in the region where the insulator element is located, expressed as:

[0051] ;

[0052] ;

[0053] In the formula, and They respectively represent the temperatures corresponding to the highest and lowest brightness values ​​in the area where the insulator element is located.

[0054] Furthermore, the specific process of step S9 is as follows:

[0055] Step S9.1: Calculate the temperature difference in the area where the insulator element is located according to the temperatures corresponding to the highest and lowest brightness values ​​in the area where the insulator element is located;

[0056] Step S9.2: Compare the temperature difference in the area where the insulator element is located with the current temperature difference to judge the current temperature measurement range;

[0057] Step S9.3: When it is determined that the current temperature measurement range does not meet the requirements, the highest temperature in the area where the insulator element is located is compared with the highest temperature in the current temperature measurement range, and the lowest temperature in the area where the insulator element is located is compared with the lowest temperature in the current temperature measurement range, to obtain the position where the temperature in the area where the insulator element is located is within the current temperature measurement range;

[0058] Step S9.4: adjusting the current temperature measurement range according to the position of the temperature in the area where the insulator element is located within the current temperature measurement range;

[0059] Step S9.5: recalculate the adjusted current temperature measurement range to obtain the optimal temperature measurement range, adjust the insulator string image to the optimal temperature measurement range through the optimal temperature measurement range, and obtain the optimal insulator string image.

[0060] The deterioration identification system based on the temperature difference adjustment of the infrared thermal image of the insulator string includes:

[0061] An image acquisition module is used to acquire an image of an insulator string; the image of the insulator string is an infrared thermal image;

[0062] A scanning module, used for scanning the insulator string image to obtain a pixel matrix of the insulator string image;

[0063] A preprocessing module, used to fill the pixel points whose pixel value deviation is greater than a preset value in the pixel matrix, and filter the insulator string image according to the filled pixel matrix;

[0064] A feature acquisition module, used for acquiring edge features of the insulator string image after filtering using a Sobel operator;

[0065] A comparison module is used to obtain the real-time insulator string shape, compare the real-time insulator string shape with the obtained edge features, and obtain the area where the insulator string is located in the insulator string image after filtering;

[0066] A division module is used to divide the area where the insulator string is located according to the real-time insulator element appearance, and obtain the area where each insulator element is located in the insulator string image after filtering;

[0067] A brightness mean module is used to perform mean processing on the brightness of the pixels in the area where each insulator element is located, and obtain the highest and lowest brightness values ​​in the area where each insulator element is located according to the brightness mean of the pixels in the area where each insulator element is located;

[0068] A temperature calculation module is used to calculate the temperature corresponding to the highest and lowest brightness values ​​in the area where the insulator element is located;

[0069] A temperature range determination module is used to calculate the temperature difference in the area where the insulator element is located according to the temperatures corresponding to the highest and lowest brightness values ​​in the area where the insulator element is located, and to obtain the optimal temperature measurement range in combination with the current temperature. The insulator string image is adjusted to the optimal temperature measurement range through the optimal temperature measurement range to obtain the optimal insulator string image;

[0070] The degradation identification module is used to obtain the area where the insulator string is located in the optimal insulator string image, calculate the abnormal area in the area where the insulator string is located through the multi-head attention mechanism and multi-scale convolutional neural network, perform adaptive connected domain analysis on the abnormal area, and optimize the connected domain analysis through the conditional random field CRF, merge adjacent abnormal areas, obtain the complete degraded area, and highlight it.

[0071] Compared with the existing technology, the present invention has the following beneficial effects:

[0072] (1) The present invention effectively enhances the accuracy of the image by using the Sobel operator to extract edge features, ensuring accurate identification of the insulator string position; the extraction of edge features not only improves the clarity of the image, but also provides a reliable basis for subsequent regional division and brightness analysis, avoiding the influence of background interference and noise, so that degradation identification can focus on the target area. This precise regional segmentation provides strong support for subsequent temperature measurement and analysis, and improves the efficiency and accuracy of degradation identification; the accuracy of temperature measurement is further optimized through the analysis of brightness difference and the dynamic adjustment of the temperature acquisition range. By calculating the brightness difference, the present invention can automatically adjust the temperature acquisition range to ensure that the temperature difference range of the image is always in the most suitable range when different infrared thermal images are obtained.

[0073] (2) Through the self-attention mechanism and temperature gradient analysis, the present invention can accurately identify local abnormal areas after image segmentation and automatically focus on areas with significant temperature changes. The self-attention mechanism can weightedly calculate the temperature value of each position based on the global context information, avoiding the problems of misidentification and missed identification that may occur in traditional methods. In addition, the combination of the multi-scale convolutional neural network MSCNN and local area temperature gradient analysis not only enhances the ability to recognize temperature anomalies at different levels, but also can capture tiny temperature fluctuations and large-scale temperature changes, thereby improving the accuracy of degraded area detection.

[0074] (3) By comparing the shape parameters of the insulator elements with the edge features of the insulator string image, the number of insulator elements in the insulator string in the image can be quickly obtained, which is convenient for obtaining the temperatures of multiple insulator elements and arranging the insulator elements in the insulator string. The status and changes of each element can be easily tracked and analyzed, providing strong support for subsequent insulator string monitoring and maintenance.

[0075] (4) The present invention performs mean processing on the areas where multiple insulator elements are located, thereby preventing the low brightness of some insulator elements due to surface dirt and obstruction of the insulator elements, thereby affecting subsequent judgment. By comparing the brightness mean values ​​of different insulator element areas, the surface state differences of each insulator element can be preliminarily judged. By comparing the highest and lowest values, the overall state difference and uniformity of the insulator string can be evaluated, providing a decision-making basis for subsequent maintenance and replacement. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] Figure 1 The figure is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0077] like Figure 1 As shown, the present invention provides a technical solution: a degradation identification method based on temperature difference adjustment of infrared thermal images of insulator strings:

[0078] Step S1: Acquire an insulator string image; the insulator string image is an infrared thermal image.

[0079] Step S2: Scan the insulator string image to obtain a pixel matrix of the insulator string image.

[0080] Step S3: Filling the pixel points in the pixel matrix whose pixel value deviation is greater than a preset value, and filtering the insulator string image according to the filled pixel matrix.

[0081] Step S4: using the Sobel operator to obtain edge features of the filtered insulator string image.

[0082] Step S5: obtaining the real-time insulator string shape, comparing the real-time insulator string shape with the edge features obtained in step S4, and obtaining the region where the insulator string is located in the insulator string image after filtering.

[0083] Step S6: Divide the area where the insulator string is located according to the real-time insulator element appearance, and obtain the area where each insulator element is located in the insulator string image after filtering.

[0084] Step S7: performing mean value processing on the brightness of the pixels in the area where each insulator element is located, and obtaining the highest and lowest values ​​of the brightness in the area where each insulator element is located according to the mean value of the brightness of the pixels in the area where each insulator element is located.

[0085] Step S8: Calculate the temperature corresponding to the highest and lowest brightness values ​​in the area where the insulator element is located.

[0086] Step S9: Calculate the temperature difference in the area where the insulator element is located according to the temperatures corresponding to the highest and lowest brightness values ​​in the area where the insulator element is located, and obtain the optimal temperature measurement range in combination with the current temperature. Adjust the insulator string image to the optimal temperature measurement range through the optimal temperature measurement range to obtain the optimal insulator string image.

[0087] Step S10: Obtain the area where the insulator string is located in the optimal insulator string image, calculate the abnormal area in the area where the insulator string is located through a multi-head attention mechanism and a multi-scale convolutional neural network, perform adaptive connected domain analysis on the abnormal area, and optimize the connected domain analysis through the conditional random field CRF, merge adjacent abnormal areas, obtain a complete degraded area, and highlight it.

[0088] Those skilled in the art can understand that by filtering the thermal imaging image, the thermal imaging image is prevented from being interfered by other heat source information during the acquisition process, which makes it easy to introduce more noise into the image. The filtered image is processed to obtain the edge features of the image and compared with the shape of the insulator string to determine the area where the insulator string is located in the thermal imaging image. The number and position of the insulator elements in the insulator string can be obtained by analyzing the insulator string. In the thermal imaging image, different temperatures make the brightness of the corresponding pixels different. Therefore, by analyzing the brightness values ​​of the areas where multiple insulator elements are located, different temperatures of the insulator elements can be quickly obtained, and the current temperature measurement can be judged by the temperature between the insulator elements. If the current temperature measurement range is too large, the current temperature is adjusted based on the temperature between the insulator elements.

[0089] The specific process of step S2 is:

[0090] Step S2.1: Scan each pixel in the insulator string image to construct a pixel matrix.

[0091] Step S2.2: Assume that the lower left corner of the pixel matrix is ​​the origin, and construct an x-axis and y-axis coordinate system with the origin along the length direction of the pixel matrix as the x-axis direction and the origin along the width direction of the pixel matrix as the y-axis direction, and the scale of two adjacent pixel points on the coordinate axis is 1.

[0092] It can be understood by those skilled in the art that by scanning the insulator string image and obtaining information about each pixel, a matrix containing all pixel data of the image can be constructed. This pixel matrix is ​​the basis of digital image processing, which enables the image to be understood and processed by a computer. Each pixel corresponds to an element in the matrix, and the value of the element represents the brightness, color or other attributes of the pixel. At the same time, constructing the pixel matrix helps to locate subsequent pixels. In the present method, this step provides a reliable data basis for subsequent temperature acquisition range adjustment, status monitoring of insulator elements, etc.

[0093] The specific process of filling the pixel points with large pixel value deviation in the pixel matrix is ​​as follows:

[0094] Step S3.11: Obtain the coordinates of the pixel points with the largest pixel value deviation in the pixel matrix.

[0095] Step S3.12: Construct a 3×3 matrix B with the pixel coordinates obtained in step S3.11 as the center.

[0096] Step S3.13: Calculate the mean of the pixels in matrix B to obtain the mean pixel, which is expressed as:

[0097] ;

[0098] In the formula, is the mean pixel; It is the coordinates of the pixel points with large pixel value deviation in the pixel matrix.

[0099] Those skilled in the art can understand that, for each identified pixel with a large deviation, a 3×3 matrix B is constructed with its coordinates as the center. This matrix B includes the pixel and its 8 neighboring pixels. The 3×3 matrix is ​​selected because it is a smaller local area that can reflect the relationship between the pixel and its neighboring points. At the same time, the amount of calculation is relatively small. Using the mean pixel point filling can significantly reduce the noise and outliers in the image, making the image look smoother and more continuous. At the same time, since the replacement is based on the mean of the local area, this method can retain most of the details and structural information of the image.

[0100] Step S3.14: Use the mean pixel to replace the pixel with a large pixel value deviation in the pixel matrix.

[0101] The specific process of step S4 is:

[0102] Step S4.1: Convert the filtered insulator string image into a grayscale image, scan each pixel in the grayscale image to construct a matrix C, and obtain the grayscale value of each pixel in the grayscale image.

[0103] Step S4.2: Take the lower left corner of the matrix C as the origin, the origin along the length direction of the matrix C as the x-axis direction, and the origin along the width direction of the matrix C as the y-axis direction; combine the grayscale value of each pixel point, and take the origin as the starting point, perform convolution calculation on each pixel point in the grayscale image along the x-axis and y-axis directions respectively, and obtain the gradient of each pixel point in the grayscale image in the x-axis and y-axis directions.

[0104] Among them, taking the origin as the starting point, convolution calculation is performed on each pixel in the grayscale image along the x-axis and y-axis directions respectively, and the gradient of each pixel in the grayscale image in the x-axis and y-axis directions is obtained, which is expressed as:

[0105] ;

[0106] ;

[0107] ;

[0108] In the formula, and are the approximate values ​​of the gradient of the pixel points in the grayscale image in the x-axis and y-axis directions respectively; is the gray value of the pixel in the gray image; It is the gradient of the pixel in the grayscale image in the x-axis and y-axis directions.

[0109] Step S4.3: Combine the gradients of each pixel in the grayscale image in the x-axis and y-axis directions to calculate the gradient direction of each pixel in the grayscale image, expressed as:

[0110] ;

[0111] In the formula, is the gradient direction of the pixel in the grayscale image.

[0112] Step S4.4: Obtain edge features of the grayscale image, that is, edge features of the insulator string image after filtering, according to the gradient direction of each pixel point in the grayscale image.

[0113] It is understood by those skilled in the art that the Sobel operator is a first-order discrete difference operator used to calculate the approximate value of the gradient of the image brightness function. It uses a 3×3 mask to perform a convolution operation on the image. The calculation process is relatively simple and the execution speed is fast. It is suitable for scenes that require fast processing. The Sobel operator can produce relatively smooth and continuous edges while detecting edges, which is helpful for subsequent image analysis and processing. The Sobel operator can calculate the gradients in the horizontal and vertical directions respectively, thereby providing the direction information of the edge, which is very useful for subsequent edge connection, contour extraction and other processing.

[0114] The specific process of step S5 is:

[0115] Step S5.1: obtaining the real-time insulator string outline, scanning the real-time insulator string outline, and obtaining the real-time insulator string outline parameters;

[0116] Step S5.2: Analyze the edge features obtained in step S4, obtain feature parameters of the edge features, and compare the feature parameters of the edge features with the real-time insulator string shape parameters;

[0117] Step S5.3: Filter out the area where the difference between the feature parameters of the edge features in the filtered insulator string image and the real-time insulator string shape parameters is the lowest, which is the area where the insulator string is located in the filtered insulator string image.

[0118] It can be understood by those skilled in the art that obtaining a clear outline of the insulator string and obtaining shape parameters provide an accurate insulator string shape reference for subsequent comparison with image edge features. The shape parameters help to accurately identify the insulator string in a complex image background. A detailed analysis of the detected edge features and extraction of various parameters of the edge features, such as edge length, direction, curvature and coordinates of edge points, help to identify edge features related to the insulator string in the image. By comparing the extracted insulator string shape parameters with various parameters of the image edge features one by one, the position of the insulator string in the image can be accurately identified, providing key information for subsequent image processing and analysis. At the same time, through comparison and screening, interference factors in the image can be effectively eliminated, thereby improving the accuracy and robustness of insulator string identification.

[0119] The specific process of step S6 is:

[0120] Step S6.1: Obtain real-time insulator component shape parameters.

[0121] Step S6.2: Divide the area where the insulator string is located according to the real-time insulator element shape parameters to obtain the number of insulator elements in the area where the insulator string is located, and mark the insulator elements from top to bottom to obtain the area where each insulator element is located in the insulator string image after filtering.

[0122] Among them, the division of the area where the insulator string is located according to the real-time insulator element shape parameters can be expressed as:

[0123] ;

[0124] In the formula, The real-time insulator component shape parameters and the first The similarity of features, The real-time insulator component shape parameters The first feature The characteristic index value, The first The first feature The characteristic index value, is the total number of characteristic index values.

[0125] It will be understood by those skilled in the art that by comparing the external shape parameters of the insulator elements with the edge features of the insulator string image, the number of insulator elements in the insulator string in the image can be quickly obtained, which facilitates the subsequent acquisition of the temperatures of multiple insulator elements and the arrangement of the insulator elements in the insulator string. The status and changes of each element can be conveniently tracked and analyzed, providing strong support for subsequent monitoring and maintenance of the insulator string.

[0126] The specific process of step S7 is:

[0127] Step S7.1: Obtain the brightness value of the pixel points in the area where each insulator element is located.

[0128] Step S7.2: performing mean value processing on the brightness values ​​of the pixels in the area where each insulator element is located, respectively, to obtain the mean brightness value of the pixels in the area where each insulator element is located.

[0129] Step S7.3: Obtain the highest and lowest values ​​of brightness in the area where each insulator element is located according to the average brightness of the pixel points in the area where each insulator element is located, expressed as:

[0130] ;

[0131] ;

[0132] In the formula, is the number of insulator elements; Indicates the average brightness of the pixels in the area where the insulator element is located; Indicates the preset value of the number of insulator elements.

[0133] Those skilled in the art will appreciate that by performing mean processing on the areas where multiple insulator elements are located, it is possible to prevent the problem of low brightness of some insulator elements due to surface dirt and obstruction of the insulator elements, thereby affecting subsequent judgments. By comparing the brightness means of different insulator element areas, it is possible to preliminarily judge the surface state differences of each insulator element. By comparing the highest and lowest values, the overall state differences and uniformity of the insulator string can be evaluated, providing a decision-making basis for subsequent maintenance and replacement.

[0134] The specific process of step S8 is:

[0135] Step S8.1: Acquire the temperature acquisition range of the initially acquired insulator string image.

[0136] Step S8.2: Based on the temperature acquisition range of the initially acquired insulator string image and the brightness range, the temperature corresponding to the brightness value within the brightness range is obtained, which is expressed as:

[0137] ;

[0138] In the formula, and They respectively represent the highest temperature and the lowest temperature within the temperature acquisition range of the initially acquired insulator string image; Represents the temperature adjustment factor.

[0139] Step S8.3: Match the highest and lowest brightness values ​​in the region where the insulator element is located with the temperature corresponding to the brightness value, and obtain the temperature corresponding to the highest and lowest brightness values ​​in the region where the insulator element is located, expressed as:

[0140] ;

[0141] ;

[0142] In the formula, and They respectively represent the temperatures corresponding to the highest and lowest brightness values ​​in the area where the insulator element is located.

[0143] The specific process of step S9 is:

[0144] Step S9.1: Calculate the temperature difference in the area where the insulator element is located according to the temperatures corresponding to the highest and lowest brightness values ​​in the area where the insulator element is located.

[0145] Step S9.2: Compare the temperature difference in the area where the insulator element is located with the current temperature difference. If the current temperature difference is much larger than the temperature difference in the area where the insulator element is located, it is judged that the current temperature measurement range is too large and does not meet the requirements. If the current temperature difference is close to the temperature difference in the area where the insulator element is located, it is judged that the current temperature measurement range is appropriate and meets the requirements.

[0146] Step S9.3: When it is determined that the current temperature measurement range is too large and does not meet the requirements, the highest temperature in the area where the insulator element is located is compared with the highest temperature in the current temperature measurement range, and the lowest temperature in the area where the insulator element is located is compared with the lowest temperature in the current temperature measurement range to obtain the position where the temperature in the area where the insulator element is located is within the current temperature measurement range.

[0147] Step S9.4: According to the position where the temperature in the area where the insulator element is located is within the current temperature measurement range, the current temperature measurement range is adjusted.

[0148] Step S9.5: recalculate the adjusted current temperature measurement range to obtain the optimal temperature measurement range, adjust the insulator string image to the optimal temperature measurement range through the optimal temperature measurement range, and obtain the optimal insulator string image.

[0149] It can be understood by those skilled in the art that the temperature difference in the area where the insulator element is located reflects the uniformity of the temperature distribution inside the insulator string, and is one of the important indicators for evaluating the operating status of the insulator string. By comparison, it can be determined whether the temperature measurement range currently used is appropriate. If the current temperature difference is much larger than the temperature difference in the area where the insulator element is located, it may mean that the measurement range is too large, resulting in a decrease in measurement accuracy; if the two are close, it means that the current measurement range is more appropriate, and the relative position of the temperature difference in the area where the insulator element is located in the current temperature measurement range can be determined. The specific distribution of the temperature difference in the area where the insulator element is located in the current measurement range can be clarified, providing a basis for adjusting the measurement range. By adjusting the measurement range, the accuracy and efficiency of temperature measurement can be improved, and the actual operating status of the insulator string can be more accurately reflected. By continuously monitoring and adjusting the temperature measurement range, the temperature monitoring range of the insulator string can be continuously optimized.

[0150] The specific process of step S10 is:

[0151] The area where the insulator string is located in the optimal insulator string image is obtained, and the area where the insulator elements with abnormal temperature distribution are located in the area where the insulator string is located is calculated through the multi-head attention mechanism, which is expressed as:

[0152] ;

[0153] In the formula, is the attention weight in the region where the insulator string is located, and denote the query and key vectors respectively, is the dimension of the feature vector; Indicates the current pixel position; Indicates except Other pixel positions of ; represents the normalization function;

[0154] The area where the insulator element with abnormal temperature distribution is located is combined with its own temperature gradient information to identify the part with significant temperature change, that is, the abnormal area;

[0155] The multi-scale convolutional neural network MSCNN is used to process the area where the insulator string is located in the optimal insulator string image, and the parts with different abnormal temperatures in the area where the insulator string is located are extracted, which is expressed as:

[0156] ;

[0157] In the formula, represents the parts with different abnormal temperatures extracted by the multi-scale convolutional network (MSCNN), It is Convolution operations at different scales, The position of the part with different abnormal temperature The temperature value at Represents the total number of scales used by the multi-scale convolutional network MSCNN.

[0158] By calculating the temperature gradient of each pixel in the parts with different abnormal temperatures, the area with abnormal temperature change, that is, the abnormal area, is determined; the temperature gradient calculation formula is:

[0159] ;

[0160] In the formula, The position of the part showing different abnormal temperatures The temperature gradient at It means to find partial derivatives.

[0161] Perform adaptive connected domain analysis on abnormal areas, optimize the connected domain analysis through conditional random field (CRF), merge adjacent abnormal areas, obtain complete degraded areas, and highlight them;

[0162] The connected domain analysis is optimized by conditional random field (CRF), which is expressed as:

[0163] ;

[0164] In the formula, is the objective function of the conditional random field, Indicates abnormal area; Labels indicating degraded areas; and represents any two pixel positions in the abnormal region; express The local characteristics of express and The joint features between .

[0165] The deterioration identification system based on the temperature difference adjustment of the infrared thermal image of the insulator string includes:

[0166] The image acquisition module is used to acquire the insulator string image; the insulator string image is an infrared thermal image.

[0167] The scanning module is used to scan the insulator string image to obtain a pixel matrix of the insulator string image.

[0168] The preprocessing module is used to fill the pixel points whose pixel value deviation is greater than a preset value in the pixel matrix, and filter the insulator string image according to the filled pixel matrix.

[0169] The feature acquisition module is used to obtain the edge features of the insulator string image after filtering using the Sobel operator.

[0170] The comparison module is used to obtain the real-time insulator string shape, compare the real-time insulator string shape with the obtained edge features, and obtain the area where the insulator string is located in the insulator string image after filtering.

[0171] The division module is used to divide the area where the insulator string is located according to the real-time insulator element appearance, and obtain the area where each insulator element is located in the insulator string image after filtering.

[0172] The brightness mean module is used to perform mean processing on the brightness of the pixels in the area where each insulator element is located, and obtain the highest and lowest brightness values ​​in the area where each insulator element is located according to the brightness mean of the pixels in the area where each insulator element is located.

[0173] The temperature calculation module is used to calculate the temperature corresponding to the highest and lowest brightness values ​​in the area where the insulator element is located.

[0174] The temperature range determination module is used to calculate the temperature difference in the area where the insulator element is located according to the temperatures corresponding to the highest and lowest brightness values ​​in the area where the insulator element is located, and obtain the optimal temperature measurement range in combination with the current temperature. The insulator string image is adjusted to the optimal temperature measurement range through the optimal temperature measurement range to obtain the optimal insulator string image.

[0175] The degradation identification module is used to obtain the area where the insulator string is located in the optimal insulator string image, calculate the abnormal area in the area where the insulator string is located through the multi-head attention mechanism and multi-scale convolutional neural network, perform adaptive connected domain analysis on the abnormal area, and optimize the connected domain analysis through the conditional random field CRF, merge adjacent abnormal areas, obtain the complete degraded area, and highlight it.

[0176] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A degradation identification method based on temperature difference adjustment of infrared thermal images of insulator strings, characterized in that: The steps include: Step S1: acquiring an insulator string image; the insulator string image is an infrared thermal image; Step S2: Scanning the insulator string image to obtain a pixel matrix of the insulator string image; Step S3: Filling the pixel points in the pixel matrix whose pixel value deviation is greater than a preset value, and filtering the insulator string image according to the filled pixel matrix; Step S4: using the Sobel operator to obtain edge features of the insulator string image after filtering; Step S5: obtaining the real-time insulator string shape, comparing the real-time insulator string shape with the edge features obtained in step S4, and obtaining the region where the insulator string is located in the insulator string image after filtering; Step S6: dividing the area where the insulator string is located according to the real-time insulator element appearance, and obtaining the area where each insulator element is located in the insulator string image after filtering; Step S7: performing mean value processing on the brightness of the pixels in the area where each insulator element is located, and obtaining the highest value and the lowest value of the brightness in the area where each insulator element is located according to the mean value of the brightness of the pixels in the area where each insulator element is located; Step S8: Calculate the temperature corresponding to the highest and lowest brightness values ​​in the area where the insulator element is located; Step S9: Calculate the temperature difference in the area where the insulator element is located according to the temperatures corresponding to the highest and lowest brightness values ​​in the area where the insulator element is located, obtain the optimal temperature measurement range in combination with the current temperature, and adjust the insulator string image to the optimal temperature measurement range through the optimal temperature measurement range to obtain the optimal insulator string image; Step S10: Obtain the area where the insulator string is located in the optimal insulator string image, calculate the abnormal area in the area where the insulator string is located through a multi-head attention mechanism and a multi-scale convolutional neural network, perform adaptive connected domain analysis on the abnormal area, and optimize the connected domain analysis through the conditional random field CRF, merge adjacent abnormal areas, obtain a complete degraded area, and highlight it.

2. The degradation identification method based on temperature difference adjustment of infrared thermal images of insulator strings according to claim 1 is characterized in that: The specific process of step S10 is: Obtain the area where the insulator string is located in the optimal insulator string image, and calculate the area where the insulator elements with abnormal temperature distribution are located in the area where the insulator string is located through a multi-head attention mechanism; The area where the insulator element with abnormal temperature distribution is located is combined with its own temperature gradient information to identify the part with significant temperature change, that is, the abnormal area; The multi-scale convolutional neural network MSCNN is used to process the area where the insulator string is located in the optimal insulator string image, and the parts with different abnormal temperatures in the area where the insulator string is located are extracted, which is expressed as: ； In the formula, Represents the parts with different abnormal temperatures extracted by the multi-scale convolutional network MSCNN, It is Convolution operations at different scales, The position of the part with different abnormal temperature The temperature value at Represents the total number of scales used by the multi-scale convolutional network MSCNN; By calculating the temperature gradient of each pixel in the parts with different abnormal temperatures, the area with abnormal temperature change, i.e., the abnormal area, is determined; the temperature gradient calculation formula is: ； In the formula, The position of the part showing different abnormal temperatures The temperature gradient at It means to find partial derivatives; Perform adaptive connected domain analysis on abnormal areas, optimize the connected domain analysis through conditional random field (CRF), merge adjacent abnormal areas, obtain complete degraded areas, and highlight them; The connected domain analysis is optimized through the conditional random field CRF, which is expressed as: ； In the formula, is the objective function of the conditional random field, Indicates abnormal area; Labels indicating degraded areas; and represents any two pixel positions in the abnormal region; express The local characteristics of express and The joint features between .

3. The degradation identification method based on temperature difference adjustment of infrared thermal images of insulator strings according to claim 2 is characterized in that: The specific process of step S5 is: Step S5.1: obtaining the real-time insulator string outline, scanning the real-time insulator string outline, and obtaining the real-time insulator string outline parameters; Step S5.2: Analyze the edge features obtained in step S4, obtain feature parameters of the edge features, and compare the feature parameters of the edge features with the real-time insulator string shape parameters; Step S5.3: Filter out the area where the difference between the feature parameters of the edge features in the filtered insulator string image and the real-time insulator string shape parameters is the lowest, which is the area where the insulator string is located in the filtered insulator string image.

4. The degradation identification method based on temperature difference adjustment of infrared thermal images of insulator strings according to claim 3 is characterized in that: The specific process of step S6 is: Step S6.1: obtaining real-time insulator component shape parameters; Step S6.2: Divide the area where the insulator string is located according to the real-time insulator element shape parameters to obtain the number of insulator elements in the area where the insulator string is located, and mark the insulator elements from top to bottom to obtain the area where each insulator element is located in the insulator string image after filtering; Among them, the area where the insulator string is located is divided according to the real-time insulator element shape parameters as follows: ； In the formula, The real-time insulator component shape parameters and the first The similarity of the features; The real-time insulator component shape parameters The feature of characteristic index value; The first The feature of characteristic index value; is the total number of characteristic index values.

5. The degradation identification method based on temperature difference adjustment of insulator string infrared thermal image according to claim 4 is characterized in that: The specific process of step S7 is: Step S7.1: Obtain the brightness value of the pixel points in the area where each insulator element is located; Step S7.2: performing mean value processing on the brightness values ​​of the pixels in the area where each insulator element is located, respectively, to obtain the brightness mean value of the pixels in the area where each insulator element is located; Step S7.3: Obtain the highest and lowest values ​​of brightness in the area where each insulator element is located according to the average brightness of the pixel points in the area where each insulator element is located, expressed as: ； ； In the formula, is the number of insulator elements; Indicates the average brightness of the pixels in the area where the insulator element is located; Indicates the preset value of the number of insulator elements.

6. The degradation identification method based on temperature difference adjustment of infrared thermal images of insulator strings according to claim 5 is characterized in that: The specific process of step S8 is: Step S8.1: obtaining the temperature acquisition range of the initially acquired insulator string image; Step S8.2: Based on the temperature acquisition range of the initially acquired insulator string image and the brightness range, the temperature corresponding to the brightness value within the brightness range is obtained, which is expressed as: ； In the formula, and They respectively represent the highest temperature and the lowest temperature within the temperature acquisition range of the initially acquired insulator string image; represents the temperature adjustment factor; Step S8.3: Match the highest and lowest brightness values ​​in the region where the insulator element is located with the temperature corresponding to the brightness value, and obtain the temperature corresponding to the highest and lowest brightness values ​​in the region where the insulator element is located, expressed as: ； ； In the formula, and They respectively represent the temperatures corresponding to the highest and lowest brightness values ​​in the area where the insulator element is located.

7. The degradation identification method based on temperature difference adjustment of infrared thermal images of insulator strings according to claim 6 is characterized in that: The specific process of step S9 is: Step S9.1: Calculate the temperature difference in the area where the insulator element is located according to the temperatures corresponding to the highest and lowest brightness values ​​in the area where the insulator element is located; Step S9.2: Compare the temperature difference in the area where the insulator element is located with the current temperature difference to judge the current temperature measurement range; Step S9.3: When it is determined that the current temperature measurement range does not meet the requirements, the highest temperature in the area where the insulator element is located is compared with the highest temperature in the current temperature measurement range, and the lowest temperature in the area where the insulator element is located is compared with the lowest temperature in the current temperature measurement range, to obtain the position where the temperature in the area where the insulator element is located is within the current temperature measurement range; Step S9.4: adjusting the current temperature measurement range according to the position of the temperature in the area where the insulator element is located within the current temperature measurement range; Step S9.5: recalculate the adjusted current temperature measurement range to obtain the optimal temperature measurement range, adjust the insulator string image to the optimal temperature measurement range through the optimal temperature measurement range, and obtain the optimal insulator string image.

8. A degradation identification system based on temperature difference adjustment of infrared thermal images of insulator strings, characterized in that: include: An image acquisition module, used for acquiring an image of an insulator string; The image of the insulator string is an infrared thermal image; A scanning module, used for scanning the insulator string image to obtain a pixel matrix of the insulator string image; A preprocessing module, used to fill the pixel points whose pixel value deviation is greater than a preset value in the pixel matrix, and filter the insulator string image according to the filled pixel matrix; A feature acquisition module, used for acquiring edge features of the insulator string image after filtering using a Sobel operator; A comparison module is used to obtain the real-time insulator string shape, compare the real-time insulator string shape with the obtained edge features, and obtain the area where the insulator string is located in the insulator string image after filtering; A division module is used to divide the area where the insulator string is located according to the real-time insulator element appearance, and obtain the area where each insulator element is located in the insulator string image after filtering; A brightness mean module is used to perform mean processing on the brightness of the pixels in the area where each insulator element is located, and obtain the highest and lowest brightness values ​​in the area where each insulator element is located according to the brightness mean of the pixels in the area where each insulator element is located; A temperature calculation module is used to calculate the temperature corresponding to the highest and lowest brightness values ​​in the area where the insulator element is located; A temperature range determination module is used to calculate the temperature difference in the area where the insulator element is located according to the temperatures corresponding to the highest and lowest brightness values ​​in the area where the insulator element is located, and to obtain the optimal temperature measurement range in combination with the current temperature. The insulator string image is adjusted to the optimal temperature measurement range through the optimal temperature measurement range to obtain the optimal insulator string image; The degradation identification module is used to obtain the area where the insulator string is located in the optimal insulator string image, calculate the abnormal area in the area where the insulator string is located through the multi-head attention mechanism and multi-scale convolutional neural network, perform adaptive connected domain analysis on the abnormal area, and optimize the connected domain analysis through the conditional random field CRF, merge adjacent abnormal areas, obtain the complete degraded area, and highlight it.

Images