Surface damage detection system for power grid equipment based on image analysis

Abstract

The present invention discloses a surface damage detection system for power grid equipment based on image analysis, which belongs to the technical field of power grid equipment detection, and includes a device image collection module, a database establishment module, an image two-dimensional processing module, and an abnormality determination module. The device image collection module is used to collect images of power grid equipment in substations, including cables, transformers, and power poles and towers. The database establishment module is based on data obtained by the device image collection module. The surface damage detection system for power grid equipment based on image analysis of the present invention can more clearly identify faults of power grid equipment by performing two-dimensional conversion on images of power grid equipment. Image processing technology can automatically detect and locate various faults on equipment by converting and analyzing image data, thereby improving the efficiency and accuracy of detection, and helping to timely discover and solve equipment problems, thereby ensuring the stability and reliability of the power grid.

Description

Technical Field

[0001] The invention belongs to the technical field of power grid equipment detection, and in particular is a power grid equipment surface damage detection system based on image analysis. Background Art

[0002] Grid equipment refers to various types of equipment used in power systems. These devices play a key role in the generation, transmission, distribution and control of electricity. The main function of grid equipment is to ensure that electricity reaches end users from power plants and maintain the stability and security of the power system. In substations, grid equipment mainly includes transformers, power transmission cables and electric poles.

[0003] In order to ensure urban power transmission, it is necessary to conduct regular inspections on the power grid equipment in the substation to ensure the normal operation of the substation. Traditional power grid inspections mainly rely on manual inspections, and the equipment is inspected by operators visually or through specific instrument methods. For example, the Chinese patent with the publication number "CN116188370A" discloses "a power grid inspection method, device, computer equipment, storage medium and computer program product. The method includes: obtaining an inspection image for the power grid and inspection item information corresponding to the inspection image". Although it can complete the inspection of power grid equipment, the process is relatively complicated, and it is impossible to make regular and rapid judgments on the damage of power grid equipment, nor to predict possible faults. The existing visual inspection by personnel is simple and fast, but it is prone to misjudgment and it is not easy to find small damage, which has the problem of low practicality and functionality.

[0004] In view of the above, this case proposes a surface damage detection system for power grid equipment based on image analysis to solve the above technical problems. Summary of the invention

[0005] The present invention aims to solve at least one of the technical problems existing in the prior art. To this end, the present invention proposes a surface damage detection system for power grid equipment based on image analysis, which solves the above technical problems by improving the detection method and processing method.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] The surface damage detection system of power grid equipment based on image analysis includes equipment image collection module, database establishment module, image two-dimensional processing module and abnormality determination module;

[0008] The equipment image collection module is used to collect images of substation power grid equipment, including cables, transformers, and power poles and towers;

[0009] The database establishment module establishes an original image database based on the data obtained by the device image collection module and according to the location information of the substation;

[0010] The image two-dimensional processing module, based on the image data in the original image database, performs contour extraction and background stripping on the image to be detected through a threshold, performs two-dimensional conversion on the image through a computer vision library, and transmits the relevant data to the abnormality determination module;

[0011] The abnormality determination module performs line detection on the two-dimensional line image based on the device type of the original image, determines the device abnormality, transmits the abnormality signal into the original image database, and marks the relevant device files.

[0012] Furthermore, based on the image data in the original image database, the image to be detected is subjected to contour extraction and background stripping through a threshold, the image is subjected to two-dimensional conversion through a computer vision library, and the relevant data is transmitted to the abnormality determination module, and the specific steps are as follows:

[0013] Grayscale processing is performed on the image data in the original image database, and the background of the original image is stripped by setting the grayscale threshold, and the stripped image is marked as the image to be extracted;

[0014] The edge detection algorithm is used to detect the contour of the image to be extracted, and the computer vision library is used to perform two-dimensional conversion on the image to obtain a two-dimensional line image of the relevant equipment.

[0015] Furthermore, the grayscale processing is performed on the image data in the original image database, the background of the original image is stripped by setting the grayscale threshold, and the stripped image is marked as the image to be extracted, including the following steps:

[0016] For the color images in the original image database, traverse each pixel, substitute the RGB value of the relevant image into the grayscale weighted average formula to calculate the grayscale value, and convert the color images in the original image database into grayscale images;

[0017] According to the converted grayscale image, the grayscale threshold is set by Otsu's method and the background of the grayscale image is stripped. The specific steps are as follows:

[0018] The algorithm formula for statistical image gray value distribution is:

[0019] ;

[0020] in, Represents gray value The pixel probability, Represents gray value The number of pixels, The total number of pixels in the image is calculated by assuming a grayscale threshold , divide the image pixels into foreground and background, and calculate the weight of the foreground class and the weight of the background class;

[0021] Among them, the weight of the foreground class , that is, the gray value is less than or equal to , and its algorithm formula is:

[0022] ;

[0023] Among them, the weight of the background class , that is, the gray value is greater than , and its algorithm formula is:

[0024] ;

[0025] Calculate the grayscale average of the foreground class , and its algorithm formula is:

[0026] ;

[0027] Calculate the grayscale average of the background class , and its algorithm formula is:

[0028] ;

[0029] Calculate the total average gray value of the image , and its algorithm formula is:

[0030] ;

[0031] Weights based on foreground class , weight of background class , the average grayscale value of the foreground class , the grayscale average of the background class And the total average gray value , calculate the between-class variance , and its algorithm formula is:

[0032] ;

[0033] By between-class variance Value, determine the optimal threshold , and its algorithm formula is:

[0034] ;

[0035] By the optimal threshold , background stripping for grayscale images includes the following steps:

[0036] ;

[0037] in, Represents the binary image at position The pixel value at Represents the grayscale image at position The pixel value at .

[0038] Furthermore, the contour of the image to be extracted is detected by an edge detection algorithm, and the image is converted into two dimensions using a computer vision library to obtain a two-dimensional line image of the relevant device. The specific steps are as follows:

[0039] The Sobel operator is used to perform edge detection on the relevant image after background stripping to obtain the device outline. The steps are as follows:

[0040] ;

[0041] ;

[0042] In the above formula, is the Sobel operator kernel in the horizontal direction, is the Sobel operator kernel in the vertical direction. Based on the Sobel operator kernel in the horizontal direction and the Sobel operator kernel in the vertical direction, the horizontal gradient direction and the vertical gradient direction are calculated:

[0043] ;

[0044] ;

[0045] in, represent The horizontal gradient calculated at the coordinates, represent The vertical gradient calculated at the coordinate, Represents the grayscale image at position The pixel value at ;

[0046] Calculate the gradient magnitude With direction :

[0047] ;

[0048] ;

[0049] The edge points of the device in the image are further determined by non-maximum suppression and double threshold processing. The specific steps are as follows:

[0050] For each pixel, its gradient magnitude is compared with the gradient magnitudes of two adjacent pixels along the gradient direction. If the gradient magnitude of the selected pixel is not the local maximum, the selected pixel is set to 0, otherwise it remains unchanged.

[0051] By setting a high threshold and a low threshold, pixels with a gradient amplitude greater than the high threshold are determined as strong edge points, pixels with a gradient amplitude less than the low threshold are determined as non-edge points, and pixels between the high threshold and the low threshold are determined as strong edge points when connected to strong edge points, and as non-edge points when connected to non-edge points;

[0052] pass In Extract contours from edge pixels and organize them into a series of contours through edge points. Each contour is a continuous sequence of points, forming a closed or open curve.

[0053] pass Generate a completely black image of the same size as the original image, draw the extracted contour lines on the new image, generate a two-dimensional line image, and transmit the two-dimensional line image to the abnormality judgment module.

[0054] Furthermore, the abnormality determination module performs line detection on the two-dimensional line image based on the device type of the original image, determines the device abnormality, transmits the abnormal signal into the original image database, and marks the relevant device files. The specific steps are as follows:

[0055] Through the original image database, the types of power grid equipment in the relevant two-dimensional line images are obtained, including transformers, cables, and power poles and towers;

[0056] Based on different types of power grid equipment, different line detection methods are used for two-dimensional line images to determine the damage of power grid equipment;

[0057] Based on the results of the power grid equipment damage determination, the abnormal signal is transmitted into the original image database, the relevant equipment files are marked, and the relevant abnormal files are marked in red.

[0058] Furthermore, the method of using different line detection methods for two-dimensional line images based on different types of power grid equipment to determine the damage of power grid equipment includes the following steps:

[0059] For the two-dimensional line image, a coordinate system conversion method is adopted, the lower left corner of the two-dimensional line image is set as the origin of the coordinate system, and a plane rectangular coordinate system is established with the horizontal direction of the two-dimensional line image as the axis and the vertical direction as the axis;

[0060] For cables, obtain the position coordinates of the cables in the rectangular coordinate system through the two-dimensional line image and , the curvature of the relevant cable lines is calculated by the curvature calculation formula, and the algorithm formula is:

[0061] ;

[0062] in, Represents the curve at point The first-order derivative in the horizontal direction and the first-order derivative in the vertical direction, Represents the curve at point The horizontal second-order derivative and the vertical second-order derivative of set the curvature threshold The steps to determine the bending degree of the cable are as follows:

[0063] when ≤ , indicating that the bending of the relevant cable segment is normal;

[0064] when > , which means that the relevant cable segment is abnormally bent and there is a fault;

[0065] At the same time, for each point on the line, the distance between it and its adjacent points is calculated. The algorithm formula is:

[0066] ;

[0067] in, Represents two points , The distance between Determine the break of the line:

[0068] when ≤ , indicating that the relevant cable segment is normal;

[0069] when > , which means that the continuity of the relevant cable segment is abnormal and there is a break fault;

[0070] For the power tower, the distance between each point on the line and its adjacent points is calculated, and the continuity is determined by combining the threshold. The steps are as follows:

[0071] ;

[0072] in, Represents two points , The distance between Determine the break of the line:

[0073] when ≤ , indicating that the relevant power tower section is normal;

[0074] when > , which means that the continuity of the relevant power tower section is abnormal and there is a fracture fault;

[0075] Calculate the angle of the main line of the power tower by fitting a straight line to the contour line, and use linear regression to fit the straight line to get the slope of the straight line and the intercept , calculate the angle of the line :

[0076] ;

[0077] Set the tilt angle threshold , comparing the line angles of different parts of the power tower , for abnormal determination of power towers:

[0078] when ≤ , indicating that the relevant power tower section is normal;

[0079] when > , which means that the relevant power tower section is abnormal and there is a risk of tilting;

[0080] For the transformer, for the two-dimensional line pattern, draw the reference line according to the normal contour, calculate the distance between the actual line and the reference line, and judge the expansion and depression of the transformer shell:

[0081] ;

[0082] in, is the vertical coordinate of the point on the actual contour line, is the vertical coordinate of the corresponding point on the reference line, based on the judgment threshold , for transformer abnormality determination:

[0083] when ≤ , indicating that the external part of the relevant transformer is normal;

[0084] when > , which means the external part of the relevant transformer is abnormal, with depression or expansion;

[0085] For each point on the line, the distance between it and its adjacent point is calculated, and the integrity of the transformer casing is determined by combining the threshold. The steps are as follows:

[0086] ;

[0087] in, Represents two points , The distance between Determine the break of the line:

[0088] when ≤ , representing that the external part of the relevant transformer is intact;

[0089] when > , indicating that the relevant transformer has external abnormalities and is at risk of breakage.

[0090] Furthermore, based on the result of the power grid equipment damage determination, the abnormal signal is transmitted into the original image database, the relevant equipment files are marked, and the relevant abnormal files are marked in red. The specific steps are:

[0091] Based on the results of the power grid equipment damage judgment, when there is an abnormality in the power grid equipment, the abnormal signal is transmitted to the database establishment module, and the relevant equipment files in the relevant original image database are marked in red;

[0092] When the power grid equipment is judged to be normal, the transmission judgment related data enters the database establishment module, and the data is stored and backed up in the relevant equipment files in the relevant original image database, and the date is marked.

[0093] Furthermore, the abnormality determination module is also provided with a cable damage tendency prediction system, which includes the following steps:

[0094] Based on the relevant cable files in the original image database, the judgment result data is compared, the detection data at the same position is compared at intervals, and the damage tendency is predicted. The specific steps are as follows:

[0095] get The curvature of and The curvature of , and the time interval , calculate the curvature change rate, the algorithm formula is:

[0096] ;

[0097] ;

[0098] in, Represents the current detection time. For The current time moves forward 2 times. Represents the curvature change rate, for the curvature change rate threshold , predict the damage tendency:

[0099] when > , which means the curvature change rate is large and there is a tendency to damage. The relevant files are marked in orange:

[0100] when > , which means that the curvature change rate is large and there is a tendency to be damaged. The relevant files are marked in orange.

[0101] Further, the equipment image collection module includes a fixed camera for regularly collecting images of substation power grid equipment, including cables, transformers, and power poles and towers, and transmitting the data into the database establishment module;

[0102] The database establishment module establishes an original image database based on the data obtained by the equipment image collection module and the location information of the substation, and creates and classifies files for different equipment at different locations. At the same time, it receives the judgment-related data of the abnormal judgment module and supplements the relevant files.

[0103] Compared with the prior art, the present invention has the following beneficial effects:

[0104] 1. In the present invention, by regularly shooting the power grid equipment images, performing foreground stripping and two-dimensional line conversion on the equipment images, the shape and structural features of the equipment can be highlighted. After the power grid equipment images are converted into two-dimensional lines, the amount of image data is greatly reduced, and the system processing speed is improved. For large-scale power grid equipment inspections, image analysis can be completed more quickly and equipment problems can be discovered in a timely manner;

[0105] 2. In the present invention, by further analyzing the lines of the two-dimensional image, the looseness and broken strands of the transmission cable can be detected. If the transmission cable is loose, the line will become more curved. If the transmission cable is broken, the line will be discontinuous. By real-time monitoring of the transmission cable lines, the fault of the transmission cable can be discovered in time;

[0106] 3. In the present invention, by analyzing the angle of the power tower line, the inclination of the power tower can be detected. By drawing the reference line and detecting the depression and protrusion of the transformer line, the deformation and fracture of the power tower and the deformation of the transformer can be discovered in time, ensuring the safe and stable operation of the power line, and enhancing the practicality and functionality of the system.

[0107] 4. In the present invention, by periodically acquiring images of power grid equipment and transmitting each recorded data to the relevant equipment archive, and by comparing the curvature change rate of the transmission cable at intervals, the fault condition of the transmission cable can be predicted, helping the staff to formulate maintenance plans in advance and arrange resources reasonably;

[0108] 5. The entire power grid equipment surface damage detection system based on image analysis can more clearly identify the faults of power grid equipment by performing two-dimensional conversion on the power grid equipment image. Image processing technology can automatically detect and locate various faults on the equipment by converting and analyzing image data, thereby improving the efficiency and accuracy of detection, and helping to promptly discover and solve equipment problems, thereby ensuring the stability and reliability of the power grid. BRIEF DESCRIPTION OF THE DRAWINGS

[0109] Figure 1 This is a block diagram of the power grid equipment surface damage detection system based on image analysis of the present invention. DETAILED DESCRIPTION

[0110] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0111] Embodiment 1:

[0112] like Figure 1 As shown, the surface damage detection system of power grid equipment based on image analysis includes an equipment image collection module, a database establishment module, an image two-dimensional processing module, and an abnormality determination module;

[0113] Equipment image collection module, used to collect images of substation power grid equipment, including cables, transformers, and power towers;

[0114] The equipment image collection module includes a fixed camera, which is used to regularly collect images of substation power grid equipment, including cables, transformers, and power towers, and transmit the data to the database establishment module;

[0115] It should be noted that fixed cameras are used to regularly collect images of equipment that need to be inspected in the substation to ensure that the shooting position is fixed each time to avoid errors in subsequent judgments caused by viewing angle errors. Taking cables as an example, it is necessary to obtain the number and parameters of cable equipment inside the relevant substation. The image collection table of cable equipment in medium-sized substations can be used to obtain the image acquisition status of cable equipment inside the relevant substation. Low-voltage cables include 0.6 / 1kV, medium-voltage cables include 3.6 / 6kV and 6 / 10kV, and high-voltage cables include 35kV, 110kV and above. Different image acquisition cycles can be set based on different cable types to improve the self-inspection effect.

[0116] Table 1. Image collection of cable equipment in medium-sized substations

[0117]

[0118] A database building module, based on the data obtained by the equipment image collection module and according to the location information of the substation, builds an original image database;

[0119] The database establishment module establishes the original image database based on the data obtained by the equipment image collection module and the location information of the substation, and establishes and classifies files for different equipment at different locations. At the same time, it receives the judgment-related data of the abnormal judgment module and supplements the relevant files;

[0120] The image two-dimensional processing module, based on the image data in the original image database, extracts the contour and removes the background of the image to be detected through the threshold, converts the image into two dimensions through the computer vision library, and transmits the relevant data to the abnormality judgment module;

[0121] Based on the image data in the original image database, the contour extraction and background stripping of the image to be detected are performed through the threshold, the image is converted into two dimensions through the computer vision library, and the relevant data is transmitted to the abnormality judgment module, including the following steps:

[0122] Grayscale processing is performed on the image data in the original image database, and the background of the original image is stripped by setting the grayscale threshold, and the stripped image is marked as the image to be extracted;

[0123] The image data in the original image database is grayed out, and the background of the original image is stripped by setting the gray threshold. At the same time, the stripped image is marked as the image to be extracted. The specific steps and algorithm formula are as follows:

[0124] For the color images in the original image database, traverse each pixel, substitute the RGB value of the relevant image into the grayscale weighted average formula to calculate the grayscale value, and convert the color images in the original image database into grayscale images;

[0125] According to the converted grayscale image, the grayscale threshold is set by Otsu's method and the background of the grayscale image is stripped. The specific steps are as follows:

[0126] The algorithm formula for statistical image gray value distribution is:

[0127] ;

[0128] in, Represents gray value The pixel probability, Represents gray value The number of pixels, The total number of pixels in the image is calculated by assuming a grayscale threshold , divide the image pixels into foreground and background, and calculate the weight of the foreground class and the weight of the background class;

[0129] Among them, the weight of the foreground class , that is, the gray value is less than or equal to , and its algorithm formula is:

[0130] ;

[0131] Weight of background class , that is, the gray value is greater than , and its algorithm formula is:

[0132] ;

[0133] Calculate the grayscale average of the foreground class , and its algorithm formula is:

[0134] ;

[0135] Calculate the grayscale average of the background class , and its algorithm formula is:

[0136] ;

[0137] Calculate the total average gray value of the image , and its algorithm formula is:

[0138] ;

[0139] Weights based on foreground class , weight of background class , the average grayscale value of the foreground class , the grayscale average of the background class And the total average gray value , calculate the between-class variance , and its algorithm formula is:

[0140] ;

[0141] By between-class variance Value, determine the optimal threshold , and its algorithm formula is:

[0142] ;

[0143] It should be noted that Usually refers to the range of gray levels in an image. The pixel values ​​in a grayscale image usually range from 0 to ,in is the total number of gray levels. For an 8-bit grayscale image, the gray levels range from 0 to 255. , Indicates the maximum value of the available grayscale value in a grayscale image, usually used as a boundary for loops or calculations.

[0144] By the optimal threshold , for grayscale image background stripping, the specific steps are:

[0145] ;

[0146] in, Represents the binary image at position The pixel value at Represents the grayscale image at position The pixel value at .

[0147] It should be noted that by combining the optimal threshold For background and foreground separation, the foreground object is clearly separated and the background information is removed, making the foreground extraction and subsequent processing simpler and more intuitive.

[0148] The edge detection algorithm is used to detect the contour of the image to be extracted, and the computer vision library is used to convert the image into two dimensions to obtain the two-dimensional line image of the relevant equipment. The specific steps are as follows:

[0149] Through the Sobel operator, edge detection is performed on the relevant image after background stripping to obtain the device outline. The specific steps are as follows:

[0150] ;

[0151] ;

[0152] In the above formula, is the Sobel operator kernel in the horizontal direction, is the Sobel operator kernel in the vertical direction. Based on the Sobel operator kernel in the horizontal direction and the Sobel operator kernel in the vertical direction, the horizontal gradient direction and the vertical gradient direction are calculated:

[0153] ;

[0154] ;

[0155] in, represent The horizontal gradient calculated at the coordinates, represent The vertical gradient calculated at the coordinate, Represents the grayscale image at position The pixel value at ;

[0156] Calculate the gradient magnitude With direction :

[0157] ;

[0158] ;

[0159] The edge points of the device in the image are further determined by non-maximum suppression and double threshold processing. The specific steps are as follows:

[0160] For each pixel, its gradient magnitude is compared with the gradient magnitudes of two adjacent pixels along the gradient direction. If the gradient magnitude of the selected pixel is not the local maximum, the selected pixel is set to 0, otherwise it remains unchanged.

[0161] By setting a high threshold and a low threshold, pixels with a gradient amplitude greater than the high threshold are determined as strong edge points, pixels with a gradient amplitude less than the low threshold are determined as non-edge points, and pixels between the high threshold and the low threshold are determined as strong edge points when connected to strong edge points, and as non-edge points when connected to non-edge points;

[0162] It should be noted that non-maximum suppression can refine the edges and remove the pixels that are not local maxima caused by noise or other factors. For power grid equipment images, it helps to highlight the main edge features of the equipment and remove some blurred or unclear edges, making the outline of the equipment clearer. This makes it easier to capture the shape and structure of the equipment more accurately and retain the original appearance of the equipment during the subsequent two-dimensional conversion.

[0163] Through double threshold processing, pixels with gradient amplitude greater than the high threshold are determined as strong edge points, pixels with gradient amplitude less than the low threshold are determined as non-edge points, and pixels between the two are determined as edge points based on their connection with the strong edge points, so as to effectively remove false edges caused by noise while retaining the true device edges. The high threshold and the low threshold can be flexibly set by consulting experts in related fields and according to usage.

[0164] pass In Extract contours from edge pixels and organize them into a series of contours through edge points. Each contour is a continuous sequence of points, forming a closed or open curve.

[0165] pass Generate a completely black image of the same size as the original image, draw the extracted contour lines on the new image, generate a two-dimensional line image, and transmit the two-dimensional line image to the abnormality judgment module.

[0166] Embodiment 2:

[0167] The abnormality judgment module performs line detection on the two-dimensional line image based on the device type of the original image, judges the device abnormality, transmits the abnormal signal into the original image database, and marks the relevant device files. The specific steps are as follows:

[0168] Through the original image database, the types of power grid equipment in the relevant two-dimensional line images are obtained, including transformers, cables, and power poles and towers;

[0169] Based on different types of power grid equipment, different line detection methods are used for two-dimensional line images to determine the damage of power grid equipment, including the following steps:

[0170] For the two-dimensional line image, a coordinate system conversion method is adopted, the lower left corner of the two-dimensional line image is set as the origin of the coordinate system, and a plane rectangular coordinate system is established with the horizontal direction of the two-dimensional line image as the axis and the vertical direction as the axis;

[0171] For cables, obtain the position coordinates of the cables in the rectangular coordinate system through the two-dimensional line image and , the curvature of the relevant cable lines is calculated by the curvature calculation formula, and the algorithm formula is:

[0172] ;

[0173] in, Represents the curve at point The first-order derivative in the horizontal direction and the first-order derivative in the vertical direction, Represents the curve at point The horizontal second-order derivative and the vertical second-order derivative of set the curvature threshold The steps to determine the bending degree of the cable are as follows:

[0174] when ≤ , indicating that the bending of the relevant cable segment is normal;

[0175] when > , which means that the relevant cable segment is abnormally bent and there is a fault;

[0176] At the same time, for each point on the line, the distance between it and its adjacent points is calculated. The algorithm formula is:

[0177] ;

[0178] in, Represents two points , The distance between Determine the break of the line:

[0179] when ≤ , indicating that the relevant cable segment is normal;

[0180] when > , which means that the continuity of the relevant cable segment is abnormal and there is a break fault;

[0181] It should be noted that for overhead transmission cables, the two-dimensional line pattern can detect problems such as looseness and broken strands of the conductor. If the cable is loose, the line will become more curved. If the cable has broken strands, the line will become discontinuous. By performing threshold comparison on the relevant positions in the two-wire line pattern, cable faults can be quickly discovered. The threshold setting can refer to the manual requirements of the relevant project and equipment, or be set by consulting experts in related fields.

[0182] For the power tower, the distance between each point on the line and its adjacent points is calculated, and the continuity is determined by combining the threshold. The steps are as follows:

[0183] ;

[0184] in, Represents two points , The distance between Determine the break of the line:

[0185] when ≤ , indicating that the relevant power tower section is normal;

[0186] when > , which means that the continuity of the relevant power tower section is abnormal and there is a fracture fault;

[0187] Calculate the angle of the main line of the power tower by fitting a straight line to the contour line, and use linear regression to fit the straight line to get the slope of the straight line and the intercept , calculate the angle of the line :

[0188] ;

[0189] Set the tilt angle threshold , comparing the line angles of different parts of the power tower , for abnormal determination of power towers:

[0190] when ≤ , indicating that the relevant power tower section is normal;

[0191] when > , which means that the relevant power tower section is abnormal and there is a risk of tilting;

[0192] It should be noted that the two-dimensional line pattern can clearly show the overall shape and structure of the tower. If the tower is tilted, the lines will be asymmetric or the angle will change. If the tower is broken, the lines will be obviously interrupted. By analyzing the lines, the deformation and fracture of the tower can be discovered in time to ensure the safe and stable operation of the power line. At the same time, a tower condition monitoring record sheet is generated to facilitate the inspection of the tower condition.

[0193] Table 2 Tower Condition Monitoring Record

[0194]

[0195] For the transformer, for the two-dimensional line pattern, draw the reference line according to the normal contour, calculate the distance between the actual line and the reference line, and judge the expansion and depression of the transformer shell:

[0196] ;

[0197] in, is the vertical coordinate of the point on the actual contour line, is the vertical coordinate of the corresponding point on the reference line, based on the judgment threshold , for transformer abnormality determination:

[0198] when ≤ , indicating that the external part of the relevant transformer is normal;

[0199] when > , which means the external part of the relevant transformer is abnormal, with depression or expansion;

[0200] For each point on the line, the distance between it and its adjacent point is calculated, and the integrity of the transformer casing is determined by combining the threshold. The steps are as follows:

[0201] ;

[0202] in, Represents two points , The distance between Determine the break of the line:

[0203] when ≤ , representing that the external part of the relevant transformer is intact;

[0204] when > , indicating that the relevant transformer has external abnormalities and is at risk of breakage.

[0205] It should be noted that the deformation of the transformer shell will affect the normal operation and safety of the transformer. The two-dimensional line pattern can be used to detect the expansion and depression of the shell. If the shell expands, the lines will bulge outward, and if the shell is depressed, the lines will be depressed inward. By monitoring the shell lines, the deformation of the transformer shell can be discovered in time and corresponding measures can be taken to deal with it. It is necessary to select an appropriate threshold. If the threshold is too small, it may misreport minor deformations within the normal range. If the threshold is too large, it may miss obvious deformations. The appropriate threshold can be determined through actual measurements and historical data. .

[0206] Based on the results of the power grid equipment damage determination, the abnormal signal is transmitted into the original image database, and the relevant equipment files are marked, and the relevant abnormal files are marked in red. The specific steps are as follows:

[0207] Based on the results of the power grid equipment damage judgment, when there is an abnormality in the power grid equipment, the abnormal signal is transmitted to the database establishment module, and the relevant equipment files in the relevant original image database are marked in red;

[0208] When the power grid equipment is judged to be normal, the transmission judgment related data is entered into the database establishment module, and the data is stored and backed up in the relevant equipment files in the relevant original image database, and the date is marked;

[0209] The abnormality determination module is also provided with a cable damage tendency prediction system, which includes the following steps:

[0210] Based on the relevant cable files in the original image database, the judgment result data is compared, the detection data at the same position is compared at intervals, and the damage tendency is predicted. The specific steps are as follows:

[0211] get The curvature of and The curvature of , and the time interval , calculate the curvature change rate, the algorithm formula is:

[0212] ;

[0213] ;

[0214] in, Represents the current detection time. For The current time moves forward 2 times. Represents the curvature change rate, for the curvature change rate threshold , predict the damage tendency:

[0215] when > , which means the curvature change rate is large and there is a tendency to damage. The relevant files are marked in orange:

[0216] when > , which means that the curvature change rate is large and there is a tendency to be damaged. The relevant files are marked in orange.

[0217] It should be noted that by calculating the curvature change rate, the curvature change of the relevant cable position within three detection cycles can be determined, so as to predict the damage tendency of the cable and take timely measures to prevent cable failure. The curvature change rate threshold can be flexibly set according to the cable weight, installation location and basic parameters.

[0218] The surface damage detection system for power grid equipment based on image analysis of the present invention collects images of power grid equipment in substations, including cables, transformers, and power poles and towers, through fixed cameras at regular intervals during use, and establishes an original image database, which includes archives of various equipment;

[0219] By analyzing and processing the original image, the foreground and background of the equipment in the original image are stripped, and the equipment outline is extracted to generate a two-dimensional line image. Based on the two-dimensional line image, damage detection is performed on different equipment;

[0220] Record the relevant data of each damage detection, combine it with the previous data in the files of each equipment, and predict the failure of the transmission cable, so that operators can make maintenance plans in advance and arrange resources reasonably.

[0221] In the embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the modules is only a logical function division, and there may be other division methods in actual implementation; the modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the method of this embodiment.

[0222] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A power grid equipment surface damage detection system based on image analysis, characterized by: It includes equipment image collection module, database establishment module, image two-dimensional processing module and abnormality determination module; The equipment image collection module is used to collect images of substation power grid equipment, including cables, transformers, and power poles and towers; The database establishment module establishes an original image database based on the data obtained by the device image collection module and according to the location information of the substation; The image two-dimensional processing module, based on the image data in the original image database, performs contour extraction and background stripping on the image to be detected through a threshold, performs two-dimensional conversion on the image through a computer vision library, and transmits relevant data to the abnormality determination module, including the following steps: The image data in the original image database is grayed out, and the background of the original image is stripped by setting the gray threshold. At the same time, the stripped image is marked as the image to be extracted. The specific steps and algorithm formula are as follows: For the color images in the original image database, traverse each pixel, substitute the RGB value of the relevant image into the grayscale weighted average formula to calculate the grayscale value, and convert the color images in the original image database into grayscale images; According to the converted grayscale image, the grayscale threshold is set by Otsu's method and the background of the grayscale image is stripped. The specific steps are as follows: The algorithm formula for statistical image gray value distribution is: ； in, Represents gray value The pixel probability, Represents gray value The number of pixels, Represents the total number of pixels in the image, by assuming a grayscale threshold , divide the image pixels into foreground and background, and calculate the weight of the foreground class and the weight of the background class; Among them, the weight of the foreground class , that is, the gray value is less than or equal to , and its algorithm formula is: ； Weight of background class , that is, the gray value is greater than , and its algorithm formula is: ； Calculate the grayscale average of the foreground class , and its algorithm formula is: ； Calculate the grayscale average of the background class , and its algorithm formula is: ； Calculate the total average gray value of the image , and its algorithm formula is: ； Weights based on foreground class , weight of background class , the average grayscale value of the foreground class , the grayscale average of the background class And the total average gray value , calculate the between-class variance , and its algorithm formula is: ； By between-class variance Value, determine the optimal threshold , and its algorithm formula is: ； By the optimal threshold , for grayscale image background stripping, the specific steps are: ； in, Represents the binary image at position The pixel value at Represents the grayscale image at position The pixel value at ; The edge detection algorithm is used to detect the contour of the image to be extracted, and the computer vision library is used to convert the image into two dimensions to obtain the two-dimensional line image of the relevant equipment. The specific steps are as follows: Through the Sobel operator, edge detection is performed on the relevant image after background stripping to obtain the device outline. The specific steps are as follows: ； ； In the above formula, is the Sobel operator kernel in the horizontal direction, is the Sobel operator kernel in the vertical direction. Based on the Sobel operator kernel in the horizontal direction and the Sobel operator kernel in the vertical direction, the horizontal gradient direction and the vertical gradient direction are calculated: ； ； in, represent The horizontal gradient calculated at the coordinates, represent The vertical gradient calculated at the coordinate, Represents the grayscale image at position The pixel value at ; Calculate the gradient magnitude With direction : ； ； The edge points of the device in the image are further determined by non-maximum suppression and double threshold processing. The specific steps are as follows: For each pixel, its gradient magnitude is compared with the gradient magnitudes of two adjacent pixels along the gradient direction. If the gradient magnitude of the selected pixel is not the local maximum, the selected pixel is set to 0, otherwise it remains unchanged. By setting a high threshold and a low threshold, pixels with a gradient amplitude greater than the high threshold are determined as strong edge points, and pixels with a gradient amplitude less than the low threshold are determined as non-edge points. For pixels between the high threshold and the low threshold, if they are connected to strong edge points, they are determined as strong edge points, and if they are connected to non-edge points, they are determined as non-edge points. pass In Extract contours from edge pixels and organize them into a series of contours through edge points. Each contour is a continuous sequence of points, forming a closed or open curve. pass Generate a completely black image of the same size as the original image, draw the extracted contour lines on the new image, generate a two-dimensional line image, and transmit the two-dimensional line image to the abnormality judgment module; The abnormality determination module performs line detection on the two-dimensional line image based on the device type of the original image, determines the device abnormality, transmits the abnormality signal into the original image database, and marks the relevant device files.

2. The power grid equipment surface damage detection system based on image analysis according to claim 1 is characterized in that: The abnormality determination module performs line detection on the two-dimensional line image based on the device type of the original image, determines the device abnormality, transmits the abnormal signal into the original image database, and marks the relevant device files. The specific steps are as follows: Through the original image database, the types of power grid equipment in the relevant two-dimensional line images are obtained, including transformers, cables, and power poles and towers; Based on different types of power grid equipment, different line detection methods are used for two-dimensional line images to determine the damage of power grid equipment; Based on the results of the power grid equipment damage determination, the abnormal signal is transmitted into the original image database, the relevant equipment files are marked, and the relevant abnormal files are marked in red.

3. The power grid equipment surface damage detection system based on image analysis according to claim 2 is characterized in that: The method of using different line detection methods for two-dimensional line images based on different types of power grid equipment to determine the damage of power grid equipment includes the following steps: For the two-dimensional line image, a coordinate system conversion method is adopted, the lower left corner of the two-dimensional line image is set as the origin of the coordinate system, and a plane rectangular coordinate system is established with the horizontal direction of the two-dimensional line image as the axis and the vertical direction as the axis; For cables, obtain the position coordinates of the cables in the rectangular coordinate system through the two-dimensional line image and , the curvature of the relevant cable lines is calculated by the curvature calculation formula, and the algorithm formula is: ； in, Represents the curve at point The horizontal and vertical first-order derivatives of Represents the curve at point The horizontal second-order derivative and the vertical second-order derivative of set the curvature threshold The steps to determine the bending degree of the cable are as follows: when ≤ , indicating that the bending of the relevant cable segment is normal; when > , which means that the relevant cable segment is abnormally bent and there is a fault; At the same time, for each point on the line, the distance between it and its adjacent points is calculated. The algorithm formula is: ； in, Represents two points , The distance between Determine the break of the line: when ≤ , indicating that the relevant cable segment is normal; when > , which means that the continuity of the relevant cable segment is abnormal and there is a break fault; For the power tower, the distance between each point on the line and its adjacent points is calculated, and the continuity is determined by combining the threshold. The steps are as follows: ； in, Represents two points , The distance between Determine the break of the line: when ≤ , indicating that the relevant power tower section is normal; when > , which means that the continuity of the relevant power tower section is abnormal and there is a fracture fault; Calculate the angle of the main line of the power tower by fitting a straight line to the contour line, and use linear regression to fit the straight line to get the slope of the straight line and the intercept , calculate the angle of the line : ； Set the tilt angle threshold , comparing the line angles of different parts of the power tower , for abnormal determination of power towers: when ≤ , indicating that the relevant power tower section is normal; when > , which means that the relevant power tower section is abnormal and there is a risk of tilting; For transformers, for two-dimensional line patterns, draw reference lines according to the normal contours and calculate the distance between the actual lines and the reference lines. , to determine the expansion and depression of the transformer casing: ； in, is the vertical coordinate of the point on the actual contour line, is the vertical coordinate of the corresponding point on the reference line, based on the judgment threshold , for transformer abnormality determination: when ≤ , indicating that the external part of the relevant transformer is normal; when > , which means the external part of the relevant transformer is abnormal, with depression or expansion; For each point on the line, the distance between it and its adjacent point is calculated, and the integrity of the transformer casing is determined by combining the threshold. The steps are as follows: ； in, Represents two points , The distance between Determine the break of the line: when ≤ , representing that the external part of the relevant transformer is intact; when > , indicating that the relevant transformer has external abnormalities and is at risk of breakage.

4. The power grid equipment surface damage detection system based on image analysis according to claim 2 is characterized in that: Based on the result of the power grid equipment damage determination, the abnormal signal is transmitted into the original image database, the relevant equipment files are marked, and the relevant abnormal files are marked in red. The specific steps are: Based on the results of the power grid equipment damage judgment, when there is an abnormality in the power grid equipment, the abnormal signal is transmitted to the database establishment module, and the relevant equipment files in the relevant original image database are marked in red; When the power grid equipment is judged to be normal, the transmission judgment related data enters the database establishment module, and the data is stored and backed up in the relevant equipment files in the relevant original image database, and the date is marked.

5. The power grid equipment surface damage detection system based on image analysis according to claim 4 is characterized in that: The abnormality determination module is also provided with a cable damage tendency prediction system, which includes the following steps: Based on the relevant cable files in the original image database, the judgment result data is compared, the detection data at the same position is compared at intervals, and the damage tendency is predicted. The specific steps are as follows: get The curvature of and The curvature of , and the time interval , calculate the curvature change rate, the algorithm formula is: ； ； in, Represents the current detection time. For The current time moves forward 2 times. Represents the curvature change rate, for the curvature change rate threshold , predict the damage tendency: when > , which means that the curvature change rate is large and there is a tendency to be damaged. The relevant files are marked in orange.

6. The power grid equipment surface damage detection system based on image analysis according to claim 1 is characterized in that: The equipment image collection module includes a fixed camera for regularly collecting images of substation power grid equipment, including cables, transformers, and power poles, and transmitting the data into the database establishment module; The database establishment module establishes an original image database based on the data obtained by the equipment image collection module and the location information of the substation, and creates and classifies files for different equipment at different locations. At the same time, it receives the judgment-related data of the abnormal judgment module and supplements the relevant files.

Images