High-voltage bushing positioning system and method for transformer replacement traction

Abstract

The invention discloses a transformer replacement traction high-voltage bushing positioning system and method, and the system comprises a tail end image positioning monitoring detector which is disposed at the top end of a high-voltage bushing and is used for collecting the images of the high-voltage bushing and a GIS conductor; the mounting frame is used for fixing the tail end image positioning monitoring detector on the high-voltage bushing; the target is configured to be fixed on the GIS conductor and is used for providing a visual reference point; wherein the tail end image positioning monitoring detector measures the deviation between the axis of the high-voltage bushing and the axis of the GIS conductor through an image processing algorithm; according to the method, universal adaptation of the main transformers of different models in the whole watershed is achieved, automatic calling can be achieved in follow-up operation only by inputting parameters after first-time input, and the problems that tools need to be customized and adaptability is poor in a traditional method are solved.

Description

Technical Field

[0001] This invention belongs to the field of transformer installation technology, and in particular relates to a high-voltage bushing positioning system and method for replacing traction bushings in transformers. Background Technology

[0002] In the operation and maintenance of large hydropower stations, the main transformer, as the core equipment for power conversion and transmission, requires regular maintenance and replacement, which is a crucial and technically complex task. According to the industry standard "DL / T 573-2010 Guidelines for the Maintenance of Power Transformers," when the main transformer needs to be removed from the powerhouse for maintenance, it must be pulled to the maintenance position using a dedicated track transport system. After maintenance is completed, it is precisely pulled back to its installation position along the original track, ensuring accurate connection between the A / B / C phase bushings on the high-voltage side and the corresponding conductors in the GIS pipeline. The precision requirements for this final connection are extremely stringent, typically requiring the deviation of the center axis in both the lateral and longitudinal directions to be controlled within ±10 mm. Any deviation beyond this range can lead to stress at the electrical connection, seal failure, or partial discharge, posing serious safety hazards. Therefore, achieving high-precision and high-efficiency final positioning of the main transformer, a heavy-duty piece of equipment weighing hundreds of tons, in the complex and space-constrained environment of a hydropower station powerhouse has always been a key technical challenge that urgently needs to be solved in this field.

[0003] Currently, the closest and most widely adopted technology in the industry is a mechanical measurement system relying on purely manual operation. This technical solution is implemented as follows: First, a frame level is used to calibrate the track plane and establish a horizontal reference. Then, nylon plumb lines are used to establish a measurement reference. One set of plumb lines is suspended directly above the track wheels of the main transformer transfer trolley to monitor lateral misalignment between the wheel and rail, i.e., rail wear. The other set of plumb lines is fixed directly above the center of the GIS conductor and the main transformer's high-voltage bushing, and adjusted to sag so that they are on the same horizontal plane. Throughout the traction process, operators must stop the machine after moving a small distance, manually reading and recording the projection distance of each set of plumb lines on the track surface or marking plate using a steel ruler or steel tape measure. The offset is calculated, guiding the adjustment of the traction rope length to correct the main transformer's travel posture and final position.

[0004] While this traditional method supported the operation and maintenance of power plants for a certain period of history, its inherent technical limitations have become glaringly apparent in the context of modern industry's pursuit of "high precision, high efficiency, and high safety," revealing numerous shortcomings: Low and unstable measurement accuracy: When the core measurement benchmark plumb line is suspended at a height exceeding half a meter, it is highly susceptible to slight swaying caused by vibrations from on-site equipment and airflow disturbances, leading to baseline drift. Simultaneously, in confined spaces such as the bottom of the main transformer (only about 500 mm high), it is difficult for surveyors to maintain a perpendicular line of sight to the steel ruler's scale, inevitably causing parallax and reading errors. The combination of these factors makes it difficult to consistently guarantee overall measurement accuracy, with actual deviations often exceeding the tolerance.

[0005] The operation efficiency is extremely low: In order to achieve coplanar end faces, the entire traction process requires frequent shutdowns. If the final docking fails, rework is required. Just deploying the traction rigging in reverse may take more than half a day, seriously delaying the maintenance schedule.

[0006] Poor spatial adaptability and operability: The operating space at the bottom of the main transformer is low and narrow, and the ground is often paved with uneven cobblestones, making it difficult for personnel to move around and place tools. The high-voltage riser pipes at the top of the main transformer are densely distributed and very closely spaced, which severely restricts the position and movement of the workers, making operations such as hanging plumb lines and marking extremely inconvenient and easily obstructed.

[0007] Lack of dynamic monitoring capability: This technology is a discrete point measurement with an extremely low data update frequency. It cannot capture minute instantaneous deviations caused by track joints and other factors during the traction interval, resulting in data blind spots and potentially causing delays in correction commands.

[0008] Poor economic efficiency and safety risks: Frequent downtime leads to long idle waiting times for traction equipment, resulting in low equipment utilization and requiring significant time and labor costs for each operation. Furthermore, to ensure the plumb line is coplanar, one person must hold a plumb bob below the GIS while another marks it on top of the main transformer. Working on a wet, slippery surface with residual insulating oil poses a high risk of tools falling and personnel slipping.

[0009] In summary, existing technologies rely entirely on manual experience and mechanical measuring tools, and their technical means are significantly lagging behind the development level of modern industrial measurement. They cannot meet the urgent needs for millimeter-level accuracy, high-efficiency processes, and high-level safety assurance in the replacement of main transformers in large hydropower stations. Therefore, it is necessary to propose a high-voltage bushing positioning system and method for transformer replacement traction to solve the above problems. Summary of the Invention

[0010] The technical problem to be solved by the present invention is to provide a high-voltage bushing positioning system and method for transformer replacement traction, which aims to solve the problem that the existing technology relies entirely on manual experience and mechanical measuring tools, which has serious deficiencies in accuracy and efficiency, and poses many safety risks.

[0011] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a high-voltage bushing positioning system and method for replacing traction bushings in transformers, the system comprising: The end-point image positioning and monitoring detector is configured to be deployed at the top of the high-voltage bushing to acquire images of the high-voltage bushing and GIS conductor; Mounting bracket is used to fix the end image positioning and monitoring detector on the high-voltage bushing; The target is configured to be fixed on the GIS conductor to provide a visual reference point; The end-image positioning monitoring detector measures the deviation between the high-voltage bushing axis and the GIS conductor axis using an image processing algorithm.

[0012] Preferably, the mounting frame includes a cross structure composed of four telescopic rods, with the monitoring detector positioned at the center of the four telescopic rods; the telescopic rods are used to adjust the position of the end image positioning monitoring detector and are locked by a handle to prevent directional deviation.

[0013] Preferably, the mounting bracket further includes a tapered pin rotatably connected to the end of the telescopic rod for fixing to the mounting hole or extension of the high-pressure bushing riser.

[0014] Preferably, the target includes a weight to ensure that the target hangs naturally and is aligned with the center of the GIS conductor.

[0015] Preferably, the telescopic rod is a threaded sleeve structure, and the telescopic rod is provided with a scale, which is locked by a locking bolt.

[0016] Preferably, it also includes an image processing host, configured to process the images acquired by the end image positioning monitoring detector and calculate the deviation between the high-voltage bushing axis and the GIS conductor axis.

[0017] Preferably, the target includes a fixed plate, which is bolted to the threaded holes on the GIS conductor through the four corner slots; a weight is suspended at the center of the fixed plate and hangs naturally.

[0018] Preferably, the measurement method of the high-voltage bushing positioning system for transformer replacement traction includes the following steps: S1, The end image positioning and monitoring detector is deployed on the top of the high-voltage bushing via a mounting bracket; S2, fix the target on the GIS conductor; S3, the end image positioning and monitoring detector acquires an image containing the target and the high-voltage bushing; S4. Based on the acquired images, the spatial deviation between the high-voltage bushing axis and the GIS conductor axis is calculated using an image processing algorithm. S5. Adjust the position of the main transformer according to the deviation until the deviation is less than ±10mm.

[0019] Preferably, the image processing algorithm in step S4 includes image preprocessing of the acquired image, including grayscale conversion and image filtering and denoising, specifically: Image acquisition and input: Color images of the site are acquired by the end-point image positioning and monitoring detector. ,in The coordinates are pixel coordinates, with the y-axis representing the direction of motion of the main transformer and the x-axis representing the horizontal direction on the plane perpendicular to the direction of motion of the main transformer. Grayscale processing: Convert a color image to a grayscale image To reduce the computational load, a weighted average method is used, as shown in the following formula: ; in , , These are color images in The pixel values ​​of the red, green, and blue channels of the location; S103: Image filtering and denoising: Apply Gaussian filtering to grayscale images Smoothing is performed to suppress noise interference. The Gaussian kernel function is: ; in It's the standard deviation, which controls the smoothness; the filtered image: ; Here, * represents the convolution operation.

[0020] Preferably, the image processing algorithm also includes rapid initial localization and ROI extraction based on SIFT feature points, quickly and robustly determining the approximate area of ​​the target and high-voltage bushing, thus defining the range for subsequent precision measurements, specifically: (1) Define templates and initialize variables: Save a template image taken under ideal alignment. ; exist Precisely marked in the text: Target center point coordinates ; The center of the flange edge, i.e., the coordinates of the high-pressure bushing feature point ; Define variables: Used to store the current image A list of SIFT keypoints, where each keypoint kp contains its location. Scale and orientation; Used for storage The corresponding 128-dimensional SIFT descriptor matrix; , Similarly, it can be used to store template images. Key points and descriptors; Used to store high-quality feature point matching pairs after initial screening; (2) Feature point detection and descriptor calculation for the current image and template images Execute the SIFT algorithm separately: Constructing scale space: Detect extreme points to locate key points (x, y).

[0021] Assign direction to each key point Generate a 128-dimensional vector descriptor desc with rotation invariance; get and ; (3) Feature point matching: Using the k-nearest neighbor k-NN algorithm, for Each descriptor in, Find the two descriptors that are closest and second closest in Euclidean distance; Calculate the ratio of the nearest distance d1 to the second nearest distance d2. ; like If the match is less than the preset threshold, the match is accepted and added to the Good_Matches list; otherwise, it is considered a mismatch and is removed. (4) Geometric verification and calculation of projection transformation matrix: Extract the coordinate set of matching key points from the current image and the template image from Good_Matches. and ; Using the Random Sample Consensus algorithm, from and A 3x3 homography matrix H is estimated, which satisfies: ; in It is a point in the current image. These are the corresponding points in the template image, and ~ indicates equality in homogeneous coordinates; (5) Initial location estimation and ROI generation: Using the obtained homography matrix H, the known homography in the template image is... and Projecting these images onto the current image yields their initial estimated locations: ; ; Normalize the homogeneous coordinates: ; by and Centered on a rectangle, a fixed-size region is generated as the Region of Interest (ROI) for subsequent precise measurements. and .

[0022] Preferably, the image processing algorithm also includes sub-pixel edge precision localization based on Zernike moments. Within the calculated ROI, ultra-high precision edge point localization is performed to obtain the precise center of the target and the cannula. The specific method is as follows: (1) Pixel-level coarse edge extraction within the ROI: In respectively and Internally, the Canny edge detection algorithm is used to obtain a pixel-level set of edge points. and ; (2) Zernike moment template and parameter definition: Define a Zernike moment calculation template of size N x N, with the origin (x0, y0) located at its geometric center; The core kernel of the three complex kernels of the Zernike moment defined on the unit circle: V 00 V 11 V 20 ; Define variable: L is the subpixel distance from the edge point to the origin of the template; The angle is at the subpixel level in the direction of the edge point normal. (3) Calculate the Zernike moments: For each pixel-level edge point Take an N x N neighborhood image patch centered on it; Calculate the image patch relative to three kernels V 00 V 11 V 20 Zernike Matrix A 00 A 11 A 20 The calculation formula is: ; Where (x, y) are local coordinates within the template, and Conjugate represents taking the conjugate; (4) Solve for sub-pixel edge parameters: Using the calculated moments, solve for the boundary parameters: Edge direction angle: ; Distance from the edge to the origin: ; Calculate the sub-pixel level coordinate correction for this edge point: ; ; in It is a normalization factor related to the template size and the physical resolution of the image; Obtain the sub-pixel precise coordinates of the point: ; Perform S303 and S304 on all edge points to obtain a sub-pixel edge point set. and ; (5) Contour fitting and center calculation: For the target area: The points in the equation are fitted to an ellipse using the least squares method, and the center of the fitted ellipse is... That is, the center of the target after precise positioning. ; For the high-pressure bushing area: The points in the data are fitted to one or more straight lines to determine the boundary lines of the casing; based on the geometric relationships of the boundary lines, the characteristic points of the casing after precise positioning are calculated. ; It also includes data fusion and spatial bias calculation: (1) Pixel deviation calculation: Calculate the deviation between the target center and the cannula feature points in the image coordinate system: ; ; (2) Coordinate transformation to physical space: Using the intrinsic parameter matrix K and depth information Z obtained in advance through camera calibration, pixel deviation is converted into physical deviation in the world coordinate system; the depth information Z is a fixed installation distance. The conversion formula is: ; Final output physical space deviation ( This data is sent to the main control system to generate traction correction commands.

[0023] The beneficial effects of this invention are as follows: 1. Compared with traditional manual measurement methods, this invention achieves a revolutionary breakthrough in measurement accuracy. Existing technologies are limited by the physical swing of the plumb line and the parallax of manual readings, resulting in extremely unstable overall measurement accuracy, making it difficult to meet the stringent requirements of wheel-rail clearance. This invention eliminates these error sources in principle by introducing a feature-guided sub-pixel edge precision localization algorithm. First, initial localization is performed using SIFT feature point matching and the RANSAC algorithm, which is robust and unaffected by ambient lighting and minor occlusions, accurately locking in the Region of Interest (ROI) for precise measurement. Within the ROI, Zernike moment sub-pixel edge detection technology is used. By calculating the complex moments of the image neighborhood and substituting them into the formula, the sub-pixel position of the edge is precisely solved, improving the edge localization accuracy from the pixel level to the sub-pixel level. This technology stably controls the measurement accuracy of wheel-rail clearance within the error range, significantly improving accuracy and completely solving the problems of rework and safety risks caused by inaccurate measurements.

[0024] 2. This invention, through systematic technological innovation, completely reverses the inefficiency of traditional methods. Existing technologies rely on manual measurement at discrete points, resulting in extremely low efficiency. Furthermore, deployment, coplanar debugging, and potential rework further prolong the work cycle. This invention constructs a highly efficient end-to-end operation mode. By adopting a modular quick-installation design, all detectors are installed via tool-free quick-locking mechanisms, allowing a single person to quickly complete the deployment and calibration of the entire system, representing a qualitative leap in efficiency compared to traditional methods. The core measurement process achieves dynamic continuous monitoring, replacing the original "move-stop-measure" cycle, enabling near-uninterrupted operation of the main transformer traction process. This reduces downtime caused by measurement to zero, significantly saving labor costs and improving equipment utilization.

[0025] 3. Existing technologies rely entirely on traditional measuring tools and manual experience, resulting in outdated techniques and significant personal safety risks when operating at heights or in confined spaces. This invention integrates machine vision and edge computing to construct a non-contact, fully automated intelligent measurement system, greatly improving operational safety: operators no longer need to climb to the top of the transformer or perform dangerous visual and manual measurements in narrow spaces; all data can be wirelessly transmitted to the user terminal via cable, enabling remote and safe monitoring and operation, fundamentally eliminating risks such as falls from heights and falling objects. Attached Figure Description

[0026] Figure 1 This is a schematic diagram of the device structure of the present invention; Figure 2 This is a schematic diagram of the target structure in this invention; Figure 3This is a schematic diagram of the installation of the present invention. Detailed Implementation

[0027] Example 1: like Figures 1-3 As shown, a high-voltage bushing positioning system and method for transformer traction replacement includes: The end-point image positioning and monitoring detector is configured to be deployed at the top of the high-voltage bushing to acquire images of the high-voltage bushing and GIS conductor; Mounting bracket is used to fix the end image positioning and monitoring detector on the high-voltage bushing; The target is configured to be fixed on the GIS conductor to provide a visual reference point; The end-image positioning monitoring detector measures the deviation between the high-voltage bushing axis and the GIS conductor axis using an image processing algorithm.

[0028] Preferably, the mounting frame includes a cross structure composed of four telescopic rods, with the monitoring detector positioned at the center of the four telescopic rods; the telescopic rods are used to adjust the position of the end image positioning monitoring detector and are locked by a handle to prevent directional deviation.

[0029] Preferably, the mounting bracket further includes a tapered pin rotatably connected to the end of the telescopic rod for fixing to the mounting hole or extension of the high-pressure bushing riser.

[0030] Preferably, the target includes a weight to ensure that the target hangs naturally and is aligned with the center of the GIS conductor.

[0031] Preferably, the telescopic rod is a threaded sleeve structure, and the telescopic rod is provided with a scale, which is locked by a locking bolt.

[0032] Preferably, it also includes an image processing host, configured to process the images acquired by the end image positioning monitoring detector and calculate the deviation between the high-voltage bushing axis and the GIS conductor axis.

[0033] Preferably, the target includes a fixed plate, which is bolted to the threaded holes on the GIS conductor through the four corner slots; a weight is suspended at the center of the fixed plate and hangs naturally.

[0034] Preferably, the measurement method of the high-voltage bushing positioning system for transformer replacement traction includes the following steps: S1, The end image positioning and monitoring detector is deployed on the top of the high-voltage bushing via a mounting bracket; S2, fix the target on the GIS conductor; S3, the end image positioning and monitoring detector acquires an image containing the target and the high-voltage bushing; S4. Based on the acquired images, the spatial deviation between the high-voltage bushing axis and the GIS conductor axis is calculated using an image processing algorithm. S5. Adjust the position of the main transformer according to the deviation until the deviation is less than ±10mm.

[0035] Example 2: Preferably, the image processing algorithm in step S4 includes image preprocessing of the acquired image, including grayscale conversion and image filtering and denoising, specifically: Image acquisition and input: Color images of the site are acquired by the end-point image positioning and monitoring detector. ,in The coordinates are pixel coordinates, with the y-axis representing the direction of motion of the main transformer and the x-axis representing the horizontal direction on the plane perpendicular to the direction of motion of the main transformer. Grayscale processing: Convert a color image to a grayscale image To reduce the computational load, a weighted average method is used, as shown in the following formula: ; in , , These are color images in The pixel values ​​of the red, green, and blue channels of the location; S103: Image filtering and denoising: Apply Gaussian filtering to grayscale images Smoothing is performed to suppress noise interference. The Gaussian kernel function is: ; in It's the standard deviation, which controls the smoothness; the filtered image: ; Here, * represents the convolution operation.

[0036] Preferably, the image processing algorithm also includes rapid initial localization and ROI extraction based on SIFT feature points, quickly and robustly determining the approximate area of ​​the target and high-voltage bushing, thus defining the range for subsequent precision measurements, specifically: (1) Define templates and initialize variables: Save a template image taken under ideal alignment. ; exist Precisely marked in the text: Target center point coordinates ; The center of the flange edge, i.e., the coordinates of the high-pressure bushing feature point ; Define variables: Used to store the current image A list of SIFT keypoints, where each keypoint kp contains its location. Scale and orientation; Used for storage The corresponding 128-dimensional SIFT descriptor matrix; , Similarly, it can be used to store template images. Key points and descriptors; Used to store high-quality feature point matching pairs after initial screening; (2) Feature point detection and descriptor calculation for the current image and template images Execute the SIFT algorithm separately: Constructing scale space: Detect extreme points to locate key points (x, y).

[0037] Assign direction to each key point Generate a 128-dimensional vector descriptor desc with rotation invariance; get and ; (3) Feature point matching: Using the k-nearest neighbor k-NN algorithm, for Each descriptor in, Find the two descriptors that are closest and second closest in Euclidean distance; Calculate the ratio of the nearest distance d1 to the second nearest distance d2. ; like If the match is less than the preset threshold, the match is accepted and added to the Good_Matches list; otherwise, it is considered a mismatch and is removed. (4) Geometric verification and calculation of projection transformation matrix: Extract the coordinate set of matching key points from the current image and the template image from Good_Matches. and ; Using the Random Sample Consensus algorithm, from and A 3x3 homography matrix H is estimated, which satisfies: ; in It is a point in the current image. These are the corresponding points in the template image, and ~ indicates equality in homogeneous coordinates; (5) Initial location estimation and ROI generation: Using the obtained homography matrix H, the known homography in the template image is... and Projecting these images onto the current image yields their initial estimated locations: ; ; Normalize the homogeneous coordinates: ; by and Centered on a rectangle, a fixed-size region is generated as the Region of Interest (ROI) for subsequent precise measurements. and .

[0038] Preferably, the image processing algorithm also includes sub-pixel edge precision localization based on Zernike moments. Within the calculated ROI, ultra-high precision edge point localization is performed to obtain the precise center of the target and the cannula. The specific method is as follows: (1) Pixel-level coarse edge extraction within the ROI: In respectively and Internally, the Canny edge detection algorithm is used to obtain a pixel-level set of edge points. and ; (2) Zernike moment template and parameter definition: Define a Zernike moment calculation template of size N x N, with the origin (x0, y0) located at its geometric center; The core kernel of the three complex kernels of the Zernike moment defined on the unit circle: V 00 V 11 V 20 ; Define variable: L is the subpixel distance from the edge point to the origin of the template; The angle is at the subpixel level in the direction of the edge point normal. (3) Calculate the Zernike moments: For each pixel-level edge point Take an N x N neighborhood image patch centered on it; Calculate the image patch relative to three kernels V 00 V 11 V20 Zernike Matrix A 00 A 11 A 20 The calculation formula is: ; Where (x, y) are local coordinates within the template, and Conjugate represents taking the conjugate; (4) Solve for sub-pixel edge parameters: Using the calculated moments, solve for the boundary parameters: Edge direction angle: ; Distance from the edge to the origin: ; Calculate the sub-pixel level coordinate correction for this edge point: ; ; in It is a normalization factor related to the template size and the physical resolution of the image; Obtain the sub-pixel precise coordinates of the point: ; Perform S303 and S304 on all edge points to obtain a sub-pixel edge point set. and ; (5) Contour fitting and center calculation: For the target area: The points in the equation are fitted to an ellipse using the least squares method, and the center of the fitted ellipse is... That is, the center of the target after precise positioning. ; For the high-pressure bushing area: The points in the data are fitted to one or more straight lines to determine the boundary lines of the casing; based on the geometric relationships of the boundary lines, the characteristic points of the casing after precise positioning are calculated. ; It also includes data fusion and spatial bias calculation: (1) Pixel deviation calculation: Calculate the deviation between the target center and the cannula feature points in the image coordinate system: ; ; (2) Coordinate transformation to physical space: Using the intrinsic parameter matrix K and depth information Z obtained in advance through camera calibration, pixel deviation is converted into physical deviation in the world coordinate system; the depth information Z is a fixed installation distance. The conversion formula is: ; Final output physical space deviation ( This data is sent to the main control system to generate traction correction commands.

Claims

1. A high-voltage bushing positioning system for transformer traction replacement, characterized in that, include: The end-point image positioning and monitoring detector is configured to be deployed at the top of the high-voltage bushing to acquire images of the high-voltage bushing and GIS conductor; Mounting bracket is used to fix the end image positioning and monitoring detector on the high-voltage bushing; The target is configured to be fixed on the GIS conductor to provide a visual reference point; The end-image positioning monitoring detector measures the deviation between the high-voltage bushing axis and the GIS conductor axis using an image processing algorithm.

2. The high-voltage bushing positioning system for transformer traction replacement according to claim 1, characterized in that, The mounting frame includes a cross structure composed of four telescopic rods, with the monitoring detector positioned at the center of the four telescopic rods. The telescopic rods are used to adjust the position of the end image positioning monitoring detector and are locked by a handle to prevent directional deviation. The mounting frame also includes a tapered pin rotatably connected to the end of the telescopic rod for fixing to the mounting hole or extension of the high-pressure bushing riser pipe.

3. The high-voltage bushing positioning system for transformer traction replacement according to claim 1, characterized in that, The target includes a weight to ensure that the target hangs naturally and is aligned with the center of the GIS conductor.

4. The high-voltage bushing positioning system for transformer traction replacement according to claim 1, characterized in that, The telescopic rod is a threaded sleeve structure with graduations and is locked in place by a locking bolt.

5. The high-voltage bushing positioning system for transformer traction replacement according to claim 1, characterized in that, It also includes an image processing host, configured to process the images acquired by the end image positioning monitoring detector and calculate the deviation between the high-voltage bushing axis and the GIS conductor axis.

6. The high-voltage bushing positioning system for transformer traction replacement according to claim 1, characterized in that, The target includes a fixed plate, which is bolted to the threaded holes on the GIS conductor through the four corner slots; a weight is suspended at the center of the fixed plate and hangs naturally.

7. A measurement method for a high-voltage bushing positioning system for transformer replacement traction according to any one of claims 1-6, characterized in that, Includes the following steps: S1, The end image positioning and monitoring detector is deployed on the top of the high-voltage bushing via a mounting bracket; S2, fix the target on the GIS conductor; S3, the end image positioning and monitoring detector acquires an image containing the target and the high-voltage bushing; S4. Based on the acquired images, the spatial deviation between the high-voltage bushing axis and the GIS conductor axis is calculated using an image processing algorithm. S5. Adjust the position of the main transformer according to the deviation until the deviation is less than ±10mm.

8. A method for a transformer traction high-voltage bushing positioning system according to claim 7, characterized in that, The image processing algorithm in step S4 includes image preprocessing of the acquired image, including grayscale conversion and image filtering and denoising, specifically: Image acquisition and input: Color images of the site are acquired by the end-point image positioning and monitoring detector. ,in The coordinates are pixel coordinates, with the y-axis representing the direction of motion of the main transformer and the x-axis representing the horizontal direction on the plane perpendicular to the direction of motion of the main transformer. Grayscale processing: Convert a color image to a grayscale image To reduce the amount of calculation, a weighted average method is used, as shown in the following formula: ; in , , These are color images in The pixel values ​​of the red, green, and blue channels of the location; S103: Image filtering and denoising: Apply Gaussian filtering to grayscale images Smoothing is performed to suppress noise interference; the Gaussian kernel function is: ； in It's the standard deviation, which controls the smoothness; the filtered image: ； Here, * represents the convolution operation.

9. A method for a transformer traction high-voltage bushing positioning system according to claim 8, characterized in that, The image processing algorithm also includes rapid initial localization and ROI extraction based on SIFT feature points, quickly and robustly determining the approximate area of ​​the target and high-voltage bushing, thus defining the range for subsequent precision measurements. Specifically: (1) Define templates and initialize variables: Save a template image taken under ideal alignment. ; exist Precisely marked in the text: Target center point coordinates ; The center of the flange edge, i.e., the coordinates of the high-pressure bushing feature point ; Define variables: Used to store the current image A list of SIFT keypoints, where each keypoint kp contains its location. Scale and orientation; Used for storage The corresponding 128-dimensional SIFT descriptor matrix; , Similarly, it can be used to store template images. Key points and descriptors; Used to store high-quality feature point matching pairs after initial screening; (2) Feature point detection and descriptor calculation for the current image and template images Execute the SIFT algorithm separately: Constructing scale space: Detect extreme points to locate key points (x, y); Assign direction to each key point Generate a 128-dimensional vector descriptor desc with rotation invariance; get and ; (3) Feature point matching: Using the k-nearest neighbor k-NN algorithm, for Each descriptor in, Find the two descriptors that are closest and second closest in Euclidean distance; Calculate the ratio of the nearest distance d1 to the second nearest distance d2. ; like If the match is less than the preset threshold, the match is accepted and added to the Good_Matches list; otherwise, it is considered a mismatch and is removed. (4) Geometric verification and calculation of projection transformation matrix: Extract the coordinate set of matching key points from the current image and the template image from Good_Matches. and ; Using the Random Sample Consensus algorithm, from and A 3x3 homography matrix H is estimated, which satisfies: ； in It is a point in the current image. These are the corresponding points in the template image, and ~ indicates equality in homogeneous coordinates; (5) Initial location estimation and ROI generation: Using the obtained homography matrix H, the known homography in the template image is... and Projecting these images onto the current image yields their initial estimated locations: ； ； Normalize the homogeneous coordinates: ； by and Centered on a rectangle, a fixed-size region is generated as the Region of Interest (ROI) for subsequent precise measurements. and .

10. A method for a transformer traction high-voltage bushing positioning system according to claim 9, characterized in that, The image processing algorithm also includes sub-pixel edge precision localization based on Zernike moments. Within the calculated ROI, ultra-high precision edge point localization is performed to obtain the precise center of the target and cannula. The specific method is as follows: (1) Pixel-level coarse edge extraction within the ROI: In respectively and Internally, the Canny edge detection algorithm is used to obtain a pixel-level set of edge points. and ; (2) Zernike moment template and parameter definition: Define a Zernike moment calculation template of size N x N, with the origin (x0, y0) located at its geometric center; The core kernel of the three complex kernels of the Zernike moment defined on the unit circle: V 00 V 11 V 20 ; Define variable: L is the subpixel distance from the edge point to the origin of the template; The angle is at the subpixel level in the direction of the edge point normal. (3) Calculate the Zernike moments: For each pixel-level edge point Take an N x N neighborhood image patch centered on it; Calculate the image patch relative to three kernels V 00 V 11 V 20 Zernike Matrix A 00 A 11 A 20 The calculation formula is: ； Where (x, y) are local coordinates within the template, and Conjugate represents taking the conjugate; (4) Solve for sub-pixel edge parameters: Using the calculated moments, solve for the boundary parameters: Edge direction angle: ; Distance from the edge to the origin: ; Calculate the sub-pixel level coordinate correction for this edge point: ； ； in It is a normalization factor related to the template size and the physical resolution of the image; Obtain the sub-pixel precise coordinates of the point: ; Perform S303 and S304 on all edge points to obtain a sub-pixel edge point set. and ; (5) Contour fitting and center calculation: For the target area: The points in the equation are fitted to an ellipse using the least squares method, and the center of the fitted ellipse is... That is, the center of the target after precise positioning. ; For the high-pressure bushing area: The points in the data are fitted to one or more straight lines to determine the boundary lines of the casing; based on the geometric relationships of the boundary lines, the characteristic points of the casing after precise positioning are calculated. ; It also includes data fusion and spatial bias calculation: (1) Pixel deviation calculation: Calculate the deviation between the target center and the cannula feature points in the image coordinate system: ； ； (2) Coordinate transformation to physical space: Using the intrinsic parameter matrix K and depth information Z obtained in advance through camera calibration, pixel deviation is converted into physical deviation in the world coordinate system; the depth information Z is a fixed installation distance. The conversion formula is: ； Final output physical space deviation ( This data is sent to the main control system to generate traction correction commands.

Images