Non-contact visual dry-type transformer operation state monitoring device and method

Abstract

The invention provides a non-contact visual monitoring device and method for the running state of a dry-type transformer, and the device comprises an imaging system, a monitoring system, and a multi-point silicone oil-based fluorescent magnetofluid device disposed around a to-be-detected transformer. The multi-point silicone oil-based fluorescent magnetic fluid device is of a hollow structure composed of a plurality of insulated silicone oil-based fluorescent magnetic fluid units. A camera device of the imaging system can move in the horizontal direction and the vertical direction to shoot each silicone oil-based fluorescent magnetic fluid unit, and the monitoring system receives images of the silicone oil-based fluorescent magnetic fluid units at different positions uploaded by the imaging system in real time and processes the images in real time to form three-dimensional images of the multi-point silicone oil-based fluorescent magnetic fluid device. And the operation state of the tested transformer is judged. According to the transformer state monitoring and fault detection method, the operation state of the transformer can be reflected accurately, comprehensively and in real time, and if a fault occurs, the fault type can be determined and positioned quickly.

Description

Technical Field

[0001] This invention belongs to the technical field of dry-type transformer operation status monitoring, and particularly relates to a non-contact visual dry-type transformer operation status monitoring device and method. Background Technology

[0002] As a core device in the power grid for power distribution and voltage transformation, the operational reliability of dry-type transformers directly affects the safety and stability of the entire power system. If a transformer fails, it can not only cause localized power outages and result in significant economic losses, but in severe cases, it may also trigger secondary disasters such as fires, threatening personal safety and the safety of the power grid.

[0003] In actual operation, transformers are subjected to multiple stresses, including electrical, thermal, and mechanical stresses, over a long period, leading to gradual deterioration of their windings, insulation, and components, resulting in numerous faults. Common faults include: inter-turn or inter-layer short circuits caused by insulation aging and partial discharge in the windings; winding deformation due to unbalanced stress; core leakage flux or overheating; and increased contact resistance at connection points due to loosening or corrosion. If these faults are not detected in time, they can gradually develop and lead to catastrophic consequences such as transformer breakdown and burnout.

[0004] Currently, the main methods for condition monitoring and fault detection of transformers include: First, temperature sensor-based monitoring, which indirectly reflects local overheating problems such as overload and short circuit by placing thermocouples or infrared thermometers in key parts such as windings and cores. However, this method is a point-based measurement and cannot comprehensively perceive the overall condition, and it is not sensitive to early electromagnetic faults. Second, vibration and noise-based monitoring, which judges problems such as structural loosening and winding deformation by detecting vibration and sound signals during transformer operation. However, the on-site environment has large noise interference, making signal feature extraction difficult and easily affecting the accuracy of diagnosis. Third, electrical quantity-based monitoring, such as by analyzing... The transformer current and voltage waveforms and harmonic characteristics are used to infer faults, but the background harmonics of the power grid are complex and the noise has a significant impact, making it difficult to identify early and minor faults such as inter-turn short circuits and magnetic leakage. Fourth, detection is based on ultrasonic, ultraviolet, and ultra-high frequency signals generated when a fault occurs. These methods are mainly for partial discharge faults. For example, ultrasonic sensors can detect ultrasonic signals, ultraviolet imagers can detect ultraviolet photons, and ultra-high frequency sensors can detect ultra-high frequency signals. However, these methods are susceptible to interference, and the installation of sensors may require structural modifications to the equipment itself, introducing new sources of risk. In addition, the cost of signal acquisition and analysis equipment is high, making it difficult to popularize and apply them.

[0005] The transformer condition monitoring and fault detection methods described above are insufficient to accurately, comprehensively, and in real-time reflect the transformer's operating status. Furthermore, they cannot quickly determine the fault type and location if a fault occurs. Therefore, it is essential to propose a detection method that can accurately, comprehensively, and in real-time reflect the transformer's operating status and quickly determine the fault type and location. Summary of the Invention

[0006] Based on the shortcomings of the prior art, the present invention provides a non-contact visual monitoring device and method for the operating status of dry-type transformers. The device includes a multi-point silicone oil-based fluorescent magnetic fluid device arranged around the transformer. When a fault occurs, the magnetic field distribution in the space around the transformer will change, thereby causing the silicone oil-based fluorescent magnetic fluid device to exhibit different states, thus realizing real-time visual monitoring of the transformer's operating status. The present invention can accurately, comprehensively and in real time reflect the transformer's operating status.

[0007] To achieve the above-mentioned technical objectives, the present invention provides a non-contact visual monitoring device for the operating status of a dry-type transformer. The monitoring device includes an imaging system, a monitoring system, and a multi-point silicone oil-based fluorescent magnetic fluid device deployed around the transformer under test. The multi-point silicone oil-based fluorescent magnetic fluid device is a hollow structure composed of multiple insulating silicone oil-based fluorescent magnetic fluid units. The transformer under test is located in the hollow area of ​​the multi-point silicone oil-based fluorescent magnetic fluid device, and the height of the multi-point silicone oil-based fluorescent magnetic fluid device is greater than or equal to the height of the transformer under test. Each silicone oil-based fluorescent magnetic fluid unit includes a transparent encapsulation shell made of insulating material, silicone oil encapsulated within the transparent encapsulation shell, and fluorescent magnetic fluid suspended within the silicone oil. The fluorescent magnetic fluid forms an imaging point within the transparent encapsulation shell.

[0008] The imaging system includes a ring of tracks arranged around the multi-point silicone oil-based fluorescent magnetic fluid device, a horizontal walking mechanism installed on the tracks, a vertical lifting frame installed on the horizontal walking mechanism, and a camera device installed on the vertical lifting frame. The camera of the camera device faces the multi-point silicone oil-based fluorescent magnetic fluid device and is used to capture images of the fluorescent magnetic fluid in the multi-point silicone oil-based fluorescent magnetic fluid device. The lifting height of the vertical lifting frame is greater than or equal to the height of the multi-point silicone oil-based fluorescent magnetic fluid device.

[0009] The monitoring system is communicatively connected to the control systems of the horizontal walking mechanism, the vertical lifting frame, and the camera device. The monitoring system receives images of silicone oil-based fluorescent magnetic fluid units at different locations uploaded by the camera device in the imaging system in real time, labeling the images as p(x, y, z), where x, y, and z are the corresponding three-dimensional spatial coordinates. Simultaneously, it processes the images of silicone oil-based fluorescent magnetic fluid units at different locations in real time and determines the specific morphology of the magnetic fluid within each unit. By stitching together the processed images of silicone oil-based fluorescent magnetic fluid units at different locations, a three-dimensional image of the multi-point silicone oil-based fluorescent magnetic fluid device is formed, and then the operating status of the transformer under test is determined.

[0010] The preferred technical solution of this invention is as follows: The monitoring system monitors the transformer under test in real time and generates a three-dimensional image of a multi-point silicone oil-based fluorescent magnetic fluid device. The real-time generated three-dimensional image of the multi-point silicone oil-based fluorescent magnetic fluid device is compared with the three-dimensional images of the multi-point silicone oil-based fluorescent magnetic fluid device under different states of the transformer under test stored in the monitoring system, and the similarity between the images is calculated. When the similarity reaches a threshold, it is determined that the current operating state of the transformer under test is consistent with the transformer operating state corresponding to the three-dimensional image of the multi-point silicone oil-based fluorescent magnetic fluid device stored in the database. If the similarity does not reach the threshold, the transformer under test is disassembled and the cause and location of the fault are found. The three-dimensional image of the multi-point silicone oil-based fluorescent magnetic fluid device under the current state of the transformer under test is stored in the database. The similarity calculation algorithm between the two images includes structural similarity calculation and normalized evaluation based on coordinate distance.

[0011] The preferred technical solution of this invention is as follows: The monitoring system controls the vertical lifting frame to move the camera device up and down to capture images of silicone oil-based fluorescent magnetic fluid units at different heights, and determines the z-coordinate of the camera device when capturing images; the monitoring system controls the horizontal walking mechanism to move the camera device along the track to image silicone oil-based fluorescent magnetic fluid units at different horizontal positions, and determines the x and y coordinates of the camera device when capturing images; the core algorithm of the monitoring system to process the images p(x, y, z) of silicone oil-based fluorescent magnetic fluid units at different positions in real time to determine the specific morphology of the magnetic fluid within the silicone oil-based fluorescent magnetic fluid units at different positions includes contour extraction and centroid solving algorithms.

[0012] A preferred technical solution of the present invention is as follows: the imaging system further includes a base surrounding the transformer under test, and the track and the multi-point silicone oil-based fluorescent magnetohydrodynamic device are both mounted on the base.

[0013] The preferred technical solution of the present invention is as follows: the multi-point silicone oil-based fluorescent magnetic fluid device is hollow, and its shape matches the specifications and usage scenario of the transformer under test, and the transformer under test is surrounded in the middle; in the unfolded state of the multi-point silicone oil-based fluorescent magnetic fluid device, the multiple silicone oil-based fluorescent magnetic fluid units that make up the multi-point silicone oil-based fluorescent magnetic fluid device are arranged in an array.

[0014] The preferred technical solution of the present invention is as follows: the silicone oil-based fluorescent magnetic fluid unit is square, circular, elliptical, rectangular or other polygonal, and the transparent encapsulation shell of each silicone oil-based fluorescent magnetic fluid unit is made of transparent plastic, and adjacent silicone oil-based fluorescent magnetic fluid units are fixedly bonded together; the fluorescent magnetic fluid is black, the silicone oil is colorless, and the fluorescent magnetic fluid and silicone oil are insulating.

[0015] The preferred technical solution of the present invention is as follows: the horizontal walking mechanism is an electrically controlled automatic walking trolley, and the vertical lifting frame is an electric lifting frame; the monitoring system is connected to the control motor signals of the horizontal walking mechanism and the vertical lifting frame respectively, and the movement of the control motors of the horizontal walking mechanism and the vertical lifting frame is controlled by commands sent by the monitoring system.

[0016] To achieve the above-mentioned technical objectives, the present invention also provides a non-contact visual dry-type transformer operation status monitoring method. The method is based on the aforementioned non-contact visual dry-type transformer operation status monitoring device and specifically includes the following steps:

[0017] S1. Collect three-dimensional images of the multi-point silicone oil-based fluorescent magnetohydrodynamic device of the transformer under test under normal conditions and after operation under typical artificial faults, and store them in the database of the monitoring system as reference images.

[0018] S2. The monitoring system monitors the images of silicone oil-based fluorescent magnetic fluid units at different locations of the transformer under test in real time. After processing the images, the specific morphology of the magnetic fluid in each silicone oil-based fluorescent magnetic fluid unit is calculated and determined. The processed images of silicone oil-based fluorescent magnetic fluid units at different locations are then stitched together to form a three-dimensional image of a multi-point silicone oil-based fluorescent magnetic fluid device.

[0019] S3. Calculate the similarity between the real-time generated three-dimensional image of the multi-point fluorescent silicone oil-based magnetohydrodynamic device and the reference images of the transformer under test in different states stored in the monitoring system database;

[0020] S4. Determine the operating status of the transformer under test based on the similarity calculated in step S3. When the similarity reaches the threshold, the current operating status of the transformer under test is consistent with the operating status of the transformer corresponding to the pre-stored reference image in the database. If the similarity does not reach the threshold, disassemble the transformer under test and find the cause and location of the fault, and pre-store the three-dimensional image of the multi-point silicone oil-based fluorescent magnetohydrodynamic device under the current state of the transformer under test in the database. Complete the real-time visualization monitoring, fault detection and location of the operating status of the transformer under test.

[0021] The preferred technical solution of the present invention is as follows: the typical faults manually set in step S1 include winding short circuit, winding deformation, magnetic leakage, abnormal vibration and overheating; after the fault is determined in step S4, the three-dimensional image of the multi-point fluorescent silicone oil-based magnetofluid device corresponding to the fault is stored in the database of the monitoring system to improve the database of the monitoring system.

[0022] The preferred technical solution of this invention is as follows: In step S2, the monitoring system receives images of silicone oil-based fluorescent magnetic fluid units at different locations uploaded by the camera device in the imaging system in real time. The images are labeled as p(x, y, z), where x, y, and z are the three-dimensional spatial coordinates corresponding to the images. Image processing methods are used to process the images of silicone oil-based fluorescent magnetic fluid units at different locations in real time. Then, contour extraction and centroid solving algorithms are used to calculate and determine the specific morphology of the magnetic fluid within each silicone oil-based fluorescent magnetic fluid unit. The processed images of multiple silicone oil-based fluorescent magnetic fluid units are then stitched together to form a three-dimensional image of a multi-point silicone oil-based fluorescent magnetic fluid device. The image processing methods include grayscale conversion, Gaussian filtering, histogram equalization, and binarization.

[0023] The preferred technical solution of the present invention is as follows: the algorithm for calculating the similarity between the real-time generated three-dimensional image of the multi-point fluorescent silicone oil-based magnetohydrodynamic device and the reference image in step S3 includes structural similarity calculation and normalized evaluation based on coordinate distance.

[0024] The monitoring device in this invention consists of a multi-point silicone oil-based fluorescent magnetohydrodynamic (MHD) device. When a dry-type transformer is energized, a strong magnetic field is generated around it. This magnetic field alters the state of the MHD, and the changes in the MHD state at each monitoring point reflect the transformer's operating status. The core of this method lies in using a specially designed silicone oil-based fluorescent MHD device placed in the space surrounding the transformer. During transformer operation, the winding current and core magnetic flux generate a specific spatial magnetic field distribution. When a fault occurs, such as an inter-turn short circuit, winding deformation, or core leakage flux, the magnetic field distribution in the space surrounding the transformer changes, causing the silicone oil-based fluorescent MHD device to exhibit different states, thus achieving real-time visual monitoring of the transformer's operating status. This invention can accurately, comprehensively, and in real-time reflect the transformer's operating status. If, before the transformer is put into operation, various fault simulation experiments are conducted and the corresponding silicone oil-based fluorescent MHD device states are recorded, the type and location of the transformer fault can be determined.

[0025] Depending on the specifications of the transformer under test and its application scenario, the multi-point silicone oil-based fluorescent magnetic fluid device can be designed in different shapes. The cross-section of the multi-point silicone oil-based fluorescent magnetic fluid device can be formed into hollow squares, hollow ellipses, hollow circles, hollow gourd shapes, or other polygonal or curved shapes. The imaging system is arranged around the multi-point silicone oil-based fluorescent magnetic fluid device. The camera in the imaging system can be a high-definition, magnifiable camera. By changing the position of the camera, monitoring and imaging of all the multi-point silicone oil-based fluorescent magnetic fluid units within the device can be performed. The monitoring system processes the images of the multi-point silicone oil-based fluorescent magnetic fluid device uploaded by the imaging system and synthesizes them into a three-dimensional image. The multi-point fluorescent silicone oil-based magnetic fluid device exhibits different overall states under different operating conditions of the transformer under test. By comparing the similarity between the three-dimensional images of the multi-point silicone oil-based fluorescent magnetic fluid device under different operating conditions of the transformer under test pre-stored in the monitoring system database and the real-time image, the operating state of the transformer under test can be determined.

[0026] The transformer condition monitoring and fault detection method in this invention can accurately, comprehensively and in real time reflect the operating status of the transformer. If a fault occurs, it can also quickly determine the fault type and locate the fault. Attached Figure Description

[0027] Figure 1 This is a schematic diagram of the transformer operation status monitoring device in an embodiment of the present invention;

[0028] Figure 2 This is a top view of the transformer operation status monitoring device in an embodiment of the present invention;

[0029] Figure 3 This is a schematic diagram of the multi-point silicone oil-based fluorescent magnetofluid device in an embodiment of the present invention;

[0030] Figure 4 These are normal images and images of two different faults in the embodiments of the present invention;

[0031] Figure 5 This is a schematic diagram of the structure of the square insulating silicone oil-based fluorescent magnetofluid unit in an embodiment of the present invention;

[0032] Figure 6 This is a physical image of the square-insulated silicone oil-based fluorescent magnetofluid unit in an embodiment of the present invention;

[0033] Figure 7 This is a schematic diagram of the combined state of a square-insulated silicone oil-based fluorescent magnetohydrodynamic unit;

[0034] Figure 8 This is a schematic diagram of the structure of the elliptical silicone oil-based fluorescent magnetofluid unit in an embodiment of the present invention;

[0035] Figure 9 This is a schematic diagram of the combined state of an elliptical silicone oil-based fluorescent magnetofluid unit;

[0036] Figure 10 This is a schematic diagram of the square multi-point silicone oil-based fluorescent magnetofluid device in this invention;

[0037] Figure 11 This is a schematic diagram of the gourd-shaped multi-point silicone oil-based fluorescent magnetofluid device of the present invention;

[0038] Figure 12 This is a flowchart of the monitoring method in this invention;

[0039] Figure 13 This is a schematic diagram of the xy plane of the magnetofluid unit in an embodiment of the present invention.

[0040] In the figure: 1. Transformer under test; 2. Multi-point silicone oil-based fluorescent magnetic fluid device; 3. Insulated silicone oil-based fluorescent magnetic fluid unit; 301. Transparent encapsulation shell; 302. Silicone oil; 303. Fluorescent magnetic fluid; 4. Track; 5. Horizontal walking mechanism; 6. Vertical lifting frame; 7. Camera device; 8. Monitoring system; 9. Base. Detailed Implementation

[0041] The present invention will be further described below with reference to the accompanying drawings and embodiments. Figures 1 to 13 All accompanying drawings are simplified versions of embodiments and are intended only to clearly and concisely illustrate the embodiments of the present invention. The technical solutions shown in the drawings below are specific solutions of embodiments of the present invention and are not intended to limit the scope of the claimed invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0042] Example 1 provides a non-contact, visual monitoring device for the operating status of dry-type transformers, such as... Figure 1 and Figure 2 As shown, the monitoring device includes an imaging system, a monitoring system 8, and a multi-point silicone oil-based fluorescent magnetic fluid device 2 deployed around the transformer under test 1. The multi-point silicone oil-based fluorescent magnetic fluid device 2 is a hollow structure composed of multiple insulated silicone oil-based fluorescent magnetic fluid units 3. The shape of the multi-point silicone oil-based fluorescent magnetic fluid device 8 matches the specifications and usage scenario of the transformer under test 1, and surrounds the transformer under test 1 in the center. The transformer under test 1 is located in the hollow area of ​​the multi-point silicone oil-based fluorescent magnetic fluid device 2. The height of the multi-point silicone oil-based fluorescent magnetic fluid device 2 is greater than or equal to the height of the transformer under test 1. The shape of the multi-point silicone oil-based fluorescent magnetic fluid device 8 can be as follows: Figure 2 The center can be set to a cylindrical shape; or it can be set as follows: Figure 10 As shown, set it to a square; or as... Figure 11 As shown, it is set in a gourd shape; as Figure 3 , Figure 7 and Figure 9 As shown, in the unfolded state, the multi-point silicone oil-based fluorescent magnetic fluid device 8 comprises multiple silicone oil-based fluorescent magnetic fluid units 3 arranged in an array. Each silicone oil-based fluorescent magnetic fluid unit 3 includes a transparent encapsulation shell 301 made of insulating material, silicone oil 302 encapsulated within the transparent encapsulation shell 301, and fluorescent magnetic fluid 303 suspended within the silicone oil 302. The fluorescent magnetic fluid 303 forms an imaging point within the transparent encapsulation shell 301. The silicone oil-based fluorescent magnetic fluid unit 3 is square (e.g., ...). Figure 5 and Figure 6 (as shown), circular, oval (as shown) Figure 8As shown, each silicone oil-based fluorescent magnetic fluid unit 3 is rectangular or other polygonal, and its transparent encapsulation shell is made of transparent plastic, which can be transparent insulating acrylic material or other transparent plastics. Adjacent silicone oil-based fluorescent magnetic fluid units 3 are fixedly bonded together. The elliptical silicone oil-based fluorescent magnetic fluid units 3 form a structure similar to a gourd membrane, which can be manufactured using gourd membrane encapsulation processes. The size of the silicone oil-based fluorescent magnetic fluid unit 3 can be designed as needed, for example, 1×1×1 cm or 0.5×0.5×0.5 cm. The smaller the unit, the higher the imaging resolution. Each silicone oil-based fluorescent magnetic fluid unit 3 contains a small amount of fluorescent magnetic fluid material. The fluorescent magnetic fluid 302 is black, and the silicone oil 303 is colorless. The fluorescent magnetic fluid 302 and silicone oil 303 are insulating. The fluorescent material within the silicone oil-based fluorescent magnetic fluid unit 3 emits light in low light conditions, facilitating imaging by the camera device 7 in low light conditions and enabling all-weather monitoring of the magnetic fluid state within the silicone oil-based fluorescent magnetic fluid unit 3. The magnetic fluid in the silicone oil-based fluorescent magnetic fluid unit 3 will move and deform under the action of a magnetic field, thus the magnetic fluid will exhibit different states in the silicone oil-based fluorescent magnetic fluid unit 3 (such as...). Figure 4 (As shown).

[0043] In Example 1, the transformer under test 1 exhibits different magnetic field characteristics under different operating conditions, and the multi-point silicone oil-based fluorescent magnetohydrodynamic device 2 exhibits different overall states under different operating conditions of the transformer under test 1.

[0044] The imaging system provided in Example 1 is as follows Figure 1 and Figure 2As shown, the imaging system includes a track 4 surrounding the multi-point silicone oil-based fluorescent magnetofluid device 2, a horizontal walking mechanism 5 mounted on the track 4, a vertical lifting frame 6 mounted on the horizontal walking mechanism 5, and a camera device 7 mounted on the vertical lifting frame 6. The imaging system also includes a base 9 surrounding the transformer 1 under test. Both the track 4 and the multi-point silicone oil-based fluorescent magnetofluid device 2 are mounted on the base 9. The horizontal walking mechanism 5 is an electrically controlled automatic walking trolley, and the vertical lifting frame 6 is an electric lifting frame. The monitoring system 8 is connected to the control motors of the horizontal walking mechanism 5 and the vertical lifting frame 6, respectively. The movement of the control motors of the horizontal walking mechanism 5 and the vertical lifting frame 6 is controlled by commands sent by the monitoring system 8. Both the horizontal walking mechanism 5 and the vertical lifting frame 6 can be implemented using existing conventional technologies, mainly used to control the camera device 7 to move up and down and to move horizontally around the track 4, enabling imaging of each silicone oil-based fluorescent magnetofluid unit 3 of the multi-point silicone oil-based fluorescent magnetofluid device 2. The camera device 7 uses a high-definition magnifiable camera, which faces the multi-point silicone oil-based fluorescent magnetic fluid device 2, and is used to capture images of the fluorescent magnetic fluid 303 in the multi-point silicone oil-based fluorescent magnetic fluid device 2. The lifting height of the vertical lifting frame 6 is greater than or equal to the height of the multi-point silicone oil-based fluorescent magnetic fluid device 2.

[0045] In Example 1, the monitoring system 8 is communicatively connected to the control systems of the horizontal walking mechanism 5, the vertical lifting frame 6, and the camera device 7, respectively. The monitoring system 8 controls the vertical lifting frame 6 to move the camera device 7 up and down to capture images of the silicone oil-based fluorescent magnetic fluid unit 3 at different heights, determining the z-coordinate of the camera device 7 during image capture. The monitoring system 8 controls the horizontal walking mechanism 5 to move the camera device 7 along the track 4 to image the silicone oil-based fluorescent magnetic fluid unit 3 at different horizontal positions and determine the x and y coordinates of the camera device 7 during image capture. Figure 12 As shown, the monitoring system 8 receives images of the silicone oil-based fluorescent magnetic fluid unit 3 at different locations uploaded by the camera device 7 in the imaging system in real time, and labels the images as p(x, y, z), where x, y, and z are the three-dimensional spatial coordinates corresponding to the images. Simultaneously, it processes the images of the silicone oil-based fluorescent magnetic fluid unit 3 at different locations in real time and determines the specific morphology of the magnetic fluid within each silicone oil-based fluorescent magnetic fluid unit 3. The processed images of the silicone oil-based fluorescent magnetic fluid unit 3 at different locations are then stitched together to form a three-dimensional image of the multi-point silicone oil-based fluorescent magnetic fluid device 2, which is then used to determine the operating status of the transformer 1 under test. The method used by the monitoring system 8 to process the images p(x, y, z) of the silicone oil-based fluorescent magnetic fluid unit 3 at different locations in real time includes grayscale conversion, Gaussian filtering, histogram equalization, and binarization. The core algorithm used by the monitoring system 8 to determine the specific morphology of the magnetic fluid within the silicone oil-based fluorescent magnetic fluid unit 3 at different locations includes contour extraction and centroid solving algorithms.

[0046] like Figure 12 As shown, the monitoring system 8 monitors the transformer under test 1 in real time and generates a three-dimensional image of the multi-point silicone oil-based fluorescent magnetic fluid device 2. The real-time generated three-dimensional image of the multi-point silicone oil-based fluorescent magnetic fluid device 2 is compared with the three-dimensional images of the multi-point silicone oil-based fluorescent magnetic fluid device 2 under different states pre-stored in the monitoring system 8, and the similarity between the images is calculated. When the similarity reaches a threshold, it is determined that the operating state of the transformer under test 1 is consistent with the transformer operating state corresponding to the three-dimensional image of the multi-point silicone oil-based fluorescent magnetic fluid device 2 pre-stored in the database. If the similarity does not reach the threshold, the transformer under test 1 is disassembled and the cause and location of the fault are found. The three-dimensional image of the multi-point silicone oil-based fluorescent magnetic fluid device 2 under the current state of the transformer under test 1 is pre-stored in the database. The similarity calculation algorithm between the two images includes structural similarity calculation and normalized evaluation based on coordinate distance.

[0047] Example 2 provides a non-contact visual dry-type transformer operation status monitoring method, based on the non-contact visual dry-type transformer operation status monitoring device described in Example 1, specifically including the following steps:

[0048] S1. Collect three-dimensional images of the multi-point silicone oil-based fluorescent magnetohydrodynamic device 2 under normal conditions and after operation under typical artificially set faults of the transformer under test 1, and store them in the database of the monitoring system 8 as reference images;

[0049] S2. The monitoring system 8 monitors the images of the silicone oil-based fluorescent magnetic fluid unit 3 at different locations of the transformer under test 1 in real time, and calculates and determines the specific morphology of the magnetic fluid in each silicone oil-based fluorescent magnetic fluid unit 3 after processing the images. The processed images of the silicone oil-based fluorescent magnetic fluid unit 3 at different locations are then stitched together to form a three-dimensional image of the multi-point silicone oil-based fluorescent magnetic fluid device 2.

[0050] In this embodiment, the camera device 7 simply captures images of the silicone oil-based fluorescent magnetic fluid unit 3 on the multi-point silicone oil-based fluorescent magnetic fluid device 2, i.e., only the images in the xy plane are processed, i.e., in a two-dimensional state (e.g., ...). Figure 13 As shown), taking the cubic silicone oil-based fluorescent magnetic fluid unit 3 as an example, the size of the silicone oil-based fluorescent magnetic fluid unit 3 is 10×10×10 mm. Its specific calculation process is as follows (the image processing and data analysis algorithms involved can all be implemented by Python programming. The specific algorithm is not limited, and other algorithms and software can also be used):

[0051] S201, xy plane image calibration

[0052] Using the known physical dimensions (10×10 mm) of the xy plane of the cubic magnetohydrodynamic unit 3, the correspondence between "image pixels → xy plane physical coordinates" is established as follows:

[0053] (1) Determination of the position of the cubic magnetic fluid unit 3: The magnetic fluid device composed of cubic magnetic fluid units 3 is arranged around the transformer under test 1, and the origin of the coordinate system of the magnetic fluid device is determined. After the origin of the coordinate system is determined, the coordinates of the xy plane of each cubic magnetic fluid unit 3 can be determined, such as Figure 13 As shown, the coordinates of the four vertices of the 3xy plane of the first cubic magnetic fluid unit are O (0,0) (bottom left), P (10,0) (bottom right), Q (10,10) (top right), and R (0,10) (top left).

[0054] (2) Taking reference images: The vertical lifting frame 6 (at different heights) and the horizontal walking mechanism 5 (at different horizontal positions on the track 4) drive the camera device 7 to the coordinate area corresponding to each magnetic fluid unit 3 on the magnetic fluid device 2. The camera device 7 takes a reference image of the xy plane of each cubic magnetic fluid unit 3. The central axis of the imaging of the camera device 7 is aligned with the center of the xy plane of each magnetic fluid unit 2. For example, when imaging the first cubic magnetic fluid unit 3, the central axis is aligned with the coordinate C (5, 5) to ensure that the four vertices of the xy plane of the cubic magnetic fluid unit 2 can be clearly identified in the image.

[0055] (3) Pixel-physical coordinate mapping establishment: In the reference image, the pixel coordinates (7 pixels in size of the camera device) of the four vertices are extracted by image edge detection (such as the Canny algorithm, which is the existing technology): O (px_O, py_O), P (px_P, py_P), Q (px_Q, py_Q), R (px_R, py_R);

[0056] Calculate the pixel dimensions of the xy plane in the image: pixel length in the x-axis direction Lx = |px_P - px_O|, pixel length in the y-axis direction Ly = |py_R - py_O|;

[0057] Establish mapping relationship: Since the physical size of the cubic magnetohydrodynamic unit 3 xy plane is 10×10 mm, the conversion coefficient kx in the x-axis direction "pixel to physical coordinate" is kx = 10 / Lx (unit: mm / pixel), and the conversion coefficient ky in the y-axis direction is ky = 10 / Ly (unit: mm / pixel).

[0058] Determine the origin mapping: The pixel coordinates (px_O, py_O) of the lower left vertex O of the xy plane of the cubic magnetohydrodynamic unit 3 in the image correspond to the physical coordinates (0,0). The formula for the physical coordinates (x,y) of any pixel (px, py) on the xy plane in the image is:

[0059] x = (px - px_O) × kx

[0060] y = (py - py_O) × ky

[0061] S202, Magnetohydrodynamic xy-plane positioning algorithm

[0062] (1) Take pictures of the xy plane of each cubic magnetic fluid unit 3: Under the action of the magnetic field, use the monitoring system 8 to control the camera device 7 to take pictures of the magnetic fluid on the xy plane of each cubic magnetic fluid unit 3 (the distribution pattern of the magnetic fluid in the xy plane of the magnetic fluid unit 3 should be clearly shown).

[0063] (2) Image preprocessing: convert the captured image to grayscale (convert the RGB image to a single-channel grayscale image, existing technology), Gaussian filtering (eliminate image noise, existing technology), and histogram equalization (enhance the contrast between the magnetohydrodynamic fluid and the background, existing technology).

[0064] (3) Magnetofluid region segmentation: Otsu adaptive threshold segmentation (existing technology) is used to binarize the preprocessed image (magnetofluid region is white and background is black, existing technology) to complete the determination of magnetofluid morphology;

[0065] (4) Calculation of xy-plane coordinates of magnetohydrodynamic fluid:

[0066] The contours of the magnetic fluid region are extracted using OpenCV's findContours function (existing technology), and the pixel range of the magnetic fluid in the image is determined.

[0067] The centroid pixel coordinates (px_c, py_c) of the magnetohydrodynamic region are calculated using the image Moments moment function (existing technology);

[0068] Substituting the mapping relationship established in S3, calculate the physical coordinates (x, y) of the cubic magnetohydrodynamic unit 3 in the xy plane: x=(px_c - px_O)×kx, y=(py_c - py_O)×ky;

[0069] (5) Output of results: Mark the position of the centroid of the magnetic fluid in the cubic magnetic fluid unit 3 (such as red dots) and the corresponding physical coordinates (x, y) on the captured xy plane image of the cubic magnetic fluid unit 3, and save the coordinate data of the centroid of the magnetic fluid in the cubic magnetic fluid unit 3 to complete the extraction of the morphological data of the magnetic fluid in the cubic magnetic fluid unit 3.

[0070] (6) According to the coordinate position order of the images captured by the camera device 7, the binarized images of each magnetic fluid unit 3 obtained in step S202 (3) are directly stitched together to form a multi-point fluorescent silicone oil-based magnetic fluid device 2 overall xy plane binarized image, and the centroid position of the magnetic fluid unit forms a database to complete the data collection.

[0071] S3. Calculate the similarity between the real-time generated multi-point fluorescent silicone oil-based magnetohydrodynamic device 2 three-dimensional image and the reference images of the transformer under test 1 in different states pre-stored in the monitoring system database; the calculation data is based on the image and centroid coordinate data obtained in step S2, and the specific process is as follows:

[0072] S301. Data Definition and Preprocessing: Define the computational objects, and let the two sets of overall binary images and centroid coordinate data of the magnetohydrodynamic device 2 to be compared be data set A and data set B:

[0073] Image data: A_img (image of data group A) and B_img (image of data group B) are both grayscale images (with consistent size and pixels) of the xy plane of the same magnetohydrodynamic device 2 (but data of the transformer under test 1 in two operating states).

[0074] Centroid data: A_centroid=(x_A, y_A) (physical coordinates of the magnetohydrodynamic centroid of data group A), B_centroid=(x_B, y_B) (physical coordinates of the magnetohydrodynamic centroid of data group B).

[0075] Note: A_img and B_img have the same size and resolution (if they are different, they are scaled to the same size by interpolation, as is the case with existing technology); the centroid coordinates are based on the same coordinate system (i.e., the coordinate system of the xy plane image described in step S2).

[0076] S302. Structural Similarity (SSIM, S 11 The calculation involves evaluating the global similarity between images A_img and B_img using three dimensions: brightness, contrast, and structure. The steps are as follows:

[0077] (1) Sliding window setting: an 11×11 pixel window (step size 1 pixel) is used, and a Gaussian filter with a standard deviation of 1.5 is applied within the window (existing technology) to simulate the local perception characteristics of human vision;

[0078] (2) Local similarity calculation: For each window, calculate separately:

[0079] Brightness similarity: ,

[0080] in , C1 represents the average grayscale value of windows A_img and B_img, where C1 = (0.01 × 255). 2 ;

[0081] Contrast similarity: ,

[0082] in , Let C2 be the grayscale standard deviation of windows A_img and B_img, where C2 = (0.03 × 255). 2 ;

[0083] Structural similarity: ,

[0084] in Let C3 be the grayscale covariance between the two windows, and C2 = C2 / 2.

[0085] Local SSIM: ;

[0086] (3) Global SSIM synthesis: Take the arithmetic mean of the local SSIMs of all windows to obtain S 11 ,Right now M represents the total number of windows (e.g., a 1280×720 image corresponds to M=1270×710).

[0087] S303. Feature Matching Rate (S 12 The calculation is performed; the local detail matching degree of the magnetohydrodynamic region is evaluated using ORB features (existing technology), and the steps are as follows:

[0088] (1) Magnetofluid region segmentation: Divide the binarized images A_img and B_img obtained in steps (2) and (3) of S202 into regions A_region and B_region (target is 1, i.e., the magnetic fluid in magnetic fluid unit 3, and background is 0).

[0089] (2) ORB feature extraction: Construct 4-layer Gaussian pyramids for A_region and B_region respectively (existing technology), detect corner points using the Oriented FAST algorithm (existing technology) (maximum number of feature points 1000), and calculate the direction of the gray centroid as the main direction of the feature points;

[0090] 256 pairs of random points are generated within a 32×32 pixel neighborhood. After rotating the point pairs according to the main direction, a 256-bit binary descriptor is generated by comparing grayscale values.

[0091] (3) Feature matching and filtering: The FLANN matcher (existing technology) is used for initial matching to obtain matching pairs; the RANSAC algorithm (existing technology) is used to remove mismatches and retain inliers (correctly matched feature point pairs).

[0092] (4) Match rate calculation:

[0093] Where N inlier Let N be the number of interior points. A N B These represent the total number of feature points in region A and region B, respectively.

[0094] S304. Comprehensive similarity (S1) calculation, using weighted fusion S 11 With S 12 ,

[0095] formula: ,

[0096] The weight w1 ranges from [0.5, 0.7] (default is 0.6, focusing on the global structure), and the weight S1 ranges from [0, 1]. The closer the value is to 1, the higher the similarity between A_img and B_img.

[0097] S305. Centroid data similarity calculation (S2): Based on the normalization of coordinate distance, the centroid data is a two-dimensional coordinate system. The similarity is calculated by inverse normalization of Euclidean distance (existing technology). The smaller the distance, the higher the similarity.

[0098] Calculate the Euclidean distance (d) between the centroids.

[0099] Calculate the distance between A_centroid and B_centroid:

[0100] Calculate the distance normalization (S2).

[0101] Mapping the distance d to the interval [0, 1] as the similarity, we need to define the "maximum possible distance" d. max (That is, the maximum physical distance between two points in the xy plane of the cubic magnetohydrodynamic unit, which is approximately 14.14 mm in a 10×10 mm image):

[0102]

[0103] When d=0 (the centroids are completely coincident), S2=1; when d=d_max (the centroids are furthest apart), S2=0.

[0104] S306. Overall Similarity Calculation (S): Multi-dimensional weighted fusion, combining image similarity (S1) and centroid similarity (S2), using dynamic weight fusion. The weights are adaptively adjusted according to the scene's sensitivity to "image structure" and "centroid position". The steps are as follows:

[0105] S3061. Weighting Determination Principle: If the application scenario focuses more on the morphological consistency of the magnetic fluid, then the image weight w2 is larger (e.g., w2=0.6); if the focus is more on the positional consistency of the magnetic fluid, then the centroid weight 1-w2 is larger (e.g., w2=0.4); in the default scenario, w2=0.5 (equal weight).

[0106] S3062. Overall Similarity Formula:

[0107] ,

[0108] The value range is [0, 1]. The closer S is to 1, the more similar the two sets of data are. The overall similarity S of the multi-point silicone oil-based fluorescent magnetic fluid device 2 can be taken as the weighted average.

[0109] S4. Based on the similarity S calculated in step S3, determine the operating status of the transformer under test; a threshold T can be set (e.g., T=90%). When S is greater than or equal to T, the two sets of data are determined to be "highly similar". When the similarity reaches the threshold, the operating status of the transformer under test 1 is consistent with the operating status of the transformer corresponding to the pre-stored image in the database; if the similarity does not reach the threshold (set according to the pre-experiment test results and experience, such as 60%), disassemble the transformer under test 1 and find the cause and location of the fault.

[0110] S5. After the fault is determined, the corresponding three-dimensional image of the multi-point fluorescent silicone oil-based magnetohydrodynamic device 2 is stored in the database of the monitoring system 8 to improve the database of the monitoring system 8; and the real-time visualization monitoring, fault detection and location of the operating status of the transformer under test 1 are completed.

[0111] The above description is merely one embodiment of the present invention, and while it is detailed and specific, it should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the appended claims.

Claims

1. A non-contact, visual monitoring device for the operating status of a dry-type transformer, characterized in that: The monitoring device includes an imaging system, a monitoring system (8), and a multi-point silicone oil-based fluorescent magnetic fluid device (2) arranged around the transformer under test (1). The multi-point silicone oil-based fluorescent magnetic fluid device (2) is a hollow structure composed of multiple insulating silicone oil-based fluorescent magnetic fluid units (3). The transformer under test (1) is located in the hollow area of ​​the multi-point silicone oil-based fluorescent magnetic fluid device (2). The height of the multi-point silicone oil-based fluorescent magnetic fluid device (2) is greater than or equal to the height of the transformer under test (1). The silicone oil-based fluorescent magnetic fluid unit (3) includes a transparent encapsulation shell (301) made of insulating material, silicone oil (302) encapsulated in the transparent encapsulation shell (301), and fluorescent magnetic fluid (303) suspended in the silicone oil (302). The fluorescent magnetic fluid (303) forms an imaging point in the transparent encapsulation shell (301). The imaging system includes a track (4) surrounding the multi-point silicone oil-based fluorescent magnetic fluid device (2), a horizontal walking mechanism (5) mounted on the track (4), a vertical lifting frame (6) mounted on the horizontal walking mechanism (5), and a camera device (7) mounted on the vertical lifting frame (6). The camera of the camera device (7) faces the multi-point silicone oil-based fluorescent magnetic fluid device (2) and is used to capture images of the fluorescent magnetic fluid (303) in the multi-point silicone oil-based fluorescent magnetic fluid device (2). The lifting height of the vertical lifting frame (6) is greater than or equal to the height of the multi-point silicone oil-based fluorescent magnetic fluid device (2). The monitoring system (8) is connected to the control systems of the horizontal walking mechanism (5), the vertical lifting frame (6), and the camera device (7) respectively. The monitoring system (8) receives images of silicone oil-based fluorescent magnetic fluid units (3) at different positions uploaded by the camera device (7) in the imaging system in real time, marks the images as p(x, y, z), where x, y, and z are the three-dimensional spatial coordinates corresponding to the images. It also processes the images of silicone oil-based fluorescent magnetic fluid units (3) at different positions in real time, calculates the specific morphology of the magnetic fluid in each silicone oil-based fluorescent magnetic fluid unit (3), and splices the processed images of silicone oil-based fluorescent magnetic fluid units at different positions to form a three-dimensional image of the multi-point silicone oil-based fluorescent magnetic fluid device (2). Then, it judges the operating status of the transformer (1) under test.

2. The non-contact visual dry-type transformer operation status monitoring device according to claim 1, characterized in that: The monitoring system monitors the transformer under test (1) in real time and generates a three-dimensional image of the multi-point silicone oil-based fluorescent magnetic fluid device (2). The three-dimensional image of the multi-point silicone oil-based fluorescent magnetic fluid device generated in real time is compared with the three-dimensional images of the multi-point silicone oil-based fluorescent magnetic fluid device under different states of the transformer under test (1) stored in the monitoring system to calculate the similarity between them. When the similarity reaches the threshold, it is determined that the operating state of the transformer under test (1) is consistent with the operating state of the transformer corresponding to the three-dimensional image of the multi-point silicone oil-based fluorescent magnetic fluid device stored in the database. If the similarity does not reach the threshold, the transformer under test (1) is disassembled and the cause and location of the fault are found. The three-dimensional image of the multi-point silicone oil-based fluorescent magnetic fluid device under the current state of the transformer under test (1) is stored in the monitoring system (8). The similarity calculation algorithm between the two images includes structural similarity calculation and normalization evaluation based on coordinate distance.

3. A non-contact visual dry-type transformer operation status monitoring device according to claim 1 or 2, characterized in that: The monitoring system (8) controls the vertical lifting frame (6) to drive the camera device (7) to move up and down to capture images of the silicone oil-based fluorescent magnetic fluid unit (3) at different heights, and determines the z coordinate of the camera device (7) when taking pictures; the monitoring system (8) controls the horizontal walking mechanism (5) to drive the camera device (7) to move along the track (4) to image the silicone oil-based fluorescent magnetic fluid unit (3) at different horizontal positions, and determines the x and y coordinates of the camera device (7) when taking pictures; the core algorithm of the monitoring system (8) to process the images p(x, y, z) of the silicone oil-based fluorescent magnetic fluid unit (3) at different positions in real time and calculate the specific morphology of the magnetic fluid inside the silicone oil-based fluorescent magnetic fluid unit (3) includes contour extraction and centroid solving algorithms.

4. A non-contact visual dry-type transformer operation status monitoring device according to claim 1 or 2, characterized in that: The imaging system also includes a base (9) surrounding the transformer (1) under test, and the track (4) and the multi-point silicone oil-based fluorescent magnetohydrodynamic device (2) are both mounted on the base (9).

5. A non-contact visual dry-type transformer operation status monitoring device according to claim 1 or 2, characterized in that: The multi-point silicone oil-based fluorescent magnetic fluid device (8) is hollow, and its shape matches the specifications and usage scenario of the transformer under test (1), and surrounds the transformer under test (1) in the middle. In the unfolded state, the multiple silicone oil-based fluorescent magnetic fluid units (3) that make up the multi-point silicone oil-based fluorescent magnetic fluid device (8) are arranged in an array.

6. A non-contact visual dry-type transformer operation status monitoring device according to claim 1 or 2, characterized in that: The silicone oil-based fluorescent magnetic fluid unit (3) is square, round, elliptical, rectangular or other polygonal. The transparent encapsulation shell of each silicone oil-based fluorescent magnetic fluid unit (3) is made of transparent plastic. Adjacent silicone oil-based fluorescent magnetic fluid units (3) are fixedly bonded together. The fluorescent magnetic fluid (302) is black and the silicone oil (303) is colorless. The fluorescent magnetic fluid (302) and the silicone oil (303) are insulating.

7. A non-contact visual dry-type transformer operation status monitoring device according to claim 1 or 2, characterized in that: The horizontal walking mechanism (5) is an electrically controlled automatic walking trolley, and the vertical lifting frame (6) is an electric lifting frame; the monitoring system (8) is connected to the control motor signals of the horizontal walking mechanism (5) and the vertical lifting frame (6) respectively, and the movement of the control motors of the horizontal walking mechanism (5) and the vertical lifting frame (6) is controlled by the commands sent by the monitoring system (8).

8. A non-contact, visual monitoring method for the operating status of a dry-type transformer, characterized in that: The method is based on the non-contact visual dry-type transformer operation status monitoring device according to any one of claims 1 to 7, and specifically includes the following steps: S1. Collect three-dimensional images of the multi-point silicone oil-based fluorescent magnetohydrodynamic device of the transformer under test under normal conditions and after operation under typical artificial faults, and store them in the database of the monitoring system as reference images. S2. The monitoring system monitors the images of silicone oil-based fluorescent magnetic fluid units at different locations of the transformer under test in real time. After processing the images, the specific morphology of the magnetic fluid in each silicone oil-based fluorescent magnetic fluid unit is calculated and determined. The processed images of silicone oil-based fluorescent magnetic fluid units at different locations are then stitched together to form a three-dimensional image of a multi-point silicone oil-based fluorescent magnetic fluid device. S3. Calculate the similarity between the real-time generated three-dimensional image of the multi-point fluorescent silicone oil-based magnetohydrodynamic device and the reference images of the transformer under test in different states stored in the monitoring system database; S4. Determine the operating status of the transformer under test based on the similarity calculated in step S3. When the similarity reaches the threshold, the current operating status of the transformer under test is consistent with the operating status of the transformer corresponding to the pre-stored reference image in the database. If the similarity does not reach the threshold, disassemble the transformer under test and find the cause and location of the fault, and pre-store the three-dimensional image of the multi-point silicone oil-based fluorescent magnetohydrodynamic device under the current state of the transformer under test in the database. Complete the real-time visualization monitoring, fault detection and location of the operating status of the transformer under test.

9. A non-contact visual dry-type transformer operation status monitoring device according to claim 8, characterized in that: Typical faults manually set in step S1 include winding short circuit, winding deformation, magnetic leakage, abnormal vibration, and overheating. After the fault is determined in step S4, the three-dimensional image of the multi-point fluorescent silicone oil-based magnetohydrodynamic device corresponding to the fault is stored in the database of the monitoring system to improve the database of the monitoring system.

10. A non-contact visual dry-type transformer operation status monitoring device according to claim 8 or 9, characterized in that: In step S2, the monitoring system receives images of silicone oil-based fluorescent magnetic fluid units at different locations uploaded by the camera device in the imaging system in real time. These images are labeled p(x, y, z), where x, y, and z are the corresponding three-dimensional spatial coordinates. Image processing methods are used to process the silicone oil-based fluorescent magnetic fluid unit images at different locations in real time. Then, contour extraction and centroid solving algorithms are used to calculate and determine the specific morphology of the magnetic fluid within each silicone oil-based fluorescent magnetic fluid unit. The processed images of multiple silicone oil-based fluorescent magnetic fluid units are then stitched together to form a three-dimensional image of the multi-point silicone oil-based fluorescent magnetic fluid device. The image processing methods include grayscale conversion, Gaussian filtering, histogram equalization, and binarization. The algorithm for calculating the similarity between the real-time generated 3D image of the multi-point fluorescent silicone oil-based magnetohydrodynamic device and the reference image in step S3 includes structural similarity calculation and normalized evaluation based on coordinate distance.

Images