Method for detecting theft-proof television in high-voltage metering state

Abstract

This invention relates to the field of machine vision inspection technology and discloses a visual detection method for preventing electricity theft in high-voltage metering conditions. It aims to solve the problems of existing anti-theft monitoring methods being susceptible to misjudgments due to mechanical vibration and light interference in complex industrial environments, and the difficulty in effectively capturing microscopic and concealed traces of electricity theft. This invention acquires reference images and establishes an archive using a pre-set fixed imaging unit and positioning markers; performs quality screening and spatial position correction on the current image to output a standardized image; extracts specific physical state features of the sealing area, meter area, and terminal pressure plate area, and performs item-by-item quantitative comparison with the reference archive; performs joint judgment based on feature difference results, and outputs suspected electricity theft results and generates evidence records when progressive correlation conditions are met. This method filters out environmental displacement and light and shadow interference, achieves non-contact quantitative detection of minute abnormal tampering traces, reduces the false alarm rate, and provides objective comparison evidence.

Description

Technical Field

[0001] This invention relates to the field of machine vision inspection technology, and in particular to a visual inspection method for preventing electricity theft in high-voltage metering conditions. Background Technology

[0002] High-voltage metering devices are core equipment for power supply companies to measure electricity consumption from large industrial and commercial users. The safety and legality of their operation directly affect the economic benefits of the power grid company. Currently, anti-theft inspections of high-voltage metering cabinets mainly rely on two conventional methods: one is to conduct on-site inspections by manually opening the cabinets periodically; the other is to passively screen for suspected electricity theft clues by analyzing user load data (such as abnormal fluctuations in current and voltage) in the main station system. In recent years, some distribution rooms have also begun to introduce conventional video surveillance equipment, mainly for basic video recording of the external environment or macroscopic operation process of the metering cabinets.

[0003] However, with the increasing sophistication and concealment of electricity theft methods, violators often resort to extremely minor physical damage, such as fine-tuning meter fixing screws, forging anti-counterfeiting seals, or partially short-circuiting the secondary circuit of terminal pressure plates, to steal electricity and then meticulously restore the stolen devices afterward. Traditional manual inspections, limited by safety regulations for high-voltage live-line work and the frequency of inspections, cannot achieve real-time online monitoring, and the aforementioned microscopic signs of damage are extremely difficult for inspectors to detect with the naked eye. Passive analysis relying on load data suffers from significant lag and cannot provide direct physical visual evidence to prove the theft. Attempts to automate monitoring using conventional image comparison techniques are hampered by the enclosed and dimly lit interior of high-voltage cabinets, coupled with the unavoidable mechanical vibrations caused by the operation of large industrial equipment. This easily leads to spatial misalignment and light interference in the monitoring images, resulting in widespread false alarms or missed alarms. Therefore, how to overcome the physical interference of complex industrial environments and objectively, accurately, and promptly capture extremely small and hidden physical traces of electricity theft without power interruption or contact has become a substantial technical problem that urgently needs to be solved in the field of high-voltage metering and anti-theft. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, in order to solve the problem that existing anti-theft monitoring methods are easily misjudged due to mechanical vibration and light interference in complex industrial environments, and are difficult to effectively capture microscopic and hidden traces of electricity theft, this invention provides a visual detection method for anti-theft of electricity under high voltage metering conditions.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: This invention provides a visual detection method for preventing electricity theft under high-voltage metering conditions, which includes the following steps: S1. Obtain a reference image of the high-voltage metering device when it is in a legal and normal state and establish a reference file; wherein, the reference image is acquired by a fixed shooting unit preset in the high-voltage metering cabinet with the cooperation of a fixed supplementary lighting unit, and the high-voltage metering cabinet is preset with positioning marks; S2. Acquire the current image captured by the fixed shooting unit at a preset position, and use the positioning mark and the fixed structure inside the high-voltage metering cabinet to perform position correction and quality screening on the current image to obtain a standardized image; S3. Extract the sealing area, meter area and terminal pressure plate area from the standardized image, and extract the state features of the corresponding areas respectively; S4. Compare the state characteristics of the sealing area, the meter area and the terminal pressure plate area with the corresponding area state characteristics in the reference file item by item to obtain the sealing area difference results, meter area difference results and terminal pressure plate area difference results. S5. Determine sealing operation traces based on the difference results of the sealing area, determine meter operation traces based on the difference results of the meter area, determine terminal rewiring traces based on the difference results of the terminal pressure plate area, and output suspected electricity theft results when the sealing operation traces, the meter operation traces, and the terminal rewiring traces meet preset association conditions.

[0007] As a preferred embodiment of the anti-theft visual detection method for high-voltage metering status described in this invention, the method involves: acquiring a reference image of the high-voltage metering device in a legal and normal state and establishing a reference file; wherein the reference image is acquired by a fixed imaging unit preset inside the high-voltage metering cabinet with the assistance of a fixed supplementary lighting unit, and the high-voltage metering cabinet has preset positioning marks. The specific steps are as follows: A first fixed shooting unit is preset inside the high-voltage metering cabinet facing the sealing area; a second fixed shooting unit is preset inside the high-voltage metering cabinet facing the meter area; and a third fixed shooting unit is preset inside the high-voltage metering cabinet facing the terminal pressure plate area. Fixed supplementary lighting units are preset in the shooting directions of the first, second, and third fixed shooting units, respectively. A first positioning mark is preset near the sealing hole; a second positioning mark is preset at the fixed corner of the meter; and a third positioning mark is preset at both ends of the terminal block. When the high-voltage metering device is in a legal and normal state, reference images of the sealing area, meter area, and terminal pressure plate area, respectively captured by the first, second, and third fixed shooting units, are acquired. The current high-voltage metering cabinet number, user number, shooting time, fixed shooting unit number, and fixed supplementary lighting unit working status are acquired, and the above information is stored in the reference file corresponding to the reference images of the sealing area, meter area, and terminal pressure plate area. The sealing area includes the sealing wire perforation path, sealing wire bending shape, sealing wire tightness, outer contour of the sealing component, pressure position of the sealing component, and edge marks of the sealing hole; the meter area includes the meter housing position, relative position of the meter housing and mounting base, display window boundary, nameplate position, fixing screw position, fixing screw slot orientation, and joint condition of the meter panel and mounting base; the terminal pressure plate area includes the terminal sequence, number of terminal wires, wire color, wire entry direction into the pressure plate, wire length protruding from the pressure plate, first bend position of the wire, natural bending shape of the wire, edge fit of the pressure plate, screw slot condition of the pressure plate, and end condition of the wire harness tie.

[0008] As a preferred embodiment of the visual detection method for preventing electricity theft under high-voltage metering status described in this invention, the method involves: acquiring a current image captured by the fixed imaging unit at a preset position; using the positioning marker and the fixed structure within the high-voltage metering cabinet to perform position correction and quality screening on the current image to obtain a standardized image; the specific steps are as follows: Acquire images of the current sealing area, current meter area, and current terminal pressure plate area, captured by the first fixed shooting unit, the second fixed shooting unit, and the third fixed shooting unit at preset positions; perform quality screening on the images of the current sealing area, the current meter area, and the current terminal pressure plate area; if any of the following conditions exist in the images of the current sealing area, the current meter area, and the current terminal pressure plate area: out of focus, ghosting, reflection obstruction, shadow obstruction, local overexposure, local underexposure, lens surface contamination, cabinet door obstruction, and uneven supplementary lighting, control the corresponding fixed supplementary lighting unit to re-supplement the light and trigger re-acquisition of the image. The initial reference position in the current image is determined using the first positioning mark, the second positioning mark, and the third positioning mark; then, the left-right offset, up-down offset, rotation offset, and scale offset of the current image relative to the reference image are determined using the cabinet door frame, the edge of the observation window, the edge of the cabinet mounting back panel, the features of the base mounting rail, and the end face of the terminal block; the current image is reset based on the left-right offset, the up-down offset, the rotation offset, and the scale offset, so that the sealing area, meter area, and terminal pressure plate area in the current image are aligned with the corresponding areas in the reference image.

[0009] As a preferred embodiment of the visual detection method for preventing electricity theft in high-voltage metering status according to the present invention, the following steps are taken: Seal area, meter area, and terminal pressure plate area are extracted from the standardized image, and the status features of the corresponding areas are extracted respectively. Extract the sealing wire perforation path, sealing wire bending shape, sealing wire tightness, outer contour of the sealing component, pressure position of the sealing component, and edge traces of the sealing hole in the sealing area; extract the meter housing position, relative position of the meter housing and mounting base, display window boundary, nameplate position, fixing screw position, fixing screw slot orientation, and joint state of the meter panel and mounting base in the meter area; extract the terminal sequence, number of terminal wires, wire color, wire entry direction into the pressure plate, wire length exposed from the pressure plate, first bend position of the wire, natural bending shape of the wire, edge contact state of the pressure plate, screw slot state of the pressure plate, and end state of the wire harness tie in the terminal pressure plate area.

[0010] As a preferred embodiment of the visual detection method for preventing electricity theft in high-voltage metering status according to the present invention, the method involves comparing the state characteristics of the sealing area, the meter area, and the terminal pressure plate area with the corresponding area state characteristics in the reference file item by item to obtain the difference results of the sealing area, the meter area, and the terminal pressure plate area. The specific steps are as follows: The extracted sealing area features, including the sealing wire perforation path, sealing wire bending shape, sealing wire tightness, outer contour of the sealing component, pressure position of the sealing component, and sealing hole edge marks, are compared item by item with the corresponding area state features in the reference file to obtain the sealing area difference results. Similarly, the extracted meter area features, including the meter housing position, relative position between the meter housing and the mounting base, display window boundary, nameplate position, fixing screw position, fixing screw slot orientation, and the joint state between the meter panel and the mounting base, are compared item by item with the corresponding area state features in the reference file to obtain the meter area difference results. Finally, the extracted terminal plate area features, including the terminal sequence, number of terminal wires, wire color, wire entry direction into the plate, wire length protruding from the plate, first bend position of the wire, natural bending shape of the wire, plate edge fit state, plate screw slot state, and wire harness tie end state, are compared item by item with the corresponding area state features in the reference file to obtain the terminal plate area difference results.

[0011] As a preferred embodiment of the visual detection method for preventing electricity theft under high-voltage metering conditions described in this invention, the method involves: determining sealing operation traces based on the difference results of the sealing area, determining meter operation traces based on the difference results of the meter area, determining terminal rewiring traces based on the difference results of the terminal pressure plate area, and outputting a suspected electricity theft result when the sealing operation traces, the meter operation traces, and the terminal rewiring traces meet preset correlation conditions. The specific steps are as follows: In the results of the differences in the sealing area, if two or more of the following characteristics change: the perforation path of the sealing line, the bending shape of the sealing line, the pressure position of the sealing component, and the edge marks of the sealing hole, then the sealing operation traces are determined to exist. In the results of the differences in the meter area, if two or more of the following characteristics change: the relative position of the meter housing and the mounting base, the orientation of the fixing screw slot, the state of the joint between the meter panel and the mounting base, and the boundary of the display window, then the meter operation traces are determined to exist. In the results of the differences in the terminal pressure plate area, if two or more of the following characteristics change: the number of terminal wires, the color of the wires, the length of the wires exposed on the pressure plate, the first bend position of the wires, the natural bending shape of the wires, the edge contact state of the pressure plate, and the state of the pressure plate screw slot, then the terminal rewiring traces are determined to exist. After determining the presence of traces of sealing and metering operations, the high-voltage metering cabinet is marked as a key inspection area. If, under this key inspection area, traces of terminal rewiring are further determined, a suspected electricity theft result is output. When the suspected electricity theft result is output, an evidence record is generated, including a baseline image, current image, magnified view, differences in the sealing area, differences in the metering area, differences in the terminal pressure plate area, the high-voltage metering cabinet number, the user number, and the shooting time. On-site inspection feedback instructions are received, and the on-site inspection results are written into the historical record according to the feedback instructions. If the on-site inspection result indicates legitimate maintenance, images under legitimate and normal conditions are re-acquired, and the baseline file is updated. If the on-site inspection result indicates electricity theft, the evidence record is retained.

[0012] The beneficial effects of this invention are: This invention acquires reference images and establishes an archive using a pre-set fixed shooting unit and positioning markers; it performs quality screening and spatial position correction on the current image to output a standardized image; it extracts specific physical state features of the sealing area, meter area, and terminal pressure plate area, and performs item-by-item quantitative comparison with the reference archive; based on the feature difference results, it performs joint judgment, and outputs suspected electricity theft results and generates evidence records when progressive correlation conditions are met. This method filters out environmental displacement and light and shadow interference, achieves non-contact quantitative detection of minor abnormal tampering traces, reduces the false alarm rate, and provides objective comparison basis. Attached Figure Description

[0013] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 The overall process of visual inspection method for preventing electricity theft under high voltage metering conditions; Figure 2 This is a schematic diagram illustrating the acquisition and detection of state features in the sealed area. Figure 3 This is a schematic diagram of image comparison and difference detection for the meter area; Figure 4 A schematic diagram showing the detection of rewiring traces in the terminal pressure plate area; Figure 5 This is a schematic diagram illustrating its application in a high-voltage power distribution room in an industrial park. Detailed Implementation

[0015] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0016] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0017] Secondly, the term "one embodiment" or "example" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the invention. The appearance of an embodiment in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that mutually excludes other embodiments.

[0018] Example 1 Reference Figures 1-4 This is the first embodiment of the present invention, which provides a visual detection method for preventing electricity theft under high voltage metering conditions, including the following steps: S1. Obtain a reference image of the high-voltage metering device when it is in a legal and normal state and establish a reference file; wherein, the reference image is acquired by a fixed imaging unit preset in the high-voltage metering cabinet with the cooperation of a fixed supplementary lighting unit, and the high-voltage metering cabinet is preset with positioning marks, including the following steps: A first fixed imaging unit is pre-set inside the high-voltage metering cabinet facing the sealing area; a second fixed imaging unit is pre-set inside the high-voltage metering cabinet facing the meter area; and a third fixed imaging unit is pre-set inside the high-voltage metering cabinet facing the terminal pressure plate area. Fixed supplementary lighting units are pre-set in the imaging directions of the first, second, and third fixed imaging units, respectively. A first positioning mark is pre-set near the sealing hole; a second positioning mark is pre-set at the fixed corner of the meter; and a third positioning mark is pre-set at both ends of the terminal block. When the high-voltage metering device is in a legal and normal state, reference images of the sealing area, meter area, and terminal pressure plate area are acquired by the first, second, and third fixed imaging units, respectively. The current high-voltage metering cabinet number, user number, shooting time, fixed imaging unit number, and working status of the fixed supplementary lighting units are acquired, and the above information is stored in a reference file corresponding to the reference images of the sealing area, meter area, and terminal pressure plate area. The sealing area includes the sealing wire perforation path, sealing wire bending shape, sealing wire tightness, outer contour of the sealing component, pressure position of the sealing component, and edge marks of the sealing hole; the meter area includes the meter housing position, relative position of the meter housing and mounting base, display window boundary, nameplate position, fixing screw position, fixing screw slot orientation, and joint condition of the meter panel and mounting base; the terminal plate area includes the terminal sequence, number of terminal wires, wire color, wire entry direction into the plate, wire length protruding from the plate, first bend position of the wire, natural bending shape of the wire, edge fit of the plate, screw slot condition, and cable tie end condition.

[0019] It should be noted that during the system initialization and data archiving phases, the system first needs to establish a reference benchmark for subsequent comparisons and optimize the visual acquisition environment at the hardware level. The specific operation steps are as follows: Construction and Marking Deployment of Multi-Dimensional Visual Physical Environment Inside the Metering Cabinet: Due to the limited internal space and structural obstructions within the high-voltage metering cabinet, a single camera node cannot cover the entire area. Therefore, the system adopts a multi-node directional shooting architecture, independently setting up corresponding fixed shooting units for localized focused acquisition of the multi-layered protective structure inside the high-voltage metering cabinet from the outside in. Simultaneously, to ensure spatial consistency in subsequent image registration, high-contrast positioning markers are placed at holes, corners, and terminal block ends within the cabinet, thereby establishing a mapping benchmark between the three-dimensional physical space and the two-dimensional image pixels.

[0020] Initial state of reference image acquisition and metadata association: When power supply personnel complete the standardized installation and apply physical seals, or when legitimate maintenance work is completed and the equipment is restored, the system enters a legitimate and normal state. At this time, the main control computing device sends acquisition commands to the underlying imaging units and fixed supplementary lighting units to acquire high-definition reference images. The system synchronously reads the current cabinet number, user identification code, system absolute timestamp, and auxiliary equipment operating parameters, and fuses and encrypts the above textual metadata with the reference images to generate a uniquely identified reference digital archive.

[0021] S2. Acquire the current image captured by the fixed imaging unit at a preset position, and use the positioning markers and the fixed structure inside the high-voltage metering cabinet to perform position correction and quality screening on the current image to obtain a standardized image, including the following steps: The system acquires images of the current sealing area, the current meter area, and the current terminal pressure plate area, captured by the first, second, and third fixed shooting units at preset positions. It then performs quality screening on these images. If any of the following conditions are present in the images: out of focus, motion blur, reflection obstruction, shadow obstruction, local overexposure, local underexposure, lens surface contamination, cabinet door obstruction, or uneven lighting, the corresponding fixed supplementary lighting unit is controlled to re-light and triggers re-image acquisition. The initial reference position in the current image is determined using the first, second, and third positioning marks. Then, the left-right, up-down, rotational, and dimensional offsets of the current image relative to the reference image are determined using the cabinet door frame, the edge of the observation window, the edge of the cabinet mounting back panel, the features of the base mounting rail, and the end face of the terminal block. The current image is then reset based on the left-right, up-down, rotational, and dimensional offsets to align the sealing area, meter area, and terminal pressure plate area in the current image with the corresponding areas in the reference image.

[0022] It should be noted that during the routine automated inspection phase, the system needs to filter out visual interference caused by fluctuations in industrial lighting and minor mechanical displacements, and perform quality screening and spatial registration. The specific operation steps are as follows: Environmentally Adaptive Image Quality Screening and Light Source Closed-Loop Control: After the system acquires the current image, in order to objectively and quantitatively evaluate the image's sharpness and determine whether defocusing or blurring exists, the system internally uses the dispersion of extracted high-frequency edge information as the evaluation criterion. Its Laplacian variance evaluation formula is as follows:

[0023] This represents an objective evaluation value of the current image's sharpness. A higher value indicates sharper edges and higher clarity. The value range is [range missing]. ≥0.

[0024] , : These represent the total horizontal pixel width and total vertical pixel height of the currently acquired image, respectively, and both are positive integers.

[0025] , : Represents the two-dimensional pixel coordinates in the image matrix.

[0026] : Represents coordinates The edge gradient value of the pixel at that location after being processed by the Laplacian operator convolution.

[0027] : Represents the global average of the Laplacian gradient values ​​for the entire image.

[0028] The system calculates the sharpness rating value. The constraint is compared with the system's preset threshold. If... If the threshold is not met (i.e., if local darkness, reflection, or ghosting is detected, which does not meet the requirements for feature extraction), the system will block the subsequent processing of the image and dynamically adjust the output brightness or flicker parameters of the corresponding fixed supplementary lighting unit through control commands, and then trigger re-acquisition until all image quality indicators meet the requirements.

[0029] Secondary spatial registration based on physical anchor points and rigid structures: Due to environmental vibrations and thermal expansion and contraction of equipment, the camera optical axis often experiences microscopic shifts. After identifying the positioning anchor points and extracting the feature references of the rigid fixed structure, the system needs to eliminate perspective distortion and displacement errors in four-dimensional space. The algebraic expansion formula for resetting the two-dimensional pixel coordinates, derived based on the homography transformation principle, is as follows:

[0030]

[0031] , : These represent the horizontal and vertical coordinates of the uncorrected original pixels in the currently acquired skewed image, respectively.

[0032] , : These represent the horizontal and vertical coordinates of the pixel in the aligned reference state after the reset and correction process.

[0033] ~ : Represents the perspective transformation coefficients calculated by the system through matching feature points of a rigid, fixed structure. , Characterizes the horizontal and vertical translational shift of the image; , , , Control the rotation and scaling of the image; , Controlling perspective distortion of the viewpoint.

[0034] The system uses bilinear or bicubic interpolation algorithms to iterate through the coordinates of all pixels in the current image and substitute them into the formula for inverse resampling. This operation eliminates geometric deviation errors caused by mechanical misalignment, and finally outputs a standardized image that precisely overlaps with the reference image in the spatial pixel coordinate system.

[0035] S3. Extract the sealing area, meter area, and terminal pressure plate area from the standardized image, and extract the state features of the corresponding areas respectively, including the following steps: Extract the sealing wire perforation path, sealing wire bending shape, sealing wire tightness, outer contour of the sealing component, pressure position of the sealing component, and edge traces of the sealing hole in the sealing area; extract the meter housing position, relative position of the meter housing and mounting base, display window boundary, nameplate position, fixing screw position, fixing screw slot orientation, and joint condition of the meter panel and mounting base in the meter area; extract the terminal sequence, number of terminal wires, wire color, wire entry direction into the pressure plate, wire length protruding from the pressure plate, first bend position of the wire, natural bending shape of the wire, edge contact condition of the pressure plate, screw slot condition of the pressure plate, and end condition of the wire harness tie in the terminal pressure plate area.

[0036] It should be noted that, in order to convert the abnormal operational behavior of the physical implementation into computer-analyzable quantitative data, the system needs to perform feature analysis on the surface details and spatial geometric parameters of the core components. The specific operation steps are as follows: Fitting and Analysis of Morphological Parameters of Surface Physical Protection Components: The system uses an instance segmentation algorithm to extract the sealing area and the connected domain of a sealing line in two-dimensional space. A smooth curve equation for this connected domain is fitted using a polynomial, and the coefficients and radius of curvature of this equation are used as quantitative outputs representing the bending morphology of the sealing line. For the tightness of the sealing line, the system constructs a straightening rate quantification model, which is defined by calculating the ratio of the absolute straight-line distance between the holes at both ends of the sealing line to the actual fitted arc length of the sealing line. The calculation formula is as follows:

[0037] : This represents a quantitative indicator of the tightness of the sealing line, with a value range of (0,1). The closer the value is to 1, the tighter the sealing line is.

[0038] and : These represent the two-dimensional pixel coordinates of the sealing wire insertion hole and the sealing wire exit hole in the image, respectively.

[0039] : Represents the equation of the sealed smooth curve obtained by system fitting.

[0040] Simultaneously, the deformation indentation area generated by the sealing clamp pressure is extracted, and its relative pixel coordinates with respect to the reference feature point are output to characterize the pressure position of the seal. For the meter area, the system uses a line detection algorithm (such as Hough transform) to extract a line vector from the screw slot of the fixed cover, and outputs the absolute deflection angle of the line relative to the horizontal reference plane in the interval [0, 180∘), as the quantified value of the orientation of the fixed screw slot; at the same time, the edge detection operator is used to identify the contour lines of the meter panel and the mounting base, and the pixel-level average distance between the two contour lines is calculated as a quantified index of the joint state to quantify possible disassembly or displacement traces.

[0041] Sub-pixel level extraction of the physical morphology of secondary circuit wiring: For the terminal plate area, in addition to basic wire color extraction, the system employs sub-pixel edge measurement technology to calculate the vertical pixel distance between the wire insulation endpoint and the tangent of the metal plate. This distance is multiplied by the camera calibration coefficient and output as the actual physical distance in millimeters, serving as the feature of the wire's exposed length on the plate. To capture irreversible plastic deformation, the system tracks the contour of the wire from the terminal exit point and extracts the spatial coordinates and tangent angle of the first bend by locating the curvature extremum. For the edge-fitting state of the plate, the system introduces the Intersection over Union (IoU) model from target detection for quantification. The calculation formula is as follows:

[0042] : This indicates the degree of fit between the outer contour of the conductor and the edge of the pressure plate.

[0043] : Represents the bounding box at the end of the conductor insulation layer selected by the system.

[0044] : This indicates the lower edge boundary frame of the metal pressure plate selected by the system.

[0045] : indicates calculating the intersection ( ) or union ( The pixel area of ​​the region.

[0046] When a wire is improperly pulled out or short-circuited, the original overlapping area of ​​the bounding box will be significantly altered, leading to... The numerical value drops precipitously, allowing the system to capture microscopic physical deformation characteristics.

[0047] S4. Compare the state characteristics of the sealing area, meter area, and terminal pressure plate area with the corresponding area state characteristics in the reference file item by item to obtain the difference results of the sealing area, meter area, and terminal pressure plate area, including the following steps: The extracted sealing area features, including the sealing wire perforation path, sealing wire bending shape, sealing wire tightness, outer contour of the sealing component, pressure position of the sealing component, and sealing hole edge marks, are compared item by item with the corresponding area features in the benchmark archive to obtain the sealing area difference results. Similarly, the extracted meter area features, including the meter housing position, relative position between the meter housing and the mounting base, display window boundary, nameplate position, fixing screw position, fixing screw slot orientation, and the joint state between the meter panel and the mounting base, are compared item by item with the corresponding area features in the benchmark archive to obtain the meter area difference results. Finally, the extracted terminal pressure plate area features, including the terminal sequence, number of terminal wires, wire color, wire entry direction into the pressure plate, wire length protruding from the pressure plate, first bend position of the wire, natural bending shape of the wire, pressure plate edge fit state, pressure plate screw slot state, and wire harness tie end state, are compared item by item with the corresponding area features in the benchmark archive to obtain the terminal pressure plate area difference results.

[0048] It should be noted that, during the feature difference determination stage, to avoid the problem of poor anti-interference capability in the overall image comparison mode, the system performs a dimensionality reduction tolerance test on the specific geometric parameters and physical values ​​extracted from the standardized image. The quantitative comparison determination formula is as follows:

[0049] ,when

[0050] : indicates the first The absolute difference of the physical characteristic parameters always has ≥0.

[0051] : indicates the th element extracted from the current image Numerical features (such as angle scalar, subpixel length, etc.).

[0052] : Indicates the corresponding number that is pre-existing in the baseline file. The baseline value of each feature.

[0053] : Indicates the state change flag for this feature, with a value of only 0 or 1. It is assigned a value of 1 when the absolute difference condition within the parentheses exceeds the limit, otherwise it defaults to 0.

[0054] : indicates that the system is targeting this first A tolerance threshold is independently assigned to each specific physical characteristic to absorb reasonable physical thermal expansion and contraction as well as quantification errors.

[0055] Regarding the tolerance threshold The dynamic calibration rules are as follows: Under initial legal and normal conditions, the system acquires multiple frames of legal sample images over several consecutive time periods without any human interference, and calculates the variance of fluctuations in various feature parameters under natural conditions. Based on the discreteness criteria of data statistics, the system automatically establishes the threshold boundaries for each feature by combining the built-in maximum allowable deformation constant for thermal expansion and contraction. For example, for the orientation angle of the fixing screw slot, the corresponding independent tolerance threshold is usually calibrated to a natural vibration error range of 3°~5°; for the length of the wire exposed on the pressure plate, the corresponding independent tolerance threshold is calibrated to a sub-pixel offset equivalent to the actual physical size of 0.5mm~1.0mm.

[0056] The system calculates the absolute difference between various characteristic values ​​and performs constraint operations against the aforementioned scientifically calibrated independent tolerance thresholds. This processing method enables the system to accurately locate the specific component parts where physical changes have occurred.

[0057] S5. Determine sealing operation traces based on the difference results of the sealing area, determine meter operation traces based on the difference results of the meter area, and determine terminal rewiring traces based on the difference results of the terminal pressure plate area. When the sealing operation traces, meter operation traces, and terminal rewiring traces meet the preset correlation conditions, output a suspected electricity theft result, including the following steps: In the results of differences in the sealing area, if two or more of the following characteristics change: the path of the sealing wire perforation, the bending shape of the sealing wire, the pressure position of the sealing component, and the edge marks of the sealing hole, then traces of sealing operation are identified. In the results of differences in the meter area, if two or more of the following characteristics change: the relative position of the meter housing and the mounting base, the orientation of the fixing screw slots, the state of the joint between the meter panel and the mounting base, and the boundary of the display window, then traces of meter operation are identified. In the results of differences in the terminal pressure plate area, if two or more of the following characteristics change: the number of terminal wires, the color of the wires, the length of the wires exposed on the pressure plate, the position of the first bend of the wires, the natural bending shape of the wires, the edge contact state of the pressure plate, and the state of the pressure plate screw slots, then traces of terminal wire rewiring are identified. After determining the presence of seal and meter tampering traces, the high-voltage metering cabinet is marked as a key inspection area. If further investigation reveals terminal rewiring issues under this key inspection status, a suspected electricity theft result is output. When a suspected electricity theft result is output, an evidence record is generated, including a baseline image, current image, enlarged view, seal area difference results, meter area difference results, terminal plate area difference results, high-voltage metering cabinet number, user number, and shooting time. On-site inspection feedback instructions are received, and the on-site inspection results are written to the historical record based on these instructions. If the on-site inspection result indicates legitimate maintenance, images under legitimate and normal conditions are re-acquired, and the baseline file is updated. If the on-site inspection result indicates electricity theft, the evidence record is retained.

[0058] It should be noted that, in the process of abnormal result output and business closure, in order to filter out single-point misjudgments caused by environmental noise and form a complete chain of evidence, the system performs multi-variable logical verification and adaptive updates. The specific operation steps are as follows: Multi-feature joint judgment excluding natural displacement errors: In complex industrial environments, changes in the state of a single component may originate from mechanical vibration of the equipment. Therefore, the system incorporates a multi-feature joint fault-tolerant judgment mechanism to exclude interference from single physical noise points. The logical summation formula and associated constraints for regional anomaly judgment are as follows:

[0059]

[0060] : indicates the first The total number of features identified as having undergone a state change within a specific detection region, taking the value of a natural number.

[0061] : Indicates the current detection area category, such as sealing area, meter area, or terminal pressure plate area.

[0062] : indicates that the first The total number of features extracted and verified within each detection area, and It is a positive integer.

[0063] : indicates the first in this region A state mutation identifier for a specific characteristic.

[0064] The system accumulates and sums the characteristic mutation statuses within the target area. The system determines that the physical change in the area conforms to the linkage characteristics of human-forced disassembly (such as the deflection of the fixing screw and the synchronous increase of the panel joint seam) only when the total number of characteristic items with mutations in the area satisfies the above inequality correlation condition. This eliminates the possibility of occasional single component displacement errors caused by mechanical vibration of the equipment.

[0065] Progressive correlation early warning and business evidence solidification: Abnormal operations typically exhibit a progressive characteristic, moving from the surface to the core. Based on this, the system configures business linkage judgment: upon detecting peripheral seals and meter manipulation traces, the equipment is first changed to a priority inspection status; only when traces of terminal rewiring at the core level are also detected does the system confirm the complete chain of abnormal operation behavior, and then output a suspected electricity theft result. The system simultaneously encapsulates basic comparison diagrams and magnified local comparison diagrams, generating objective and tamper-proof evidence records to provide reliable evidentiary support for subsequent business processes.

[0066] Adaptive updates to data archives based on on-site feedback: For legitimate on-site maintenance authorized by power supply company personnel, the physical form of equipment will undergo compliant changes. Upon receiving an external on-site verification feedback instruction, if the instruction indicates legitimate maintenance, the main control unit will automatically trigger a new round of baseline image acquisition. The newly acquired data will overwrite the original baseline archive, enabling the system to adaptively and dynamically learn to adapt to compliant changes in form, ensuring the accuracy of long-term online monitoring.

[0067] Example 2 Reference Figure 5 In order to further clarify the technical solution of the present invention, the following example, taking a specific scenario of detecting and locking hidden electricity theft in a high-voltage power distribution room in an industrial park as the second embodiment of the present invention, will provide a more detailed description of the technical solution of the present invention.

[0068] In this embodiment, the anti-theft visual detection method for high-voltage metering status of the present invention was deployed in the high-voltage power distribution room of a large metal processing enterprise. One early morning, the high-voltage metering cabinet in the power distribution room was opened without authorization. Its internal anti-counterfeiting seal was damaged, the meter casing was disassembled, and the secondary circuit voltage lines on the terminal blocks were stripped, short-circuited, and then re-disguised.

[0069] Step S1: Pre-build a baseline file in a legal and normal state In the initial phase of equipment operation three months prior to the incident, power company personnel completed the standardized meter installation, power connection, and anti-counterfeiting seal application for the high-voltage metering cabinet. The system was triggered into a legitimate and normal operating state. The first, second, and third fixed imaging units, in conjunction with a fixed supplementary lighting unit, respectively captured high-definition images of the intact seals, meters, and terminal pressure plate areas. The system bound and encrypted the captured baseline images with the company's user number, the high-voltage metering cabinet number, and the system's absolute timestamp, generating a baseline digital file for the metering cabinet. Among the baseline characteristics recorded in this file, the seal curve was continuous and smooth, the meter screw slots maintained their initial diagonal angle, and there were no abnormally exposed metal segments at the root of the terminal block conductors.

[0070] Step S2: Adaptive Acquisition and Spatial Registration in Complex Environments In the early morning hours following the incident, the system performed a routine automated inspection. Due to the extremely low ambient light in the power distribution room and the slight mechanical vibration caused by the operation of large processing equipment in the vicinity, the system's image quality assessment module detected that the initial image's Laplacian variance was below the sharpness threshold. The system then issued a closed-loop control command to increase the output brightness of the fixed supplementary lighting unit by 30%, thereby re-acquiring a current image that met the feature extraction requirements. Addressing the microscopic shift in the camera's optical axis caused by the mechanical vibration, the system extracted positioning marks placed at the sealing holes, meter corners, and both ends of the terminal blocks, and calculated the perspective transformation matrix by combining this with the rigid linear features of the cabinet door frame. By performing inverse pixel resampling processing, the system eliminated translation and rotation errors caused by physical offsets, generating a standardized image precisely aligned with the reference image in the spatial coordinate system.

[0071] Step S3: Deep extraction of microscopic anti-theft features The system performs parallel feature extraction on the standardized image after position correction: For the sealing area, the instance segmentation model detects breakpoints and shape distortions in the original smooth curve, and the system extracts the current bending shape of the sealing line accordingly; simultaneously, the system detects a shift in the coordinates of the clamp indentation on the sealing component, and extracts the current pressure position of the sealing component. For the meter area, the system applies a straight line detection algorithm to calculate the notch angle of the fixing screw in the upper right corner, and uses it as the current fixing screw notch orientation feature; the edge detection operator calculates that the gap width at the junction of the meter panel and the base has significantly increased, and the system extracts it as the junction state feature. For the terminal pressure plate area, the system performs sub-pixel level edge detection on the wires, calculates the length of the metal segment exposed outside the pressure plate after the insulation layer is peeled off, and uses it as the wire exposed length feature; simultaneously, the system extracts the first bending position feature of the wire root with non-smooth curvature caused by abnormal external force pulling.

[0072] Step S4: Tolerance-based item-by-item fine-grained comparison Entering the feature comparison stage, the system substitutes the feature parameters extracted in step S3 into the absolute difference calculation and tolerance test models, respectively. When comparing the orientation of the fixing screw slot, the system calculates the absolute difference between the currently extracted angle value and the reference angle value. Since this difference exceeds the system's set natural tolerance threshold, the system assigns a change value to the state change flag for this feature. When comparing the exposed copper wire length, the system calculates the difference between the current exposed copper length and the reference no-exposed copper state. This difference also exceeds the quantization error threshold, and the system assigns a change value to the state change flag for this feature as well. Through the above quantization calculation process, the system clarifies the exceeding of limits for each specific physical morphological parameter and outputs the sealing area difference results, meter area difference results, and terminal pressure plate area difference results, respectively.

[0073] Step S5: Multidimensional Association Determination and Evidence Chain Consolidation Output The system ultimately executes a logical summation and comprehensive judgment process. The system performs multi-feature joint judgment to filter out occasional environmental noise interference: in the meter area difference results, since the state change indicators for both the fixing screw slot orientation and the joint seam status have changed, satisfying the constraint condition that at least two features have changed, the system determines that there are traces of meter operation. Similarly, based on the corresponding multi-feature change judgment rules, the system confirms the existence of sealing operation traces and terminal rewiring traces. The system triggers a progressive business association judgment logic: based on the confirmed sealing operation traces and meter operation traces, the system first marks the high-voltage metering cabinet as a key verification state. In this state, since the inner core area is also determined to have terminal rewiring traces, the complete chain of unauthorized operation behavior is confirmed and closed by the system. The system then outputs the highest-level suspected electricity theft warning through the interface. Simultaneously, the system backend automatically encapsulates an unalterable evidence record package with an absolute timestamp. This evidence record includes a magnified and highlighted comparison image of the screw with the deflected angle and the exposed metal terminal wiring. The evidence record was simultaneously pushed to the monitoring center, providing clear image comparison basis and technical data support for the power supply company's on-site verification and subsequent investigation of illegal electricity use.

[0074] In summary, this invention acquires reference images and establishes an archive using a pre-set fixed shooting unit and positioning markers; performs quality screening and spatial position correction on the current image to output a standardized image; extracts specific physical state features of the sealing area, meter area, and terminal pressure plate area, and performs item-by-item quantitative comparison with the reference archive; performs joint judgment based on feature difference results, and outputs suspected electricity theft results and generates evidence records when progressive correlation conditions are met. This method filters out environmental displacement and light and shadow interference, achieves non-contact quantitative detection of minor abnormal tampering traces, reduces false alarm rate, and provides objective comparison basis.

[0075] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A visual detection method for preventing electricity theft under high-voltage metering conditions, characterized in that, Performed by a computing device, including the following steps: S1. Obtain a reference image of the high-voltage metering device when it is in a legal and normal state and establish a reference file; wherein, the reference image is acquired by a fixed shooting unit preset in the high-voltage metering cabinet with the cooperation of a fixed supplementary lighting unit, and the high-voltage metering cabinet is preset with positioning marks; S2. Acquire the current image captured by the fixed shooting unit at a preset position, and use the positioning mark and the fixed structure inside the high-voltage metering cabinet to perform position correction and quality screening on the current image to obtain a standardized image; S3. Extract the sealing area, meter area and terminal pressure plate area from the standardized image, and extract the state features of the corresponding areas respectively; S4. Compare the state characteristics of the sealing area, the meter area and the terminal pressure plate area with the corresponding area state characteristics in the reference file item by item to obtain the sealing area difference results, meter area difference results and terminal pressure plate area difference results. S5. Determine sealing operation traces based on the difference results of the sealing area, determine meter operation traces based on the difference results of the meter area, determine terminal rewiring traces based on the difference results of the terminal pressure plate area, and output suspected electricity theft results when the sealing operation traces, the meter operation traces, and the terminal rewiring traces meet preset association conditions.

2. The visual detection method for preventing electricity theft under high-voltage metering status as described in claim 1, characterized in that, The fixed imaging unit, fixed supplementary lighting unit, and positioning markers pre-installed in the high-voltage metering cabinet in S1 include: A first fixed shooting unit is preset in the high-voltage metering cabinet facing the sealing area; a second fixed shooting unit is preset in the high-voltage metering cabinet facing the meter area; a third fixed shooting unit is preset in the high-voltage metering cabinet facing the terminal pressure plate area; fixed supplementary lighting units are preset in the shooting directions of the first fixed shooting unit, the second fixed shooting unit, and the third fixed shooting unit respectively; a first positioning mark is preset near the sealing hole; a second positioning mark is preset at the fixed corner of the meter; and a third positioning mark is preset at both ends of the terminal block.

3. The visual detection method for preventing electricity theft under high-voltage metering conditions as described in claim 2, characterized in that, The step S1, which involves acquiring a reference image and establishing a reference file, includes: When the high-voltage metering device is in a legal and normal state, the following reference images are acquired: the sealing area reference image, the meter area reference image, and the terminal pressure plate area reference image, respectively captured by the first fixed shooting unit, the second fixed shooting unit, and the third fixed shooting unit; the current high-voltage metering cabinet number, user number, shooting time, fixed shooting unit number, and fixed supplementary lighting unit working status are acquired, and the above information is stored in the reference file in correspondence with the sealing area reference image, the meter area reference image, and the terminal pressure plate area reference image.

4. The visual detection method for preventing electricity theft under high-voltage metering conditions as described in claim 3, characterized in that, The corresponding region status characteristics in the benchmark archive include: The sealing area includes the sealing wire perforation path, sealing wire bending shape, sealing wire tightness, outer contour of the sealing component, pressure position of the sealing component, and edge marks of the sealing hole; the meter area includes the meter housing position, relative position of the meter housing and mounting base, display window boundary, nameplate position, fixing screw position, fixing screw slot orientation, and joint condition of the meter panel and mounting base; the terminal pressure plate area includes the terminal sequence, number of terminal wires, wire color, wire entry direction into the pressure plate, wire length protruding from the pressure plate, first bend position of the wire, natural bending shape of the wire, edge fit of the pressure plate, screw slot condition of the pressure plate, and end condition of the wire harness tie.

5. The visual detection method for preventing electricity theft under high-voltage metering status as described in claim 2, characterized in that, S2 includes: Acquire images of the current sealing area, current meter area, and current terminal pressure plate area, captured by the first fixed shooting unit, the second fixed shooting unit, and the third fixed shooting unit at preset positions; perform quality screening on the images of the current sealing area, the current meter area, and the current terminal pressure plate area; if any of the following conditions exist in the images of the current sealing area, the current meter area, and the current terminal pressure plate area: out of focus, ghosting, reflection obstruction, shadow obstruction, local overexposure, local underexposure, lens surface contamination, cabinet door obstruction, and uneven supplementary lighting, control the corresponding fixed supplementary lighting unit to re-supplement the light and trigger re-acquisition of the image.

6. The visual detection method for preventing electricity theft under high-voltage metering status as described in claim 5, characterized in that, The position correction of the current image in step S2 includes: The initial reference position in the current image is determined using the first positioning mark, the second positioning mark, and the third positioning mark; then, the left-right offset, up-down offset, rotation offset, and scale offset of the current image relative to the reference image are determined using the cabinet door frame, the edge of the observation window, the edge of the cabinet mounting back panel, the features of the base mounting rail, and the end face of the terminal block; the current image is reset based on the left-right offset, the up-down offset, the rotation offset, and the scale offset, so that the sealing area, meter area, and terminal pressure plate area in the current image are aligned with the corresponding areas in the reference image.

7. The visual detection method for preventing electricity theft under high-voltage metering conditions as described in claim 1, characterized in that, The state features of the corresponding region are extracted in S3, including: Extract the sealing wire perforation path, sealing wire bending shape, sealing wire tightness, outer contour of the sealing component, pressure position of the sealing component, and edge traces of the sealing hole in the sealing area; extract the meter housing position, relative position of the meter housing and mounting base, display window boundary, nameplate position, fixing screw position, fixing screw slot orientation, and joint state of the meter panel and mounting base in the meter area; extract the terminal sequence, number of terminal wires, wire color, wire entry direction into the pressure plate, wire length exposed from the pressure plate, first bend position of the wire, natural bending shape of the wire, edge contact state of the pressure plate, screw slot state of the pressure plate, and end state of the wire harness tie in the terminal pressure plate area.

8. The visual detection method for preventing electricity theft under high-voltage metering status as described in claim 7, characterized in that, S4 includes: The extracted sealing area features, including the sealing wire perforation path, sealing wire bending shape, sealing wire tightness, outer contour of the sealing component, pressure position of the sealing component, and sealing hole edge marks, are compared item by item with the corresponding area state features in the reference file to obtain the sealing area difference results. Similarly, the extracted meter area features, including the meter housing position, relative position between the meter housing and the mounting base, display window boundary, nameplate position, fixing screw position, fixing screw slot orientation, and the joint state between the meter panel and the mounting base, are compared item by item with the corresponding area state features in the reference file to obtain the meter area difference results. Finally, the extracted terminal plate area features, including the terminal sequence, number of terminal wires, wire color, wire entry direction into the plate, wire length protruding from the plate, first bend position of the wire, natural bending shape of the wire, plate edge fit state, plate screw slot state, and wire harness tie end state, are compared item by item with the corresponding area state features in the reference file to obtain the terminal plate area difference results.

9. The visual detection method for preventing electricity theft under high-voltage metering conditions as described in claim 8, characterized in that, S5 includes: In the results of the differences in the sealing area, if two or more of the following characteristics change: the perforation path of the sealing line, the bending shape of the sealing line, the pressure position of the sealing component, and the edge marks of the sealing hole, then the sealing operation traces are determined to exist. In the results of the differences in the meter area, if two or more of the following characteristics change: the relative position of the meter housing and the mounting base, the orientation of the fixing screw slot, the state of the joint between the meter panel and the mounting base, and the boundary of the display window, then the meter operation traces are determined to exist. In the results of the differences in the terminal pressure plate area, if two or more of the following characteristics change: the number of terminal wires, the color of the wires, the length of the wires exposed on the pressure plate, the first bend position of the wires, the natural bending shape of the wires, the edge contact state of the pressure plate, and the state of the pressure plate screw slot, then the terminal rewiring traces are determined to exist.

10. The visual detection method for preventing electricity theft under high-voltage metering conditions as described in claim 9, characterized in that, The output of suspected electricity theft results includes: After determining the presence of traces of sealing and metering operations, the high-voltage metering cabinet is marked as a key inspection area. If, under this key inspection area, traces of terminal rewiring are further determined, a suspected electricity theft result is output. When the suspected electricity theft result is output, an evidence record is generated, including a baseline image, current image, magnified view, differences in the sealing area, differences in the metering area, differences in the terminal pressure plate area, the high-voltage metering cabinet number, the user number, and the shooting time. On-site inspection feedback instructions are received, and the on-site inspection results are written into the historical record according to the feedback instructions. If the on-site inspection result indicates legitimate maintenance, images under legitimate and normal conditions are re-acquired, and the baseline file is updated. If the on-site inspection result indicates electricity theft, the evidence record is retained.

Images