Method and system for detecting arc striking phenomenon in power distribution cabinet

By analyzing images within the distribution cabinet to identify elongated bright areas and brightness variation characteristics, and combining this with airflow contact and arcing phenomena, the problem of arc initiation being difficult to detect during routine inspections has been solved, enabling continuous monitoring and fault prevention within the distribution cabinet.

WO2026237436A1PCT designated stage Publication Date: 2026-11-19ELECTRIC POWER RES INST STATE GRID SHANXI ELECTRIC POWER +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
PCT/CN2026/072899
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2025-05-14
Filing Date
2026-01-15
Publication Date
2026-11-19

AI Technical Summary

Technical Problem

Conventional inspection methods are insufficient to detect discontinuous arc initiation phenomena, making it difficult to detect and prevent equipment failures in a timely manner.

Method used

By acquiring images inside the power distribution cabinet, long, bright strips are identified, divided into multiple sub-regions, and brightness differences and changes are judged. Combined with a pre-trained target detection model, the arc initiation phenomenon is identified, and through multiple image analyses, airflow contact and strong arc blowing phenomena are judged to achieve continuous monitoring.

Benefits of technology

It enables accurate identification of arcing phenomena in power distribution cabinets, breaking through the time limitations of conventional inspections. It can accurately judge in complex environments, prevent equipment failures, and reduce the complexity of failures and maintenance costs.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN2026072899_19112026_PF_FP_ABST
    Figure CN2026072899_19112026_PF_FP_ABST
Patent Text Reader

Abstract

A method and system for detecting arc striking phenomenon in a power distribution cabinet. The method comprises: acquiring an in-cabinet image of a power distribution cabinet, and when there is a long strip-shaped highlight area on the in-cabinet image, constructing end areas corresponding to two end positions of the long strip-shaped highlight area (S1); and dividing the long strip-shaped highlight area into a plurality of sub-areas, determining whether the number of long strip-shaped highlight targets contained in each sub-area is 1, if so, determining whether a brightness difference between the two end areas is greater than a preset brightness difference threshold, if so, determining whether the brightness from one end to the other end increases monotonically or decreases monotonically, and if so, determining that an arc striking phenomenon has occurred in the power distribution cabinet (S2).
Need to check novelty before this filing date? Find Prior Art

Description

Methods and systems for detecting arcing phenomena in power distribution cabinets

[0001] This application claims priority to Chinese Patent Application No. 202510618039.8, filed with the Chinese Patent Office on May 14, 2025, the entire contents of which are incorporated herein by reference. Technical Field

[0002] This application relates to the field of electrical distribution cabinet safety testing technology, for example to a method and system for detecting arcing phenomena in electrical distribution cabinets. Background Technology

[0003] In modern power systems, distribution cabinets are one of the key facilities in the power transmission and distribution process. Distribution cabinets typically contain many critical components, such as busbars and terminal blocks. These components play an important role in power transmission and distribution, working together to ensure that power can be safely, stably, and efficiently delivered to various electrical devices and areas.

[0004] In actual operation of power distribution stations, accidents caused by condensation frequently occur due to weather conditions. This is mainly because distribution cabinets are usually connected to cable channels for cable entry and exit. During rainy seasons such as summer, the humidity inside cable channels is typically high due to rainwater accumulation and poor ventilation. The accumulation of large amounts of moisture within the cable channels creates potential hazards for future problems. The situation is equally concerning during cold seasons such as autumn and winter. The temperature inside the cable channels is higher than the outdoor ambient temperature. When the cable openings are not properly sealed, hot and humid air enters the distribution cabinet from the bottom. In the low-temperature environment inside the cabinet, this air becomes supersaturated with water vapor, and some of the moisture condenses and covers the equipment surface. This condensed moisture reduces the insulation performance of cable terminals, insulating sleeves, and other equipment. When the insulation performance deteriorates to a certain level, it can easily trigger an electric arc. For example, at electrical connection points of equipment, where the insulation was originally good, the current flows steadily. However, when condensation causes the insulation performance to deteriorate, the air is broken down, forming an electric arc. This type of electric arc initiation can easily lead to abnormal phenomena such as flashover, which can not only directly cause equipment malfunctions but may also induce inherent defects in the equipment to develop into operational failures prematurely. In severe cases, electric arc initiation can cause insulation breakdown, and the powerful current can cause the terminal to burn out or even cause the cabinet to explode, posing a serious threat to the safe and stable operation of the power distribution station.

[0005] The discontinuous nature of electric arc initiation poses a significant challenge to conventional inspection methods. Conventional inspections largely rely on periodic manual checks, typically on a daily, weekly, or even monthly basis. Within such long time intervals, arc initiation may occur only for a brief moment, and if the initiation time falls outside the inspector's inspection period, it is easily missed. Furthermore, the environment inside the distribution cabinet is complex, containing numerous electrical components and wiring. These devices generate various sounds and heat signals during normal operation, which can easily interfere with the inspector's judgment of arc initiation, making it difficult for inspectors, even when on-site, to accurately identify the arc initiation from the complex background information. Summary of the Invention

[0006] Therefore, the technical problem to be solved by this application is to overcome the shortcomings of conventional inspection methods, which make it difficult to detect discontinuous arc initiation phenomena, resulting in equipment failures being difficult to detect and prevent in a timely manner.

[0007] To address the aforementioned technical problems, this application provides a method for detecting arcing phenomena in power distribution cabinets, comprising the following steps:

[0008] Acquire an image of the inside of the power distribution cabinet. When there is a long, bright strip in the image, construct the end regions corresponding to the two ends of the long, bright strip.

[0009] The elongated bright area is divided into multiple sub-regions. It is determined whether the number of elongated bright targets in each sub-region is 1. If so, it is determined whether the brightness difference between the two end regions of the elongated bright area is greater than the preset brightness difference threshold. If it is greater, it is determined whether the brightness change from one end to the other end is monotonically increasing or monotonically decreasing. If so, it is determined that an arcing phenomenon has occurred in the power distribution cabinet.

[0010] In some embodiments, the image inside the cabinet is used to identify the highlighted box region in the image inside the cabinet through a pre-trained object detection model;

[0011] Determine whether the aspect ratio of the highlighted area in the image inside the cabinet is greater than a preset ratio. If it is greater, determine that the highlighted area in the highlighted area is a long strip, extract the highlighted area, and obtain the long strip of highlighted area.

[0012] In some embodiments, after determining that an arcing phenomenon has occurred in the distribution cabinet, it is determined whether there is a highlighted box area in the cabinet image after a time interval that has an aspect ratio within a preset range. If so, it is determined whether there is a long strip of highlighted area used in the previous time interval when the arcing phenomenon was determined. If so, it is determined whether the highlighted box area intersects with the long strip of highlighted area used in the previous time interval when the arcing phenomenon was determined. If there is an intersection, it is determined that an airflow contact arcing phenomenon has occurred in the distribution cabinet in the current time interval. The upper limit of the preset range is less than the preset ratio.

[0013] After determining that an airflow contact arc phenomenon has occurred in the power distribution cabinet, it is determined whether there is a highlighted box area in the cabinet image after a time interval that has an aspect ratio within a preset range. If it exists, it is determined whether an arc ignition phenomenon has occurred in the power distribution cabinet in the current time sequence. If it has not occurred, it is determined that a strong airflow arc phenomenon has occurred in the power distribution cabinet in the current time sequence.

[0014] When the distribution cabinet experiences arc initiation, airflow contact with arc, and strong airflow blowing arc in three consecutive time sequences, it is determined that the distribution cabinet has experienced a flashover.

[0015] In some embodiments, an end region corresponding to each end is constructed based on the two end positions of the elongated highlighted region and the average width of the elongated highlighted region.

[0016] In some embodiments, the process of obtaining the average width of the elongated highlighted area includes:

[0017] Based on the direction of the longest side of the smallest bounding rectangle perpendicular to the highlighted area of ​​the long strip, draw multiple line segments that run through the highlighted area of ​​the long strip at preset intervals. Calculate the average distance between the two points where each line segment intersects the boundary of the highlighted area of ​​the long strip as the average width of the highlighted area of ​​the long strip.

[0018] In some embodiments, constructing the end region corresponding to each end based on the two end positions of the elongated highlighted region and the average width of the elongated highlighted region includes:

[0019] Using each end position as the center, construct a square or circular area with the average width of the elongated highlighted area as the side length or diameter, which serves as the end area corresponding to that end.

[0020] In some embodiments, if the angle between the line connecting one end of the elongated highlight area to the other end and the horizontal direction is less than 45°, the elongated highlight area is vertically divided into multiple sub-regions.

[0021] If the angle between the line connecting one end of the long strip of highlighted area and the horizontal direction is greater than 45°, then the long strip of highlighted area is horizontally divided into multiple sub-regions.

[0022] If the angle between the line connecting one end of a long strip of highlighted area and the horizontal direction is 45°, then the long strip of highlighted area is divided into multiple sub-regions horizontally or vertically.

[0023] In some embodiments, if the number of elongated bright targets contained in a sub-region of the current elongated bright region is greater than 1, the elongated bright region is determined not to meet the conditions for arc initiation, and no further judgment is made, and the next elongated bright region is processed.

[0024] In some embodiments, the average brightness of each pixel within each sub-region is taken as the brightness of each sub-region;

[0025] Based on the extension direction of the elongated bright region from one end to the other, the brightness difference between the brightness of the subsequent sub-region and the brightness of the preceding sub-region is determined, thus obtaining... There are several brightness differences; among them... Number of sub-regions;

[0026] statistics The number of values ​​greater than 0 and less than 0 in each brightness difference, if When the number of brightness differences greater than 0 exceeds a preset brightness difference threshold, the brightness of the elongated bright area is determined to be monotonically increasing from one end to the other; if... When the number of brightness differences less than 0 is greater than the preset brightness difference threshold, the brightness of the long strip of bright area is determined to be monotonically decreasing from one end to the other.

[0027] This application also provides a system for detecting arc initiation phenomena in power distribution cabinets, including:

[0028] The image acquisition module is configured to acquire images of the inside of the power distribution cabinet.

[0029] The end region construction module is configured to construct the end regions corresponding to the two ends of the long strip of highlighted area when there is a long strip of highlighted area in the image inside the cabinet.

[0030] The judgment module is set to divide the elongated bright area into multiple sub-regions, and determine whether the number of elongated bright targets contained in each sub-region is 1. If so, it determines whether the brightness difference between the two end regions of the elongated bright area is greater than the preset brightness difference threshold. If it is greater, it determines whether the brightness change from one end to the other end is monotonically increasing or monotonically decreasing. If so, it determines that an arcing phenomenon has occurred in the power distribution cabinet.

[0031] Compared with related technologies, the above-described technical solution of this application has the following advantages:

[0032] This application describes a method and system for detecting arcing phenomena in power distribution cabinets. By performing image detection within the cabinet, it overcomes the time limitations of conventional inspections and enables continuous monitoring of the cabinet's internal conditions. Since power distribution cabinets are typically dark, arcing will produce a bright glow, usually appearing as a long strip shape due to factors such as current and gas flow. This application utilizes this characteristic; when a long strip of bright area is present in the cabinet image, it preliminarily determines that arcing may have occurred. Considering that other factors such as light leakage from the distribution box or equipment reflection may also cause bright glow, this application further combines other characteristics of arcing for precise judgment. Based on the characteristic that arcing is curved but not spiral, the long strip of bright area is divided into multiple sub-regions. It is determined whether the number of long strip-shaped bright targets in each sub-region is always 1, thus eliminating some interference. Meanwhile, based on the characteristic of arcing phenomena where brightness gradually decreases from head to tail, if the brightness difference between the two end regions exceeds a preset brightness difference threshold, and the brightness from one end to the other is monotonically increasing or decreasing, by comprehensively considering various characteristics of arcing phenomena, high-brightness interference caused by other factors can be effectively eliminated. This achieves accurate identification of arcing phenomena within the distribution cabinet, breaking through the time limitations of conventional inspections and enabling continuous monitoring of the internal conditions of the distribution cabinet. This significantly improves the accuracy of arcing phenomenon judgment in complex environments. Furthermore, by accurately identifying arcing phenomena within the distribution cabinet, intervention can be carried out before equipment failure occurs, effectively preventing equipment failure, reducing the complexity of failures, avoiding extensive component damage and increased repair difficulty due to worsening failures, thereby reducing maintenance costs.

[0033] Furthermore, when airflow comes into contact with an electric arc, it disturbs the arc, altering the gas flow field and energy distribution around it. In the cabinet image, this disturbance appears as a large area that intersects with the original elongated, highlighted arc area. Therefore, after determining that an arc has occurred in the distribution cabinet, this application checks if a large area exists in the cabinet image after a time interval. This large area refers to a region where the aspect ratio of the elongated, highlighted arc is within a set range, and if it contains the elongated, highlighted arc area used in the previous time interval for determining arc initiation. If so, it checks if the highlighted area intersects with the elongated, highlighted arc area used in the previous time interval for determining arc initiation. If an intersection exists, it determines that an airflow-arc contact has occurred in the distribution cabinet in the current time interval. When an airflow-arc contact occurs, the powerful airflow disperses the arc, extinguishing it. At this point, no new arcs occur in the distribution cabinet, but the energy of the airflow remains. This will form a large area on the image inside the cabinet. Therefore, after determining that an airflow contact arc phenomenon has occurred inside the distribution cabinet, this application determines whether there is a highlighted area in the cabinet image after a time interval whose aspect ratio is within a preset range and no arc ignition phenomenon has occurred inside the distribution cabinet. If so, it is determined that a strong airflow arc phenomenon has occurred inside the distribution cabinet in the current time sequence. Arc ignition is the initial stage of flashover. Subsequently, airflow contacts the arc, causing changes in the arc shape and energy distribution. Finally, strong airflow arc extinguishes the arc or significantly weakens the arc energy. Therefore, this application determines whether a flashover phenomenon has occurred in the distribution cabinet by judging whether the distribution cabinet has arc ignition, airflow contact arc, and strong airflow arc phenomenon in three consecutive adjacent time sequences. This can prevent flashover phenomena from threatening the safe and stable operation of the substation and ensure the safe and stable operation of the distribution cabinet. Attached Figure Description

[0034] To make the content of this application easier to understand, the following detailed description is provided based on specific embodiments and accompanying drawings, wherein:

[0035] Figure 1 is a flowchart illustrating a method for detecting arcing phenomena in a power distribution cabinet according to this application;

[0036] Figure 2 is a schematic diagram of extracting the elongated highlight area;

[0037] Figure 3 is a schematic diagram of dividing a long strip of highlighted area into multiple sub-regions along the horizontal direction. Figure 3(a) is a schematic diagram in which the number of long strips of highlighted targets in each sub-region is 1. Figure 3(b) is a schematic diagram in which there are two or more long strips of highlighted targets in the sub-region.

[0038] Figure 4 is a flowchart illustrating the flashover phenomenon identification process;

[0039] Figure 5 is a schematic diagram of the highlighted box area identified when the image inside the cabinet contains a current image.

[0040] Figure 6 is a structural schematic diagram of a power distribution cabinet arc ignition detection device provided in an embodiment of this application. Detailed Implementation

[0041] The present application will be further described below with reference to the accompanying drawings and specific embodiments, so that those skilled in the art can better understand and implement the present application, but the embodiments are not intended to limit the present application.

[0042] In power systems, arc initiation phenomena often occur hidden inside distribution cabinets. Due to their discontinuous nature, they are difficult to detect using conventional inspection methods. Therefore, this embodiment proposes a method for detecting arc initiation phenomena in distribution cabinets, which can accurately and promptly detect whether abnormal arc initiation phenomena have occurred inside the distribution cabinet.

[0043] Referring to Figure 1, Embodiment 1 of this application provides a method for detecting arcing phenomena in a power distribution cabinet, comprising the following steps:

[0044] Step S1: Obtain an image of the inside of the power distribution cabinet. When there is a long, bright strip in the image, construct the end regions corresponding to the two ends of the long, bright strip.

[0045] In this embodiment, specifically, the process of obtaining the elongated highlighted area in the cabinet image is as follows:

[0046] Based on the image inside the cabinet, detect the highlighted bounding box region in the image inside the cabinet; within the highlighted bounding box region in the image inside the cabinet, extract the elongated highlighted region using an image segmentation model; the image segmentation model can be any one of the U-Net model, SegNet model, or FCN model.

[0047] The formula for identifying the highlighted area within the box in the cabinet image is as follows:

[0048] ;

[0049] Where box represents the highlighted area in the image inside the cabinet, bx represents the x-axis coordinate of the vertex of the highlighted area, by represents the y-axis coordinate of the vertex of the highlighted area, bw represents the width of the highlighted area, and bh represents the height of the highlighted area.

[0050] The method for detecting the highlighted box region in the image inside the cabinet can be any one of the pre-trained target detection model or region detection algorithm;

[0051] The target detection model is any one of the YOLOv10_L target detection model, Faster R-CNN model, and Mask R-CNN model; the region detection algorithm is any one of the OpenCV region detection algorithm and Canny edge detection algorithm.

[0052] In this embodiment, specifically, when the aspect ratio of the highlighted area in the frame of the image inside the cabinet is greater than a preset ratio, the highlighted area in the frame is determined to be a long strip; wherein, the preset ratio ranges from 3 to 8, and other larger ratios may also be selected, and this application does not specifically limit it.

[0053] The elongated shape can be a straight elongated shape or a curved elongated shape. This application does not specifically limit the shape. Since the arc initiation is usually a curved elongated shape, the highlighted area of ​​the above-mentioned elongated shape is usually a curved elongated area.

[0054] Figure 2 shows a schematic diagram of extracting elongated highlight areas. The elongated highlight areas in Figure 2 are curved elongated areas.

[0055] This embodiment detects the highlighted areas within the framed regions of the image inside the cabinet and then extracts elongated highlighted regions from these areas. This accurate extraction of elongated highlighted regions effectively eliminates interference from other irrelevant information in the image, focusing only on the key parts related to the anomaly. Compared to traditional methods that directly analyze the entire image, this method significantly improves the accuracy of anomaly detection. Because the complex environment inside the cabinet contains many factors that can affect judgment, such as reflections from the equipment itself and the illumination of normal indicator lights, the method in this embodiment can accurately identify the highlighted parts that truly represent the anomaly, reducing false positives. This provides a more reliable data foundation for the analysis and processing of arc initiation phenomena, ensuring the accuracy and reliability of the arc initiation phenomenon identification results.

[0056] In this embodiment, specifically, the end regions corresponding to the two ends of the elongated highlighted area are constructed, including:

[0057] Each end is centered on a region consisting of pixels with a range less than or equal to a preset distance, which is taken as the end region corresponding to that end.

[0058] The two ends of the elongated highlighted area are located at... , , The coordinates are , The coordinates are The two ends of the elongated highlighted area are the locations of the two endpoints of the elongated highlighted area.

[0059] In this embodiment, an end region corresponding to each end is constructed based on the two end positions of the elongated highlighted area and the average width of the elongated highlighted area.

[0060] Optionally, a square region is constructed with the average width of the elongated highlighted area as the side length, centered at each end position, to serve as the end region corresponding to that end, as shown in the formula:

[0061] ;

[0062] ;

[0063] in, This refers to the end region corresponding to the head end. The end region corresponding to the tail end, and the head end position. The coordinates are Tail end position The coordinates are , For the head end position x-coordinate For the head end position The ordinate, Tail end position x-coordinate Tail end position The ordinate, The average width of the elongated highlighted area. A function representation for constructing a rectangular region.

[0064] Optionally, a circular region is constructed with the average width of the elongated highlighted area as the diameter, centered on each end position, to serve as the end region corresponding to that end.

[0065] This embodiment constructs an end region corresponding to each end based on the two end positions of the elongated bright area and the average width of the elongated bright area. This ensures that the determined end region is consistent with the overall width of the elongated bright area, thus fully reflecting the overall brightness of each end region and preventing the accuracy of the end region brightness from being affected by the excessive size of the end region.

[0066] In this application, specifically, the process of obtaining the average width of the elongated highlighted area includes:

[0067] Draw multiple line segments that run through the long side of the smallest bounding rectangle of the long strip of highlighted area at preset intervals, perpendicular to the direction of the smallest bounding rectangle of the long strip of highlighted area. Calculate the average distance between the two points where each line segment intersects the boundary of the long strip of highlighted area, and use this average distance as the average width of the long strip of highlighted area.

[0068] Optionally, based on the direction perpendicular to the minor semi-major axis of the smallest circumscribed ellipse of the elongated highlighted area, multiple line segments are drawn at preset intervals through the elongated highlighted area. The average distance between the two points where each line segment intersects the boundary of the elongated highlighted area is calculated as the average width of the elongated highlighted area.

[0069] Optionally, the elongated highlighted area is divided into sections along its extension direction from one end to the other. A highlighted section, The average width of each highlighted segment is used as the average width of the elongated highlighted area, calculated using the following formula:

[0070] ;

[0071] In the formula, The average width of the elongated highlighted area. It is a function of average value. For the first The width of each highlighted section, ;

[0072] There are several ways to determine the width of each highlighted segment, as follows:

[0073] For each highlighted segment, arbitrarily select a position within it. Starting from the selected position, measure the distance from the selected position to one side boundary of the current highlighted segment along the extension direction of the elongated highlighted area from one end to the other. This distance is taken as the width of the current highlighted segment.

[0074] For each highlighted segment, determine its inner center position. Starting from this center position, along the extension direction of the elongated highlighted area from one end to the other, measure the distance from this center position to one side boundary of the current highlighted segment, which is taken as the width of the current highlighted segment.

[0075] For each highlighted segment, arbitrarily select multiple positions within it. Starting from each selected position, along the extension direction of the elongated highlighted area from one end to the other, calculate the average distance from each position to one side boundary of the current highlighted segment, which is used as the width of the current highlighted segment.

[0076] Step S2: Divide the elongated bright area into multiple sub-regions, and determine whether the number of elongated bright targets in each sub-region is 1. If so, determine whether the brightness difference between the two end regions of the elongated bright area is greater than the preset brightness difference threshold. If it is greater, determine whether the brightness change from one end to the other end is monotonically increasing or monotonically decreasing. If so, determine that an arcing phenomenon has occurred in the power distribution cabinet.

[0077] In this embodiment, specifically, the dividing direction is determined according to the extension direction from one end to the other end, and the elongated highlighted area is divided into multiple sub-regions;

[0078] The extension direction from one end to the other is based on the curved contour of the elongated highlighted area, starting from the starting end and following the curve's trajectory throughout until reaching the other end. For ease of calculation, the extension direction from one end to the other can also be a straight line connecting the two ends.

[0079] If the angle between the line connecting one end of the elongated highlighted area and the horizontal direction is less than 45°, then the elongated highlighted area is vertically divided into multiple sub-regions.

[0080] If the angle between the line connecting one end of the long strip of highlighted area and the horizontal direction is greater than 45°, then the long strip of highlighted area is horizontally divided into multiple sub-regions.

[0081] If the angle between the line connecting one end of a long strip of highlighted area and the horizontal direction is 45°, then the long strip of highlighted area is divided into multiple sub-regions horizontally or vertically.

[0082] In this case, the angle between the straight line connecting one end of the elongated highlighted area and the horizontal direction can be adjusted by tilting the angle. Indicates the tilt angle The formula for the tangent value is:

[0083] ;

[0084] in, The tilt angle, tilt angle The tangent value of .

[0085] If the angle between the direction of the straight line connecting one end of the long, bright strip and the horizontal direction is less than... ,Right now This indicates that the elongated highlighted area tends to extend horizontally. Therefore, the dividing direction is set to vertical to divide the elongated highlighted area into multiple sub-regions along its length. Similarly, if the angle between the straight line connecting one end of the elongated highlighted area and the horizontal direction is greater than [missing information], then [missing information]. ,Right now or This indicates that the elongated highlighted area tends to extend vertically, so the dividing direction is set to horizontal.

[0086] Different methods are used to divide the elongated highlighted areas at different angles, allowing the elongated highlighted areas to be divided along their length. This division allows the sub-regions to better reflect the actual shape of the electric arc. This enables more accurate capture of the monotonic changes in brightness from one end to the other, as well as the brightness of the end regions, when analyzing brightness distribution. It avoids detection loopholes due to the diversity of arc shapes, ensuring accurate identification of arc initiation phenomena within the distribution cabinet under various complex conditions.

[0087] As shown in Figure 3, Figure 3 is a schematic diagram of dividing the long strip of highlighted area into multiple sub-regions along the horizontal direction.

[0088] This application can divide a long strip of highlighted area into multiple sub-regions, or divide the image inside the cabinet into multiple sub-regions along the extension direction from one end to the other. Each sub-region contains a long strip of highlighted area, thereby dividing the long strip of highlighted area into multiple parts. Alternatively, the long strip of highlighted area can be divided into multiple sub-regions of unequal size. This application does not specifically limit the division.

[0089] Figure 3(a) shows a schematic diagram where the number of elongated bright targets in each sub-region is 1, and Figure 3(b) shows a schematic diagram where there are two or more elongated bright targets in a sub-region. When the elongated bright area is spiral-shaped, there will be two or more elongated bright targets in the sub-region. In this case, it indicates that the elongated bright area is not caused by an electric arc, but may be caused by other reasons, such as light leakage from the distribution box. Since the characteristic of an electric arc is that it is curved but not spiral, when the number of elongated bright targets in each sub-region is 1, it indicates that the characteristics of an arc are met, which can eliminate some interference and more accurately determine whether an electric arc has occurred in the distribution cabinet.

[0090] Since the arc initiation phenomenon is usually characterized by a large brightness at the beginning of the arc and a small brightness at the end, when determining whether a suspected arc initiation area is actually an arc initiation area, it is necessary to detect whether the brightness difference between the two end areas of the suspected arc initiation area is greater than a preset brightness difference threshold. If it is greater, it means that the suspected arc initiation area meets the condition that the brightness difference between the beginning and end of the arc is large.

[0091] In this embodiment, specifically, the brightness of each end region is any one of the following: the average brightness of all pixels in the end region, the brightness of the pixel at the middle position in the end region, or the brightness of the pixel at any position in the end region. Those skilled in the art can flexibly choose how to determine the brightness of the two end regions.

[0092] For each end region, the average brightness of all pixels in that region can be calculated. The average brightness of all pixels is then determined as the brightness of the end region. The average brightness value can more accurately and comprehensively reflect the overall brightness of the end region.

[0093] The formula for calculating the average brightness of each pixel in the two end regions is:

[0094] ;

[0095] ;

[0096] in, The end region corresponding to the head end The average brightness, End region corresponding to the tail end The average brightness, The end region corresponding to the head end The number of pixels, End region corresponding to the tail end The number of pixels, This refers to the pixel index in the end region corresponding to the head end. This refers to the pixel index in the end region corresponding to the tail end. for brightness, for The brightness.

[0097] Since the brightness of an electric arc typically decreases gradually from the beginning to the end, determining whether arc initiation has occurred requires checking whether the brightness from one end to the other is monotonically increasing or decreasing. If this is met, the elongated bright region satisfies the condition of gradually decreasing brightness from the beginning to the end of the arc. Specifically, if the brightness of the elongated bright region gradually decreases from one end to the other, it indicates that the starting end is the beginning of the arc and the ending end is the end of the arc; conversely, if the brightness of the elongated bright region gradually increases from one end to the other, it indicates that the starting end is the end of the arc and the ending end is the beginning of the arc.

[0098] In this embodiment, specifically, determining whether the brightness of the elongated bright area from one end to the other is monotonically increasing or monotonically decreasing includes:

[0099] Construct a brightness profile line that extends from one end of a long, bright strip to the other. Sample pixels uniformly along the brightness profile line. If the brightness of all sampled pixels is monotonically increasing or monotonically decreasing, then the brightness from one end of the long, bright strip to the other is monotonically increasing or monotonically decreasing.

[0100] The brightness profile line can be any one of a straight line, a broken line, or a curve.

[0101] Optionally, the method for determining whether the brightness of the elongated highlighted area increases or decreases monotonically from one end to the other is as follows:

[0102] The average brightness of each pixel within each sub-region is taken as the brightness of that sub-region, using the following formula:

[0103] ;

[0104] in, For the first The average brightness of each pixel within a sub-region , For the first The number of pixels in each sub-region For the first Pixel index within each sub-region for The brightness.

[0105] Based on the extension direction from one end to the other, the brightness of adjacent sub-regions is compared sequentially;

[0106] If the brightness of the next sub-region is always greater than that of the previous sub-region, the brightness from one end of the elongated bright region to the other is determined to be monotonically increasing; if the average brightness of the next sub-region is always less than that of the previous sub-region, the brightness from one end of the elongated bright region to the other is determined to be monotonically decreasing.

[0107] To more accurately detect whether the brightness of a long, bright strip gradually decreases or increases, the number of sub-regions can be chosen as an integer such as 4, 5, 6, 7, 8, 9, or 10. However, the number of sub-regions should not be too large to avoid computational complexity and improve computational efficiency and anomaly detection efficiency.

[0108] In some embodiments, the method for determining whether the brightness of the elongated bright area increases monotonically or decreases monotonically from one end to the other is as follows:

[0109] The average brightness of each pixel within each sub-region is taken as the brightness of each sub-region.

[0110] Based on the extension direction from one end to the other, the brightness difference between the brightness of the subsequent sub-region and the brightness of the preceding sub-region is determined, and thus... One brightness difference;

[0111] For example, the elongated highlighted area is divided into 4 sub-regions, namely There is a brightness difference between the second sub-region and the first sub-region, a brightness difference between the third sub-region and the second sub-region, and a brightness difference between the fourth sub-region and the third sub-region, resulting in a total of 3 brightness differences.

[0112] In this embodiment, the brightness difference between the brightness of the next sub-region and the brightness of the previous sub-region is: the difference obtained by subtracting the brightness of the previous sub-region from the brightness of the next sub-region.

[0113] statistics The formula for the number of brightness differences greater than 0 and less than 0 is:

[0114] ;

[0115] ;

[0116] in, for The number of brightness differences greater than 0 for The number of brightness differences less than 0 For sub-region indexing, The number of sub-regions. For the first The brightness of each sub-region For the first The brightness of each sub-region.

[0117] like When the number of brightness differences greater than 0 or less than 0 exceeds a preset brightness increment threshold, the brightness from one end of the elongated bright area to the other is determined to be monotonically increasing or monotonically decreasing.

[0118] For example, a long strip of bright area is divided into 4 sub-regions. The brightness values ​​of the first to fourth sub-regions are 15, 12, 20, and 25, respectively, and the three brightness differences are -3, 8, and 5, respectively. Among the three brightness differences, the number of values ​​greater than 0 is 2, and the number of values ​​less than 0 is 1. If the set brightness increment threshold is 1.8, and the number of values ​​greater than 0 among the three brightness differences is greater than 1.8, then the brightness from one end of the long strip of bright area to the other end is determined to be monotonically increasing or monotonically decreasing.

[0119] In practical applications, the brightness of the arc-shaped image captured inside the cabinet may be uneven. While the overall brightness of the arc-shaped light generally increases or decreases during arc formation, localized brightness flickering may occur. For example, the arc may show an overall decreasing brightness trend, but due to the instability of the arc formation, the brightness of the latter area may be higher than the former, failing to meet the decreasing trend. In such cases, the overall trend should be judged. In this embodiment, in... The number of brightness differences greater than 0 and If the number of any one of the brightness differences less than 0 is greater than the brightness increase threshold, it indicates that the proportion of the number of values ​​satisfying either a decrease or an increase is relatively large. Thus, the overall trend can satisfy either an increase or decrease. This setting method can effectively avoid missing the arc initiation and improve the accuracy of arc initiation detection.

[0120] In this embodiment, optionally, if the image inside the cabinet contains only a long strip of highlighted area, the end areas corresponding to the two ends of the long strip of highlighted area are constructed.

[0121] The elongated bright area is divided into multiple sub-regions. It is determined whether the number of elongated bright targets in each sub-region is 1. If so, it is determined whether the brightness difference between the two end regions is greater than the preset brightness difference threshold. If so, it is determined whether the brightness from one end to the other is monotonically increasing or monotonically decreasing. If so, it is determined that an arcing phenomenon has occurred in the power distribution cabinet.

[0122] Because various factors can cause bright areas to appear in the image within a distribution cabinet. For example, normal indicator lights or reflections from equipment surfaces can create bright areas, but these are not indicative of arcing. Only when the bright areas in the image are entirely due to arcing, without any other causes, can it be accurately determined that arcing has indeed occurred within the distribution cabinet. Therefore, ensuring that the image contains only one bright area effectively avoids misjudging arcing due to interference from other unrelated bright areas.

[0123] Figure 4 shows a flowchart illustrating the process of flashover identification.

[0124] As shown in Figure 5, Figure 5 is a schematic diagram of the highlighted box area identified when the image inside the cabinet contains an arcing phenomenon.

[0125] In this embodiment, specifically, step S31: after determining that an arcing phenomenon has occurred in the power distribution cabinet, it is determined whether there is a highlight box area in the cabinet image after an interval of one time sequence whose aspect ratio is in the range of 0.7 to 1.3, and whether there is a long strip of highlight area based on the previous time sequence when the arcing phenomenon was determined. If so, it is determined whether the highlight box area and the long strip of highlight area based on the previous time sequence when the arcing phenomenon was determined intersect. If there is an intersection, it is determined that an airflow contact arcing phenomenon has occurred in the power distribution cabinet in the current time sequence.

[0126] When airflow comes into contact with a burning electric arc, it disturbs the arc, altering the gas flow field and energy distribution around it. In the image inside the cabinet, this disturbance appears as a new bright area, with the aspect ratio of the frame containing this bright area ranging from 0.7 to 1.3, intersecting with the original long, bright area of ​​the electric arc. This is because the airflow brings a new energy source or changes the material distribution of the original arc area, resulting in a bright area that interacts with the original arc area. Therefore, when this image feature appears, it can be determined that airflow contact with an electric arc has occurred inside the distribution cabinet.

[0127] Step S32: After determining that an airflow contact arc phenomenon has occurred in the power distribution cabinet, determine whether there is a highlighted area in the cabinet image after a time interval whose aspect ratio is within a preset range and no arc ignition phenomenon has occurred in the power distribution cabinet. If so, it is determined that a strong airflow arc phenomenon has occurred in the power distribution cabinet in the current time sequence.

[0128] When a strong airflow causes an electric arc to form, the powerful airflow disperses the arc, extinguishing it and preventing new arcing from occurring within the distribution cabinet. However, the energy of the airflow remains, forming a bright area on the cabinet's image, with the aspect ratio of the box containing this bright area within a preset range. This is because the airflow may carry substances heated by the electric arc, or the airflow itself may interact with other substances within the distribution cabinet, producing a luminescent phenomenon. Therefore, the appearance of a bright area of ​​a specific shape in the image, even without an electric arc, indicates that a strong airflow causing an electric arc has occurred within the distribution cabinet.

[0129] Step S33: When the distribution cabinet experiences arc initiation, airflow contact with arc, and strong airflow blowing arc in three consecutive adjacent time sequences, it is determined that the distribution cabinet has experienced a flashover.

[0130] Flashover is a gas discharge phenomenon along an insulating surface, typically accompanied by a series of complex physical processes. Arc initiation is the initial stage of flashover; subsequently, the gas flow contacts the arc, altering its shape and energy distribution; finally, the strong gas flow extinguishes the arc or significantly weakens its energy. This series of phenomena, occurring sequentially, reflects the generation, transfer, and dissipation of energy during flashover. Therefore, when these three phenomena are detected occurring in sequence, it can be determined that a flashover has occurred in the distribution cabinet.

[0131] The pre-trained highlight region detection model is used to detect the highlighted bounding box regions in the image. The aforementioned large-area regions are those where the maximum circumscribed ellipse of the elongated highlight region is close to a circle, that is, the absolute value of the difference between the length of the major axis and the minor axis of the maximum circumscribed ellipse of the elongated highlight region is less than a preset first length threshold, and the ratio of the area of ​​the elongated highlight region to the area of ​​its maximum circumscribed ellipse is greater than 0.7; or large-area regions where the maximum circumscribed rectangle of the elongated highlight region is close to a square, that is, the absolute value of the difference between the length and the width of the maximum circumscribed rectangle of the elongated highlight region is less than a preset second length threshold, and the ratio of the area of ​​the elongated highlight region to the area of ​​its maximum circumscribed rectangle is greater than 0.7.

[0132] In this context, the preceding time sequence and the next adjacent time sequence in three consecutive time sequences are two time sequences separated by a preset time period. That is, the interval between two adjacent time sequences is within the preset time period, which can be any duration from 1 second to 5 seconds. Those skilled in the art can set the interval between two adjacent time sequences according to the specific duration of the flashover. By judging whether three characteristics—arc initiation, airflow contact with arc, and strong airflow blowing arc—occur sequentially within the distribution cabinet in the three adjacent time sequences, it is possible to accurately determine whether a flashover has occurred.

[0133] This embodiment can detect various types of abnormalities based on the different highlights in the image inside the cabinet, and promptly issue prompts to maintenance personnel as needed.

[0134] In this embodiment, it further includes: when an arcing phenomenon is detected in the power distribution cabinet, an abnormal prompt message is sent to the user terminal equipment, the abnormal prompt message being used to indicate the abnormal arcing situation occurring in the power distribution cabinet.

[0135] In practical applications, power distribution cabinet fault warnings are usually only known to maintenance personnel after users report that electrical equipment has malfunctioned or circuit components have reported errors. At this point, because the components have already failed, the repair is more complex and costly. In addition, it also has a significant impact on the user's power consumption.

[0136] This application overcomes the shortcomings of conventional inspection methods, which make it difficult to detect discontinuous arcing phenomena and thus difficult to detect and prevent equipment failures in a timely manner. By visually identifying arcing phenomena, potential hazards in the distribution cabinet can be detected in a timely manner. Before the equipment fails, abnormalities can be detected quickly, thus better avoiding further equipment failures, allowing for early repairs, reducing maintenance costs, and minimizing the impact on users' power supply.

[0137] Embodiment 2 of this application provides a system for detecting arcing phenomena in power distribution cabinets, including:

[0138] The image acquisition module is configured to acquire images of the inside of the power distribution cabinet.

[0139] The end region construction module is configured to construct the end regions corresponding to the two ends of the long strip of highlighted area when there is a long strip of highlighted area in the image inside the cabinet.

[0140] The judgment module is set to divide the elongated bright area into multiple sub-regions, and determine whether the number of elongated bright targets contained in each sub-region is 1. If so, it determines whether the brightness difference between the two end regions of the elongated bright area is greater than the preset brightness difference threshold. If it is greater, it determines whether the brightness change from one end to the other end is monotonically increasing or monotonically decreasing. If so, it determines that an arcing phenomenon has occurred in the power distribution cabinet.

[0141] In this embodiment, specifically, the image acquisition module can be a binocular camera, a monocular camera, or other types of cameras. By installing the camera inside the power distribution cabinet, images inside the cabinet can be captured. The camera can be installed inside the power distribution cabinet by magnetic attraction, or by bolt connection, snap-fit, or other connection methods. The number of cameras can be one or more, with cameras installed in multiple locations to capture images inside the power distribution cabinet from different angles. This application does not specifically limit the number of cameras.

[0142] Referring to Figure 6, Embodiment 3 of this application also provides a device for detecting arcing phenomena in a power distribution cabinet, including: a processor 10, a memory 20, a communication bus 30, and a communication interface 40;

[0143] The memory 20 is configured to store computer programs, and the processor 10 is configured to implement the steps of the above-mentioned method for detecting arcing phenomena in a power distribution cabinet when executing the computer program.

[0144] Embodiment 4 of this application provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the steps of the above-described method for detecting arcing phenomena in a power distribution cabinet.

[0145] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0146] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in one or more blocks of the flowchart illustrations and / or one or more blocks of the block diagrams.

[0147] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means that implement the functions specified in one or more flowcharts and / or one or more block diagrams.

[0148] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, such that the instructions, which execute on the computer or other programmable apparatus, provide steps for implementing the functions specified in one or more flowcharts and / or one or more block diagrams.

Claims

1. A method for detecting arc initiation in a power distribution cabinet, comprising the following steps: Acquire an image of the inside of the power distribution cabinet. When there is a long, bright strip in the image, construct the end regions corresponding to the two ends of the long, bright strip. The elongated bright area is divided into multiple sub-regions. It is determined whether the number of elongated bright targets in each sub-region is 1. If so, it is determined whether the brightness difference between the two end regions of the elongated bright area is greater than the preset brightness difference threshold. If it is greater, it is determined whether the brightness change from one end to the other end is monotonically increasing or monotonically decreasing. If so, it is determined that an arcing phenomenon has occurred in the power distribution cabinet.

2. The power distribution cabinet arc flash detection method of claim 1, wherein, The image inside the cabinet is used to identify the highlighted box area in the image through a pre-trained object detection model; Determine whether the aspect ratio of the highlighted area in the image inside the cabinet is greater than a preset ratio. If it is greater, determine that the highlighted area in the highlighted area is a long strip, extract the highlighted area, and obtain the long strip of highlighted area.

3. The method for detecting arc initiation in a power distribution cabinet according to claim 2, further comprising, after determining that arc initiation has occurred in the power distribution cabinet: Determine whether there is a highlighted box area with an aspect ratio within a preset range in the cabinet image after a time interval. If so, determine whether there is a long strip of highlighted area used in the previous time interval when the arc initiation phenomenon was determined. If so, determine whether the highlighted box area intersects with the long strip of highlighted area used in the previous time interval when the arc initiation phenomenon was determined. If there is an intersection, determine that an airflow contact arc phenomenon has occurred in the distribution cabinet in this time interval. The upper limit of the preset range is less than the preset ratio. After determining that an airflow contact arc phenomenon has occurred in the distribution cabinet, it is determined whether there is a highlighted box area with an aspect ratio within a preset range in the cabinet image after a time interval. If it exists, it is determined whether an arc ignition phenomenon has occurred in the distribution cabinet at that time interval. If it has not occurred, it is determined that a strong airflow arc phenomenon has occurred in the distribution cabinet at that time interval. When the distribution cabinet experiences arc initiation, airflow contact with arc, and strong airflow blowing arc in three consecutive time sequences, it is determined that the distribution cabinet has experienced a flashover.

4. The power distribution cabinet arc flash detection method of claim 1, wherein, Based on the two end positions of the elongated highlighted area and the average width of the elongated highlighted area, an end region corresponding to each end is constructed.

5. The power distribution cabinet arc flash detection method of claim 4, wherein, The process of obtaining the average width of the elongated highlighted area includes: Based on the direction of the longest side of the smallest bounding rectangle perpendicular to the highlighted area of ​​the long strip, draw multiple line segments that run through the highlighted area of ​​the long strip at preset intervals. Calculate the average distance between the two points where each line segment intersects the boundary of the highlighted area of ​​the long strip as the average width of the highlighted area of ​​the long strip.

6. The power distribution cabinet arc flash detection method of claim 5, wherein, The method for constructing an end region corresponding to each end based on the two end positions of the elongated highlighted region and the average width of the elongated highlighted region includes: Using each end position as the center, construct a square or circular area with the average width of the elongated highlighted area as the side length or diameter, which serves as the end area corresponding to that end.

7. The power distribution cabinet arc flash detection method of claim 1, wherein, If the angle between the line connecting one end of the elongated highlighted area and the horizontal direction is less than 45°, then the elongated highlighted area is vertically divided into multiple sub-regions. If the angle between the line connecting one end of the long strip of highlighted area and the horizontal direction is greater than 45°, then the long strip of highlighted area is horizontally divided into multiple sub-regions. If the angle between the line connecting one end of a long strip of highlighted area and the horizontal direction is 45°, then the long strip of highlighted area is divided into multiple sub-regions horizontally or vertically.

8. The power distribution cabinet arc flash detection method of claim 1, wherein, If the number of elongated bright areas contained in a sub-region is greater than 1, then the elongated bright area is determined not to meet the conditions for arc initiation, and no further judgment is made; the next elongated bright area is then processed.

9. The power distribution cabinet arc flash detection method of claim 1, wherein, Methods for determining whether the brightness of a long, bright strip region increases or decreases monotonically from one end to the other include: The average brightness of each pixel within each sub-region is taken as the brightness of each sub-region. According to the extension direction of the long strip-shaped highlight region from one end to the other end, the luminance difference between the luminance of the latter sub-region and the luminance of the former sub-region is determined, to obtain a difference in brightness; wherein This represents the number of sub-regions. Statistics the number of the luminance differences greater than 0 and the number of the luminance differences less than 0, and if the number of the luminance differences greater than 0 is greater than a predetermined luminance difference number determination threshold, determining that the luminance from one end portion to the other end portion of the long strip-shaped highlight region is monotonously increasing; and if the number of the luminance differences greater than 0 is less than the predetermined luminance difference number determination threshold, determining that the luminance from one end portion to the other end portion of the long strip-shaped highlight region is monotonously decreasing. When the number of brightness differences less than 0 is greater than the preset brightness difference threshold, the brightness of the long strip of bright area is determined to be monotonically decreasing from one end to the other.

10. A system for detecting arc initiation in a power distribution cabinet, comprising: The image acquisition module is configured to acquire images of the inside of the power distribution cabinet; The end region construction module is configured to construct the end regions corresponding to the two ends of the long strip of highlighted area when there is a long strip of highlighted area in the image inside the cabinet. The judgment module is set to divide the elongated bright area into multiple sub-regions, and determine whether the number of elongated bright targets contained in each sub-region is 1. If so, it determines whether the brightness difference between the two end regions of the elongated bright area is greater than the preset brightness difference threshold. If it is greater, it determines whether the brightness change from one end to the other end is monotonically increasing or monotonically decreasing. If so, it determines that an arcing phenomenon has occurred in the power distribution cabinet.