A multi-specification control cabinet fault visual diagnosis system
By dividing the industrial control cabinet into zones and dynamically adjusting the frame rate, the resource shortage problem caused by high frame rate acquisition is solved, enabling effective capture and diagnosis of transient electrical faults and ensuring long-term stable operation of the system.
Patent Information
- Application Number
- CN202611146119.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-30
- Publication Date
- 2026-11-17
AI Technical Summary
In the current technology for diagnosing transient electrical faults in industrial control cabinets, high frame rate acquisition leads to excessive pressure on transmission, storage and heat dissipation, and conventional frame rates cannot effectively capture transient signals.
The control cabinet panel is divided into areas for indicator lights, analog instruments, and digital displays using a panoramic partitioning module. The frame rate of the local area is increased only when there is a suspected transient event, and the frame rate is reduced according to the frequency of events when resources are scarce. Combined with the load reduction and recovery module, resource usage is optimized.
With limited resources, a reliable transient fault capture and visual diagnostic system was successfully deployed for long-term stability, reducing transmission and storage pressure and improving system reliability and stability.
Abstract
Description
Technical Field
[0001] This invention relates to the field of visual diagnostic technology, and more specifically, to a multi-specification control cabinet fault visual diagnostic system. Background Technology
[0002] Electrical faults in industrial control cabinets often manifest themselves in transient forms, such as intermittent flickering of tens of milliseconds caused by poor contact, momentary power outages and restorations due to voltage dips, and brief arcing in the early stages of insulation degradation. These phenomena are extremely short-lived, and visual acquisition at conventional frame rates will completely miss them because the sampling interval is longer than the duration of the fault. To capture such transient signals, the intuitive approach is to continuously record at a high frame rate sufficient to resolve millisecond-level changes.
[0003] However, in actual industrial settings, transient fault characteristics are often only displayed in a specific area on the control panel, such as a particular indicator light, a display screen, or an instrument panel. The remaining large areas of the panel do not require high frame rate acquisition most of the time. If the frame rate is uniformly increased across the entire screen, a large amount of image data consumes resources in transmission, storage, and processing without contributing effective diagnostic information. At the same time, the bandwidth of available transmission cables and the capacity of edge storage media in industrial settings are very limited. Furthermore, long-term full-load operation of edge processing units can lead to heat dissipation problems, increasing protection and maintenance costs. Summary of the Invention
[0004] In order to overcome the above-mentioned defects of the prior art, embodiments of the present invention provide a multi-specification control cabinet fault visual diagnosis system to solve the problems mentioned in the background art.
[0005] To achieve the above objectives, the present invention provides the following technical solution: A multi-specification control cabinet fault visual diagnosis system includes a panoramic partitioning module, a transient marking module, a region frame-up module, and a load reduction recovery module; The panoramic partitioning module is used to acquire panoramic images of the current control cabinet, simultaneously obtain the camera's acquisition frame rate, storage bandwidth margin, and processing temperature, and identify and divide the indicator light area, analog instrument area, and digital display screen area in the panoramic image. The transient marking module is used to continuously monitor the image sequence of each area. When any event such as the indicator light turning on and off, the instrument pointer shaking, or the screen characters flashing is captured, it is marked as a suspected transient event, and the duration of a single suspected transient event and the frequency of occurrence per unit time are recorded. The regional frame-up module is used to increase the acquisition frame rate in the region where the suspected transient event occurs when the duration of a single event is less than or equal to the single frame time interval of the current acquisition frame rate, while keeping the acquisition frame rate of the remaining regions unchanged. The load reduction recovery module is used to continuously acquire real-time values of processing temperature and storage bandwidth margin during the execution of regional frame rate increase. When the processing temperature reaches the preset upper temperature threshold or the storage bandwidth margin is lower than the preset lower bandwidth threshold, the acquisition frame rate of the corresponding region is reduced sequentially from low to high according to the frequency of occurrence of suspected transient events in each region, until the processing temperature falls back below the upper temperature threshold and the storage bandwidth margin rises back above the lower bandwidth threshold, and then the process is restored.
[0006] In a preferred embodiment, the panoramic partitioning module specifically includes: The camera is triggered to capture the entire control cabinet panel, and the resulting image is used as a panoramic image. At the same time, the current acquisition frame rate of the camera is read, and the storage bandwidth usage ratio of the edge processing unit and the junction temperature of the processing chip are read simultaneously. The storage bandwidth margin is obtained by subtracting the used bandwidth from the total storage bandwidth, and the junction temperature is used as the processing temperature. Search for the nameplate area on the control cabinet panel in the panoramic image, use optical character recognition to extract the cabinet type code in the nameplate, and retrieve the area division template corresponding to the code from the pre-built benchmark template library based on the cabinet type code; The normalized bounding box coordinates recorded in the region division template are mapped to pixel coordinates in the panoramic image through the homography transformation matrix between the panoramic image and the template base image. Based on the mapped pixel coordinates, each region of interest is segmented from the panoramic image, and according to the classification labels recorded in the region division template, each region of interest is assigned to the indicator light region, analog instrument region, and digital display screen region, respectively.
[0007] In a preferred embodiment, the transient marking module specifically includes: For the indicator light area, retrieve the normal state reference of the area from the area division template. The normal state reference includes the color range of the indicator light when it is lit and the average brightness value when it is off. Compare the average brightness of the indicator light pixels in the current frame image block with the average brightness in the off state. If the current average brightness exceeds the preset multiple of the off brightness for the first time, it is determined to be a lighting event, and the current frame number is recorded as the lighting start frame. If the current average brightness value drops back to within a preset multiple and is already in the lit state, it is determined to be an off event, and the current frame number is recorded as the off frame. Divide the difference between the frame number of the off frame and the frame number of the on start frame by the sampling frame rate to obtain the duration of a single on / off transition. The frequency of occurrence per unit time is obtained by counting the number of times the screen turns on and off within the most recent observation window and dividing by the duration of the observation window.
[0008] In a preferred embodiment, the transient marking module specifically includes: For the analog instrument area, retrieve the pointer scale range of the instrument under normal conditions from the area division template; Extract the pointer centerline from the current frame image block, calculate the angle between the pointer centerline and the zero scale baseline of the dial, and map the angle to the current pointer reading; When the change in pointer reading between two adjacent frames exceeds the preset jitter judgment threshold, and the direction of change reverses at least once within three consecutive frames, it is judged as a pointer jitter event, and the starting frame number of the jitter event is recorded. When the change in pointer readings for several consecutive frames is less than the jitter detection threshold, the jitter event is determined to be over, and the end frame number is recorded. Divide the difference between the end frame number and the start frame number by the acquisition frame rate to obtain the duration of a single pointer jitter. The frequency of pointer jitter events per unit time is obtained by counting the number of times the event occurs within the most recent observation window and dividing by the duration of the observation window.
[0009] In a preferred embodiment, the transient marking module specifically includes: For the digital display screen area, retrieve the normally displayed character content of the display screen in the normal state from the area division template; The character sequence currently displayed is extracted from the image block of the current frame. The current character sequence is compared with the character sequence of the previous frame character by character. When the character sequence changes and the changed character sequence recovers to be consistent with the original character sequence within no more than two consecutive frames, it is determined as a character flashing event, and the starting frame number of the flashing event is recorded. When the character sequence remains stable for three consecutive frames after recovery, the blinking event is considered to have ended, and the end frame number is recorded. Divide the difference between the end frame number and the start frame number by the acquisition frame rate to obtain the duration of a single character flash. Count the number of times the character blinking event occurs within the most recent observation window, divide by the duration of the observation window, and obtain the frequency of occurrence per unit time.
[0010] In a preferred embodiment, the regional frame upscaling module specifically includes: Read the duration of a single suspected transient event, obtain the frame rate of the current full-frame capture by the camera, and calculate the current single-frame time interval; If the duration of a single event is less than or equal to the time interval of a single frame, a frame rate boost is triggered for the target area where a suspected transient event has occurred. From the region segmentation results of the panoramic image, the pixel coordinate range covered by the target region is extracted. The pixel coordinate range is obtained by mapping the bounding box of the target region in the panoramic image to the camera sensor pixel coordinate system through the homography transformation matrix and then rounding it.
[0011] In a preferred embodiment, the regional frame upscaling module specifically includes: A region windowing acquisition command is sent to the camera. The command includes the pixel coordinate range and the target frame rate. The camera maintains the original full-screen acquisition channel and continues to output at the current frame rate. At the same time, a new independent windowing acquisition channel is opened to output a local high frame rate image sequence at the target frame rate.
[0012] In a preferred embodiment, the load reduction recovery module specifically includes: During the windowed high frame rate acquisition in the target area, the real-time values of the processing temperature and storage bandwidth margin of the edge processing unit are continuously read at a fixed sampling period. When either the processing temperature is greater than or equal to the upper temperature threshold, or the storage bandwidth margin is less than or equal to the lower bandwidth threshold, the regional load reduction process is triggered. The regional load reduction process is as follows: record the current acquisition frame rate of each region at the start of load reduction as its own baseline frame rate, extract the occurrence frequency of suspected transient events in each region within the most recent observation window from the monitoring records of each region, sort all regions in order of occurrence frequency from low to high to form a load reduction order list, and reduce the current actual acquisition frame rate of each region to half of its baseline frame rate in order of the load reduction order list. If the reduced frame rate is lower than the acquisition frame rate of the panoramic image, it is directly reduced to the acquisition frame rate of the panoramic image. After performing a frame rate reduction operation on each region, wait for one fixed sampling period, reread the processing temperature and storage bandwidth margin, and if the triggering conditions are still met, continue to perform a frame rate reduction operation on the next region; otherwise, stop reducing the load.
[0013] In a preferred embodiment, in the load reduction recovery module, when the processing temperature drops below the upper temperature threshold and the storage bandwidth margin rises above the lower bandwidth threshold, the current actual acquisition frame rate of each reduced region is restored to its respective base frame rate in reverse order of the load reduction order list, starting from the last reduced region. After each region is restored, wait for a fixed sampling period to confirm that the processing temperature and storage bandwidth margin are still within a safe range before restoring the next region, until all deloaded regions have been restored to their respective baseline frame rates.
[0014] In a preferred embodiment, in the transient marking module, the frequency of occurrence of areas where no suspected transient events have occurred is recorded as 0.
[0015] The technical effects and advantages of this invention are as follows: 1. This invention addresses the contradiction of excessive transmission, storage, and heat dissipation pressure caused by high frame rate acquisition across the entire industrial field, making long-term system deployment difficult, while conventional frame rates inevitably miss transient electrical faults. By pre-dividing the control cabinet panel into three areas—indicator lights, analog instruments, and digital displays—the frame rate is increased only in the area where a suspected transient event occurs when the duration is less than the current frame interval, while the frame rate remains unchanged in other areas. This limits the extra data generated by the high frame rate to a localized area containing effective diagnostic information. During the high frame rate execution in a region, when the processing temperature or storage bandwidth margin reaches a preset safety threshold, the frequency is reduced sequentially from low to high according to the frequency of events in each region, prioritizing the release of bandwidth and thermal margin in low-value regions. Thus, with limited physical resources, reliable capture of transient faults and long-term stable deployment of the visual diagnostic system are achieved simultaneously. Detailed Implementation
[0016] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0017] Example: The present invention provides a multi-specification control cabinet fault visual diagnosis system, including a panoramic partitioning module, a transient marking module, a region frame-up module, and a load reduction recovery module; The panoramic partitioning module is used to acquire panoramic images of the current control cabinet, simultaneously obtain the camera's acquisition frame rate, storage bandwidth margin, and processing temperature, and identify and divide the indicator light area, analog instrument area, and digital display screen area in the panoramic image. The transient marking module is used to continuously monitor the image sequence of each area. When any event such as the indicator light turning on and off, the instrument pointer shaking, or the screen characters flashing is captured, it is marked as a suspected transient event, and the duration of a single suspected transient event and the frequency of occurrence per unit time are recorded. The regional frame-up module is used to increase the acquisition frame rate in the region where the suspected transient event occurs when the duration of a single event is less than or equal to the single frame time interval of the current acquisition frame rate, while keeping the acquisition frame rate of the remaining regions unchanged. The load reduction recovery module is used to continuously acquire real-time values of processing temperature and storage bandwidth margin during the execution of regional frame rate increase. When the processing temperature reaches the preset upper temperature threshold or the storage bandwidth margin is lower than the preset lower bandwidth threshold, the acquisition frame rate of the corresponding region is reduced sequentially from low to high according to the frequency of suspected transient events in each region until the processing temperature falls back below the upper temperature threshold and the storage bandwidth margin rises back above the lower bandwidth threshold, and then the system recovers.
[0018] In this embodiment of the invention, a panoramic image of the current control cabinet is acquired, and the camera's frame rate, storage bandwidth margin, and processing temperature are simultaneously obtained. The panoramic image is then used to identify and delineate indicator light areas, analog instrument areas, and digital display screen areas. Specifically, this includes: The camera is triggered to capture a complete image of the control cabinet panel, and the resulting image is used as a panoramic image. Simultaneously, the camera's current capture frame rate is read. (Unit: frames / second), and synchronously read the storage bandwidth usage ratio of the edge processing unit. and the junction temperature of the processing chip Total storage bandwidth Subtract used bandwidth Obtain storage bandwidth margin , will the junction temperature As the processing temperature; Search for the nameplate area on the control cabinet panel in the panoramic image, use optical character recognition to extract the cabinet type code in the nameplate, and retrieve the area division template corresponding to the code from the pre-built benchmark template library based on the cabinet type code. The construction process of the benchmark template library is as follows: For each type of control cabinet, under standard lighting conditions and normal power-on operation, a high-resolution image of the front of its panel is taken as the template base image, and the pixel width of the template base image is recorded. and pixel height Select all components with fault indication functions on the panel in the template base map, and label each component according to its functional attributes as indicator light, analog instrument, or digital display screen. Record the pixel coordinates of the top left corner of the bounding box of each component in the template base map. With bottom right pixel coordinates The pixel coordinates of the bounding box are normalized by dividing the x-coordinate of the top-left corner by 1 / 2. , y-axis divided by The normalized top-left corner coordinates are obtained. Divide the bottom right x-coordinate by , y-axis divided by The normalized lower right corner coordinates are obtained. Thus, the bounding box of each component is mapped to The normalized rectangular region in the plane is decoupled from the actual pixel size of the template base image; at the same time, the reference image features of the normal state of each component are extracted, including the on and off states of indicator lights, the pointer scale range of analog instruments, and the constantly displayed characters of digital displays; the normalized bounding box coordinates and reference image features are stored according to the cabinet type code index to form the region division template of the control cabinet of that specification, and the region division templates of all control cabinets of all specifications together constitute the reference template library; The normalized bounding box coordinates recorded in the region segmentation template are then transformed using a homography transformation matrix between the panoramic image and the template base image. Mapped to pixel coordinates in the panoramic image; Among them, homography transformation matrix The solution method is as follows: The region division template pre-sets four positioning markers. These markers are four rectangularly distributed corner points or the center points of panel fixing screw holes selected in the template base map. The normalized coordinates of the four positioning markers in the template base map are recorded, in the following order: , , , The x-coordinate of each normalized coordinate is the pixel x-coordinate of that point in the template base image divided by 1 / 2. The ordinate value is the pixel ordinate of the point in the template base map divided by . ,Right now , , ; In the panoramic image, detect the actual positions corresponding to the four positioning markers and obtain their corresponding pixel coordinates, in the following order: , , , ; Substituting the normalized coordinates and pixel coordinates of the four pairs of matching points into the planar homography transformation relationship, for each pair of matching points... and Their mapping relationship satisfies: , ,in, This indicates the proportional relationship in homogeneous coordinates. It is a 3×3 homography transformation matrix. to Given the 9 elements of the matrix, take... With fixed dimensions; Expanding the above homogeneous coordinate relationship and eliminating the scaling factor, each pair of matching points provides two coordinates related to... Linear equations of matrix elements: ; Rearranged into linear form: ; The four pairs of matching points provide eight linear equations of the above form. These eight equations can be combined to form a non-homogeneous system of linear equations. ,in The coefficient matrix is 8×8, and its construction method is as follows: For the For each matching point, its normalized coordinates in the template base map are taken. and pixel coordinates in panoramic images In the matrix The Line entry In the matrix The Line entry ; The 8-dimensional column vector to be solved is: , It is an 8-dimensional column vector, and its ______ is ______. The elements are , No. The elements are ; Solving the non-homogeneous linear equation system using the least squares method yields the following results: ,like If singular, then use singular value decomposition to find its pseudoinverse; the obtained... The 8 elements are arranged in order with Combine and refill to form a 3×3 matrix, thus obtaining the homography transformation matrix. ; Based on the mapped pixel coordinates, regions of interest are segmented from the panoramic image. Specifically, for each normalized bounding box in the template, the four corner points are used... The coordinates are mapped to pixel coordinates in the panoramic image. The smallest bounding rectangle of the mapped quadrilateral is taken as the pixel-level boundary of the region of interest. Based on the classification labels recorded in the region division template, each region of interest is assigned to the indicator light region, analog instrument region, and digital display screen region, respectively.
[0019] In this embodiment of the invention, the image sequences of each region are continuously monitored. When any event such as indicator light switching, meter pointer jitter, or screen character flashing is captured, it is marked as a suspected transient event, and the duration of a single suspected transient event and its frequency of occurrence per unit time are recorded. Specifically, this includes: Based on the frame rate of panoramic image acquisition Images are acquired frame by frame from the video stream. For each frame, the current frame image blocks of each region are extracted based on the segmentation results of the indicator light area, analog instrument area, and digital display area to form an image sequence for each region. For the indicator light area, the normal state baseline for that area is retrieved from the area segmentation template. The normal state baseline includes the color range of the indicator light in the lit state and the average brightness value in the off state. The average brightness of the indicator light pixels in the current frame image block is compared with the average brightness value in the off state. If the current average brightness value exceeds a preset multiple of the average off brightness value for the first time, it is determined as a lighting event, and the current frame number is recorded as the lighting start frame. If the current average brightness value falls back to within the preset multiple and is already in the lit state, it is determined as an off event, and the current frame number is recorded as the off frame. The difference between the frame number of the off frame and the lighting start frame is divided by the acquisition frame rate. The duration of this single brightness-to-duration transition is obtained, in seconds; the duration of the most recent observation window is calculated. The number of internal bright-to-white transitions, divided by This yields the frequency of occurrence per unit time, expressed as occurrences per second. The preset observation window duration is a fixed value between 5 and 30 seconds. For the analog instrument area, the pointer's scale range under normal conditions is retrieved from the area segmentation template. The pointer centerline is extracted using Hough line detection for the current frame image block, and the angle between the pointer centerline and the zero-scale baseline of the dial is calculated and mapped to the current pointer reading. When the change in pointer reading between two adjacent frames exceeds a preset jitter threshold, and the direction of change reverses at least once within three consecutive frames, it is determined as a pointer jitter event, and the starting frame number of the jitter event is recorded. When the change in pointer reading in several consecutive frames is less than the jitter threshold, the jitter event is determined to have ended, and the ending frame number is recorded. The difference between the ending frame number and the starting frame number is divided by the acquisition frame rate. This yields the duration of a single pointer jitter; the most recent observation window is then analyzed. The number of internal pointer bouncing events, divided by This gives the frequency of occurrence per unit time. For the digital display screen area, retrieve the normally displayed character content of the display screen in normal state from the area division template; extract the currently displayed character sequence using optical character recognition for the current frame image block, and compare the current character sequence with the character sequence of the previous frame character by character. When the character sequence changes and the changed character sequence recovers to be consistent with the previous character sequence within no more than two consecutive frames, it is determined as a character flickering event, and the starting frame number of the flickering event is recorded; when the character sequence remains stable for three consecutive frames after recovery, the flickering event is determined to have ended, and the ending frame number is recorded; divide the difference between the ending frame number and the starting frame number by the acquisition frame rate. This yields the duration of a single character blink; it also calculates the duration of the most recent observation window. The number of times the internal character blinking event occurred, divided by This gives the frequency of occurrence per unit time. Mark any of the three types of events mentioned above as a suspected transient event, and record its event type, single duration, and frequency of occurrence per unit time.
[0020] In this embodiment of the invention, when the duration of a single suspected transient event is less than or equal to the single frame time interval of the current acquisition frame rate, the acquisition frame rate is increased in the region where the suspected transient event occurred, while the acquisition frame rate of the remaining regions remains unchanged. Specifically, this includes: Read the single duration of a suspected transient event Get the frame rate of the current full-screen capture by the camera. Calculate the current single frame time interval ;like This indicates that the duration of the transient event is shorter than or equal to one frame sampling period, and the event process cannot be reliably captured at the current frame rate. In this case, a frame rate boost is triggered for the target area where the suspected transient event occurred. From the region segmentation results of the panoramic image, the pixel coordinate range covered by the target region is extracted and represented as the x-coordinate of the starting pixel of the windowing. Starting pixel ordinate Window width and window height The four parameters are obtained by mapping the bounding box of the target region in the panoramic image to the camera sensor pixel coordinate system through the homography transformation matrix and then rounding them. Send a region windowing acquisition command to the camera, the command including the windowing coordinate range. and target frame rate Target frame rate Determine as follows: First, calculate the minimum frame rate requirement for event-driven calculations. This ensures that at least two valid samples are obtained during the duration of the event. Secondly, obtain the camera's window size. The highest supported frame rate , The method to obtain this information is: query the readout time of the sensor model row in the camera's technical specifications table. The highest frame rate is from Confirmed, among which The window opening height, This represents the readout time per row of pixels for the sensor, in seconds. If the camera's windowed frame rate is provided by the manufacturer in the form of a lookup table, then the window height is directly used as the reference in the lookup table. The corresponding highest frame rate is used as ; Finally, take With the current frame rate The larger one, then with Compare the two and take the smaller one as the target frame rate: ; After receiving the window capture command, the camera maintains the original full-screen capture channel at the same frame rate. Continuous output, covering the entire indicator light area, analog instrument area, and digital display area; simultaneously, a new independent windowed acquisition channel is opened, which only reads pixel data within the windowed coordinate range, at a frame rate. Output a local high frame rate image sequence; each of the two image streams carries a unified timestamp for subsequent synchronization and comparison.
[0021] In this embodiment of the invention, during the execution of the regional frame rate increase, real-time values of processing temperature and storage bandwidth margin are continuously acquired. When the processing temperature reaches a preset upper temperature threshold, or the storage bandwidth margin falls below a preset lower bandwidth threshold, the acquisition frame rate of the corresponding region is sequentially reduced from low to high according to the frequency of suspected transient events occurring in each region, until the processing temperature falls back below the upper temperature threshold and the storage bandwidth margin rises back above the lower bandwidth threshold, after which the process is restored. Specifically, this includes: At frame rate in the target area During the high frame rate acquisition with windowing, a fixed sampling period is used. Continuously read the processing temperature of the edge processing unit and storage bandwidth margin Real-time value, fixed sampling period The value is 1 second; Preset upper temperature threshold and bandwidth lower limit threshold ,in Take 80% of the rated maximum junction temperature of the processing chip. Take total storage bandwidth 10%; When satisfied or If any of the following conditions are met, the system resources are determined to be in a state of strain, triggering a regional load reduction process. The regional unloading process is as follows: The first step is to record the current frame rate of each region at the start of load reduction, which will serve as their respective baseline frame rate: for regions currently operating at a high frame rate... The target area for windowed data acquisition has a base frame rate denoted as: For full-screen frame rate The baseline frame rate for the remaining areas captured is denoted as: ; The second step is to extract the data for each region from the monitoring records of each region within the most recent observation window. The frequency of suspected transient events occurring within the region is recorded. For regions where no suspected transient events have occurred, the frequency is recorded as 0. All regions are sorted from low to high frequency to form a load reduction order list. The third step involves sorting the load reduction order list and, starting with the region with the lowest frequency of occurrence, sequentially performing frame rate reduction operations on each region: reducing the current actual acquisition frame rate of that region to half of its baseline frame rate. If the reduced frame rate is lower than the panoramic image acquisition frame rate... Then it will directly drop to No further reduction; after each reduction operation, the current actual collection frame rate of the area is updated to the reduced value, which serves as the starting point for the next possible reduction of the load in the area; The fourth step is to wait for one sampling period after performing a frame rate reduction operation on each region. Reread and The real-time value is used to determine whether the condition is still met. or If either condition is still met, the frame rate reduction operation continues for the next region in the load reduction order list; if neither condition is met, i.e. and If the frame rate is reduced, the downloading will stop, and the regions that have already reduced the frame rate will continue to collect data while maintaining the current frame rate. When the processing temperature Falling back to Below, and storage bandwidth margin rebounded to After the above, it is determined that the system resources are no longer in a state of strain. Following the reverse order of the deload order list, starting from the last deloaded region, the current actual sampling frame rate of each deloaded region is restored to its respective baseline frame rate. After restoring each region, wait for one sampling cycle. ,confirm and While remaining within a safe range, restore the next region until all downloaded regions have been restored to their respective baseline frame rates.
[0022] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.
[0023] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions according to the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. Computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless (e.g., infrared, wireless, microwave, etc.) means. A computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. Available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media. Semiconductor media can be solid-state drives.
[0024] It should be understood that in the various embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.
[0025] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0026] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.
Claims
1. A multi-specification control cabinet fault visual diagnosis system, characterized in that: Includes a panoramic partitioning module, a transient marking module, a regional frame-up module, and a load-down recovery module; The panoramic partitioning module is used to acquire panoramic images of the current control cabinet, simultaneously obtain the camera's acquisition frame rate, storage bandwidth margin, and processing temperature, and identify and divide the indicator light area, analog instrument area, and digital display screen area in the panoramic image. The transient marking module is used to continuously monitor the image sequence of each area. When any event such as the indicator light turning on and off, the instrument pointer shaking, or the screen characters flashing is captured, it is marked as a suspected transient event, and the duration of a single suspected transient event and the frequency of occurrence per unit time are recorded. The regional frame-up module is used to increase the acquisition frame rate in the region where the suspected transient event occurs when the duration of a single event is less than or equal to the single frame time interval of the current acquisition frame rate, while keeping the acquisition frame rate of the remaining regions unchanged. The load reduction recovery module is used to continuously acquire real-time values of processing temperature and storage bandwidth margin during the execution of regional frame rate increase. When the processing temperature reaches the preset upper temperature threshold or the storage bandwidth margin is lower than the preset lower bandwidth threshold, the acquisition frame rate of the corresponding region is reduced sequentially from low to high according to the frequency of occurrence of suspected transient events in each region, until the processing temperature falls back below the upper temperature threshold and the storage bandwidth margin rises back above the lower bandwidth threshold, and then the process is restored.
2. The multi-specification control cabinet fault visual diagnosis system according to claim 1, characterized in that: The panoramic partitioning module specifically includes: The camera is triggered to capture the entire control cabinet panel, and the resulting image is used as a panoramic image. At the same time, the current acquisition frame rate of the camera is read, and the storage bandwidth usage ratio of the edge processing unit and the junction temperature of the processing chip are read simultaneously. The storage bandwidth margin is obtained by subtracting the used bandwidth from the total storage bandwidth, and the junction temperature is used as the processing temperature. Search for the nameplate area on the control cabinet panel in the panoramic image, use optical character recognition to extract the cabinet type code in the nameplate, and retrieve the area division template corresponding to the code from the pre-built benchmark template library based on the cabinet type code; The normalized bounding box coordinates recorded in the region division template are mapped to pixel coordinates in the panoramic image through the homography transformation matrix between the panoramic image and the template base image. Based on the mapped pixel coordinates, each region of interest is segmented from the panoramic image, and according to the classification labels recorded in the region division template, each region of interest is assigned to the indicator light region, analog instrument region, and digital display screen region, respectively.
3. The multi-specification control cabinet fault visual diagnosis system according to claim 1, characterized in that: The transient marking module specifically includes: For the indicator light area, retrieve the normal state reference of the area from the area division template. The normal state reference includes the color range of the indicator light when it is lit and the average brightness value when it is off. Compare the average brightness of the indicator light pixels in the current frame image block with the average brightness in the off state. If the current average brightness exceeds the preset multiple of the off brightness for the first time, it is determined to be a lighting event, and the current frame number is recorded as the lighting start frame. If the current average brightness value drops back to within a preset multiple and is already in the lit state, it is determined to be an off event, and the current frame number is recorded as the off frame. Divide the difference between the frame number of the off frame and the frame number of the on start frame by the sampling frame rate to obtain the duration of a single on / off transition. The frequency of occurrence per unit time is obtained by counting the number of times the screen turns on and off within the most recent observation window and dividing by the duration of the observation window.
4. The multi-specification control cabinet fault visual diagnosis system according to claim 1, characterized in that: The transient marking module specifically includes: For the analog instrument area, retrieve the pointer scale range of the instrument under normal conditions from the area division template; Extract the pointer centerline from the current frame image block, calculate the angle between the pointer centerline and the zero scale baseline of the dial, and map the angle to the current pointer reading; When the change in pointer reading between two adjacent frames exceeds the preset jitter judgment threshold, and the direction of change reverses at least once within three consecutive frames, it is judged as a pointer jitter event, and the starting frame number of the jitter event is recorded. When the change in pointer readings for several consecutive frames is less than the jitter detection threshold, the jitter event is determined to be over, and the end frame number is recorded. Divide the difference between the end frame number and the start frame number by the acquisition frame rate to obtain the duration of a single pointer jitter. The frequency of pointer jitter events per unit time is obtained by counting the number of times the event occurs within the most recent observation window and dividing by the duration of the observation window.
5. The multi-specification control cabinet fault visual diagnosis system according to claim 1, characterized in that: The transient marking module specifically includes: For the digital display screen area, retrieve the normally displayed character content of the display screen in the normal state from the area division template; The character sequence currently displayed is extracted from the image block of the current frame. The current character sequence is compared with the character sequence of the previous frame character by character. When the character sequence changes and the changed character sequence recovers to be consistent with the original character sequence within no more than two consecutive frames, it is determined as a character flashing event, and the starting frame number of the flashing event is recorded. When the character sequence remains stable for three consecutive frames after recovery, the blinking event is considered to have ended, and the end frame number is recorded. Divide the difference between the end frame number and the start frame number by the acquisition frame rate to obtain the duration of a single character flash. Count the number of times the character blinking event occurs within the most recent observation window, divide by the duration of the observation window, and obtain the frequency of occurrence per unit time.
6. The multi-specification control cabinet fault visual diagnosis system according to claim 1, characterized in that: The regional frame upscaling module specifically includes: Read the duration of a single suspected transient event, obtain the frame rate of the current full-frame capture by the camera, and calculate the current single-frame time interval; If the duration of a single event is less than or equal to the time interval of a single frame, a frame rate boost is triggered for the target area where a suspected transient event has occurred. From the region segmentation results of the panoramic image, the pixel coordinate range covered by the target region is extracted. The pixel coordinate range is obtained by mapping the bounding box of the target region in the panoramic image to the camera sensor pixel coordinate system through the homography transformation matrix and then rounding it.
7. The multi-specification control cabinet fault visual diagnosis system according to claim 6, characterized in that: The regional frame upscaling module specifically includes: A region windowing acquisition command is sent to the camera. The command includes the pixel coordinate range and the target frame rate. The camera maintains the original full-screen acquisition channel and continues to output at the current frame rate. At the same time, a new independent windowing acquisition channel is opened to output a local high frame rate image sequence at the target frame rate.
8. The multi-specification control cabinet fault visual diagnosis system according to claim 1, characterized in that: The load reduction recovery module specifically includes: During the windowed high frame rate acquisition in the target area, the real-time values of the processing temperature and storage bandwidth margin of the edge processing unit are continuously read at a fixed sampling period. When either the processing temperature is greater than or equal to the upper temperature threshold, or the storage bandwidth margin is less than or equal to the lower bandwidth threshold, the regional load reduction process is triggered. The regional load reduction process is as follows: record the current acquisition frame rate of each region at the start of load reduction as its own baseline frame rate, extract the occurrence frequency of suspected transient events in each region within the most recent observation window from the monitoring records of each region, sort all regions in order of occurrence frequency from low to high to form a load reduction order list, and reduce the current actual acquisition frame rate of each region to half of its baseline frame rate in order of the load reduction order list. If the reduced frame rate is lower than the acquisition frame rate of the panoramic image, it is directly reduced to the acquisition frame rate of the panoramic image. After performing a frame rate reduction operation on each region, wait for one fixed sampling period, reread the processing temperature and storage bandwidth margin, and if the triggering conditions are still met, continue to perform a frame rate reduction operation on the next region; otherwise, stop reducing the load.
9. The multi-specification control cabinet fault visual diagnosis system according to claim 8, characterized in that: In the load reduction recovery module, when the processing temperature drops below the upper temperature threshold and the storage bandwidth margin rises above the lower bandwidth threshold, the current actual acquisition frame rate of each load-reduced region is restored to its respective base frame rate in reverse order of the load reduction order list, starting from the last load-reduced region. After each region is restored, wait for a fixed sampling period to confirm that the processing temperature and storage bandwidth margin are still within a safe range before restoring the next region, until all deloaded regions have been restored to their respective baseline frame rates.
10. A multi-specification control cabinet fault visual diagnosis system according to claim 8, characterized in that: In the transient marking module, the frequency of occurrence of areas where no suspected transient events have occurred is recorded as 0.
