Individual soldier maintenance regulation and control method and system based on digitization

Abstract

The invention discloses an individual maintenance regulation and control method and system based on digitization, and relates to the technical field of individual maintenance, and the method comprises the following steps: obtaining a maintenance site image, and carrying out the segmentation and preprocessing of the image, and obtaining a first superpixel region set; obtaining first data of the first super-pixel region set, establishing an image compression evaluation model based on a convolutional neural network, and generating a picture adjustment value of each super-pixel region; respectively compressing each superpixel region of the maintenance site image based on the image adjustment value to obtain an adjustment image, and traversing each maintenance site image as an image frame to form a maintenance site video; and updating the maintenance site video according to the elimination rule, and uploading to the AR end to assist the substation in maintenance. The image compression strategy can be dynamically optimized, the real-time performance and accuracy of remote overhaul are improved, irrelevant data transmission is reduced, and efficient and reliable technical support is provided for intelligent maintenance of power grid equipment.

Description

Technical Field

[0001] The present invention relates to the technical field of individual maintenance, and more specifically, to a digital-based individual maintenance control method and system. Background Art

[0002] Smart substation maintenance usually includes on-site inspections, equipment maintenance, and troubleshooting. On-site maintenance personnel wear AR devices or use portable terminals to inspect and inspect substation equipment, use high-definition cameras to capture images and videos of key equipment in real time, and transmit them to remote experts via 5G or dedicated wireless networks. Remote experts receive on-site images through the AR terminal and use augmented reality technology to analyze the images, including equipment status assessment, fault point identification, and operation instructions. Based on real-time images, experts can overlay visual annotations, send maintenance plans, or call historical data for comparative analysis to assist on-site personnel in accurately performing maintenance tasks. The system can combine artificial intelligence algorithms to perform regional segmentation and point of interest extraction on images, improve data processing efficiency, and ensure accurate transmission of key component information. Finally, on-site personnel complete maintenance according to expert instructions, and the entire process achieves efficient and accurate remote collaboration, improving the level of intelligence in substation operation and maintenance.

[0003] For example, the invention patent with announcement number CN117193237A announces an adaptive diagnosis method and system based on hypergraph and Transformer, including: S1. Obtain fault data of equipment in the automation system and perform data preprocessing; S2. For data that does not have a graph network structure, use the preprocessed data obtained in step S1 to construct a graph, and select K nearest subsequences and the central subsequence to form a hypergraph; S3. Hypergraph data is input into hypergraph convolution and Transformer to complete the extraction of high-order, local and global information; S4. The final fusion feature is obtained based on the adaptive fusion mechanism; S5. The fusion feature is mapped to the corresponding dimensional space using a multi-layer perceptron, and finally optimized using a loss function. Experimental results in a variety of data sets and interference scenarios show that it has good generalization performance and strong robustness.

[0004] The above disclosed technical solutions have at least the following technical problems: the traditional fixed compression solution and fixed compression ratio often cannot meet the requirements of detail retention in key equipment areas and efficient transmission in background areas at the same time. Although a high compression ratio is conducive to reducing the amount of data and increasing the transmission rate, it may cause the loss of important details such as equipment edges, cracks and nameplates; on the contrary, a low compression ratio increases the burden of data transmission while ensuring image quality, which is not conducive to real-time monitoring.

[0005] In view of the above problems, the present invention proposes a solution. Summary of the invention

[0006] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a digital-based individual maintenance and control method, which uses image regional compression and control technology to solve the problem of difficulty in balancing efficiency and accuracy in processing and transmission of key equipment data of on-site image video transmission during intelligent substation maintenance.

[0007] To achieve the above object, the present invention provides the following technical solutions: The digitalized individual maintenance and control method includes the following steps: obtaining a maintenance site image, segmenting and preprocessing the image to obtain a first super-pixel region set; obtaining first data of the first super-pixel region set, and establishing an image compression evaluation model based on a convolutional neural network to generate an image adjustment value for each super-pixel region; compressing each super-pixel region of the maintenance site image separately based on the image adjustment value to obtain an adjusted image, and traversing each maintenance site image as a picture frame to form a maintenance site video; updating the maintenance site video according to the elimination rule, and uploading it to an AR terminal to assist in substation maintenance.

[0008] In a preferred embodiment, the image is segmented and preprocessed to obtain a first superpixel region set, specifically: the first superpixel region includes several superpixel regions; the maintenance site image is segmented based on the superpixel segmentation technology to obtain a second superpixel region set; each superpixel region of the second superpixel region set is converted to HSV color to obtain corresponding color features, and data analysis is performed based on the color features to mark dirty areas; based on the dirty areas and the Otsu threshold segmentation method, the dirty areas in the second superpixel region set are eliminated, and the first superpixel region set is updated.

[0009] In a preferred embodiment, the first data includes texture complexity data, and the specific acquisition method is as follows: obtain the pixel points of each superpixel area, process the pixel points through LBP, obtain the variance of the LBP value of the pixel points, and use the variance of the LBP value in each superpixel area as the texture complexity data.

[0010] In a preferred embodiment, the first data includes target area proportion data, and the specific acquisition method is as follows: obtain key device pixel data, determine the pixel ratio of key device pixels in each superpixel area based on pixel statistics, and obtain a first ratio; compare a preset pixel ratio threshold with the first ratio to determine whether each superpixel area is a target area, and use the ratio of the target area in the superpixel area as the target area proportion data.

[0011] In a preferred embodiment, the first data includes equipment failure rate weight compression rate data, and the specific acquisition method is as follows: based on the target detection network, the equipment type of each superpixel area in the maintenance site image is identified and the equipment information is obtained in combination with the equipment history database, and the failure rate weight of each device is calculated and obtained; based on the equipment failure rate weight and the preset image compression rate, a equipment failure rate weight compression data proportional model is constructed, and the equipment failure rate weight compression rate data is output.

[0012] In a preferred embodiment, the image compression evaluation model is established based on a convolutional neural network to generate a picture adjustment value for each superpixel area, specifically: according to the first data, a multidimensional image compression evaluation model is established based on a convolutional neural network, and the picture adjustment value of each superpixel area is obtained based on the multidimensional image compression evaluation model.

[0013] In a preferred embodiment, each superpixel area of ​​the maintenance site image is compressed separately based on the picture adjustment value to obtain an adjusted picture, and each maintenance site image is traversed as a picture frame to form a maintenance site video, specifically: a picture standard value of the maintenance site image is preset, and compared with the picture adjustment value of each superpixel area of ​​the maintenance site image to obtain an adjustment ratio of each superpixel area; based on the adjustment ratio, each superpixel area of ​​the maintenance site image is proportionally adjusted accordingly to obtain an adjusted picture; and each maintenance site image is traversed and adjusted as a picture frame to form a maintenance site video.

[0014] In a preferred embodiment, the elimination rule includes clarity elimination, specifically: Obtain the maintenance site video frame image and calculate the Laplacian variance of the frame image according to the Laplacian transformation as the clarity evaluation value, remove the frame image corresponding to the clarity evaluation value less than the preset first threshold, and update the maintenance site video.

[0015] In a preferred embodiment, the elimination rule includes adaptive time interval elimination, specifically: obtaining the actual time interval of the updated maintenance site video frame image as the adaptive time interval; calculating the clarity change of adjacent frame images through Laplacian transformation as the weight of the adaptive time interval, and adjusting and updating the adaptive time interval; when the actual time interval is less than the adaptive time interval, calculating the clarity change between the frame image and the previous frame image; if the clarity change is less than the set second threshold, eliminating the frame image corresponding to the actual time interval and updating the maintenance site video.

[0016] The digitalized individual maintenance and control system includes an image segmentation processing module, a data acquisition module, a picture compression evaluation module and a picture compression adjustment module; the image segmentation processing module is used to acquire maintenance site images, and perform segmentation and preprocessing to obtain a super-pixel area set; the data acquisition module is used to acquire first data of the super-pixel area set, wherein the first data includes texture complexity data, target area proportion data, task requirement weight and equipment failure rate weight compression rate data; the picture compression evaluation module is used to establish an image compression evaluation model based on a convolutional neural network according to the first data, and generate a picture adjustment value for each super-pixel area; the picture compression adjustment module is used to compress each super-pixel area of ​​the maintenance site image separately based on the picture adjustment value, obtain adjusted picture data, and upload it to an AR terminal to assist in substation maintenance.

[0017] The technical effects and advantages of the digitalized individual maintenance control method and system of the present invention are as follows: 1. The present invention obtains maintenance site image data and performs preprocessing based on superpixel segmentation, color feature analysis and Otsu threshold segmentation to obtain a superpixel region set, thereby achieving accurate division of key equipment areas and background or dirty areas in the image, thereby ensuring that key detail information is effectively retained and providing an accurate basis for the subsequent extraction of texture complexity, target area proportion data and failure rate weight data.

[0018] 2. The present invention obtains the texture complexity data, target area proportion data and fault rate weight data of the superpixel area set, and combines the task requirement weight to establish an image compression evaluation model, so that the image compression ratio can be dynamically adjusted under different maintenance tasks (such as inspection, fault diagnosis or OCR recognition) and on-site environments, which not only reduces the data transmission volume and improves the transmission efficiency, but also ensures the high-quality presentation of key areas, thereby effectively improving the real-time and responsiveness of the remote individual maintenance system. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 The figure is a flow chart of the digitalized individual maintenance and control method of the present invention.

[0020] Figure 2 It is a structural schematic diagram of the digitalized individual maintenance and control system of the present invention. DETAILED DESCRIPTION

[0021] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0022] Embodiment 1, Figure 1 The present invention provides a digitalized individual maintenance control method, which includes the following steps: S1, acquiring a maintenance site image, and segmenting and preprocessing the image to obtain a first superpixel region set; The first superpixel region includes a plurality of superpixel regions; Segment the maintenance site image based on the superpixel segmentation technology to obtain a second superpixel region set; Performing HSV color conversion on each superpixel region of the second superpixel region set to obtain corresponding color features, and performing data analysis based on the color features to mark the dirty areas; Based on the dirty area and Otsu threshold segmentation method, the dirty area in the second superpixel area set is eliminated, and the first superpixel area set is updated.

[0023] It should be noted that data analysis based on color features includes average hue, average saturation, and average brightness; It should be noted that in the above steps, the following is an overview of the calculation: The specific implementation formula of superpixel segmentation technology is:

[0024] The specific implementation formula of Otsu threshold segmentation method is:

[0025] In the formula, is the input image, is the number of superpixels, controlling the size of the segmented region, is the superpixel segmentation function, is the set of superpixel regions after segmentation, is the preset segmentation threshold, is the inter-class variance between the device area and the background, is the global variance.

[0026] The specific implementation formula of the average hue calculation formula is:

[0027] The specific implementation formula for the average saturation calculation formula is:

[0028] Average brightness calculation formula:

[0029] In the formula, is the spatial coordinate of the pixel in the superpixel area, Super pixel area The total number of pixels in Super pixel area Average tone, Super pixel area Average saturation, Super pixel area Average brightness.

[0030] S2, obtaining first data of a first superpixel region set, and establishing an image compression evaluation model based on a convolutional neural network to generate an image adjustment value for each superpixel region.

[0031] The first data includes texture complex data, and the specific acquisition method thereof is as follows: Get the pixel points of each superpixel area, process the pixel points through LBP, obtain the variance of the LBP value of the pixel point, and use the variance of the LBP value in each superpixel area as the texture complexity data.

[0032] It should be noted that in the above steps, the following is an overview of the calculation: The specific calculation formula for LBP to process pixels is:

[0033] The specific calculation formula for the variance of the LBP value is:

[0034] The specific calculation formula for the mean LBP value is:

[0035] In the formula, is the gray value of the center pixel, For centered Neighborhood pixel values, is the symbolic function, is the number of neighborhood pixels selected, is the neighborhood pixel index currently calculated, is the LBP value of the pixel, is the variance of the LBP value, is the total number of pixels in the device area, is the mean of all LBP values ​​in the device area.

[0036] It should be noted that the larger the variance of the LBP value, the greater the texture complexity, and the lower the compression is; It should be noted that ; Furthermore, analyzing texture complexity has the following advantages for evaluating image compression requirements: Identify important information areas and ensure key details; high texture complexity areas, such as cracks, wire connection points, and insulator contamination, usually contain important fault information and require a lower compression rate to maintain clarity. Low texture complexity areas, such as the uniform metal surface of the transformer, have less information and can use a higher compression rate to reduce the amount of data.

[0037] Adaptive compression improves transmission efficiency; texture complexity can be used as a dynamic adjustment factor for compression rate, ensuring that important areas have high complexity and low compression to ensure clarity; irrelevant areas have low complexity and high compression to reduce data volume.

[0038] Adapt to different environments and optimize the remote maintenance experience; the surface of the equipment may be affected by factors such as light, stains, and aging, resulting in changes in local area information; texture complexity analysis can dynamically identify these influences and adjust the compression strategy to enable remote maintenance personnel to obtain the best image quality.

[0039] The first data includes target area proportion data, and the specific acquisition method is as follows: Acquire pixel data of key devices, and determine the pixel ratio of key device pixels in each superpixel area based on a pixel statistics method to obtain a first ratio; The preset pixel ratio threshold is compared with the first ratio to determine whether each superpixel area is a target area, and the ratio of the target area to the superpixel area is used as the target area ratio data.

[0040] It should be noted that the preset pixel ratio threshold is an empirical threshold set by the system through statistics of pixel ratios of key equipment in a large number of historical maintenance images and combined with the professional knowledge of substation maintenance experts. It is used to determine whether the super-pixel area belongs to the target area, thereby guiding the selection of compression ratio in the image compression process, so as to improve data transmission efficiency while ensuring the image clarity of the target area; It should be noted that in the above steps, the following is an overview of the calculation: The specific calculation formula for the pixel ratio of key device pixels in each superpixel area is:

[0041] The specific calculation formula for the target area ratio data is:

[0042] In the formula, is a superpixel region set, for A superpixel region in a superpixel region set, Super pixel area The pixel ratio of key devices in the system, Super pixel area The number of pixels belonging to the key device, Super pixel area The total number of pixels in is the proportion of superpixel areas belonging to key devices in the set of all superpixel areas, is the judgment function, is the total number of pixels in the entire image.

[0043] It should be noted that the judgment function , It is the device judgment threshold, which can be adjusted according to the situation; It should be noted that when the superpixel region is the target region, low compression is preferred, and when the superpixel region is the non-target region, high compression is preferred; Furthermore, analyzing the target area ratio data has the following advantages for evaluating image compression requirements: Adaptive compression ratio to ensure clarity in key areas; use low compression when the target area accounts for a large proportion to retain device details; use high compression when the target area accounts for a small proportion to reduce data volume and improve transmission efficiency.

[0044] Compression strategies are adjusted intelligently based on task requirements. Quick browsing in inspection mode allows for appropriately high compression. When diagnosing faults, attention is paid to details and compression is reduced when there are many target areas. For OCR recognition, such as identifying equipment nameplates, low compression ensures text clarity.

[0045] The computational complexity is low and suitable for real-time applications. The calculation of the target area ratio is based on superpixel statistics, which is simple and suitable for AR devices and edge computing.

[0046] The specific method for obtaining the first data including the task requirement weight is as follows: The individual maintenance system includes several working modes, each of which sets a default task requirement weight; It should be noted that the working modes include inspection mode, fault diagnosis and OCR recognition; By default, the task requirement weight of the inspection mode is set to 0.5, the task requirement weight of the fault diagnosis device is set to 1, and the task requirement weight of OCR recognition is set to 1.5, and the above task requirement weights can be adjusted according to actual needs.

[0047] The first data includes equipment failure rate weight compression rate data, and the specific acquisition method is as follows: Based on the target detection network, the equipment type of each superpixel area in the maintenance site image is identified and combined with the equipment history database to obtain equipment information, and the failure rate weight of each equipment is calculated; A device failure rate weight compression data ratio model is constructed based on the device failure rate weight and a preset image compression rate, and the device failure rate weight compression rate data is output.

[0048] The equipment history database includes the failure frequency, failure severity and most recent failure time of each equipment; The device information includes standard fault frequency data, standard fault severity data and standard time decay data.

[0049] It should be noted that the target detection network is an existing technology; It should be noted that the standard fault severity data is obtained by ranking the economic losses caused by each faulty device; It should be noted that in the above steps, the following is an overview of the calculation: The specific calculation formula for the target detection network to identify the maintenance site image is:

[0050] The specific calculation formula of standard fault frequency is:

[0051] The specific calculation formula for standard fault severity is:

[0052] The specific calculation formula of standard time decay is:

[0053] The specific calculation formula for the failure rate weight of each device is:

[0054] The specific calculation formula for the equipment failure rate weight compression rate data is:

[0055] In the formula, is the device area, To segment the network, is the input maintenance site image, For equipment The candidate box, For equipment The standard failure frequency, For equipment The failure frequency, is the sum of the number of equipment failures, For equipment The standard fault severity of For equipment The severity of the fault, is the sum of the severity of equipment failures, is the standard time decay, is the fault time interval, is the time constant that controls the decay rate, is the current time, For equipment The time when the most recent failure occurred, is an exponential function, is the equipment failure rate weight, For equipment The standard fault frequency weight coefficient, For equipment The standard fault severity weight factor of For equipment The standard time decay weight coefficient of is the equipment failure rate weight compression rate data, The preset minimum compression rate is 10%. The preset maximum compression rate is 90%.

[0056] Furthermore, analyzing the device failure rate weighted compression rate data has the following advantages for evaluating image compression requirements: Quantify the risk by counting the number of historical failures of each device, normalizing it to obtain the risk weight, and clearly distinguish between high-risk and low-risk devices; Key details are retained, and a lower compression ratio is used for areas with high failure rates to ensure that equipment details and fault characteristics are clear, which is helpful for remote diagnosis; Data transmission is optimized, with a higher compression ratio used in low-risk areas to effectively reduce data volume and increase transmission speed; Dynamically adapt and adjust compression parameters based on real-time data to achieve the best balance between image quality and resource utilization.

[0057] The image compression evaluation model is established based on the convolutional neural network to generate the image adjustment value of each superpixel area, specifically: According to the first data, a multidimensional image compression evaluation model is established based on a convolutional neural network, and a picture adjustment value of each superpixel area is obtained based on the multidimensional image compression evaluation model.

[0058] It should be noted that in the above steps, the following is an overview of the calculation: The specific calculation formula for the picture adjustment value is:

[0059] In the formula, Adjust the values ​​for the image. is the target area percentage data, is the texture complexity data, is the task requirement weight, is the equipment failure rate weight compression rate data, is the adjustable weight coefficient of the target area proportion data, is the adjustable weight coefficient of the texture complexity data, is the adjustable weight coefficient of the task requirement weight, It is the adjustable weight coefficient of the failure rate weight data.

[0060] Furthermore, task requirement weights have the following advantages for evaluating image compression requirements: Flexible matching of different task objectives. In power grid maintenance, different tasks such as inspection, fault diagnosis, and OCR recognition have different requirements for image clarity. By setting task requirement weights, the system can automatically select the appropriate compression strategy based on the current task, reducing manual intervention. Effectively balance image quality and transmission efficiency. When the task demand weight is high, the system will reduce the compression ratio to retain more details; when the task demand weight is low, a higher compression ratio is allowed to reduce the data volume and increase the transmission speed; Improve resource utilization efficiency; when bandwidth or storage resources are limited, prioritize tasks based on their weights to ensure that critical tasks (high weight) receive higher quality images, while general tasks use higher compression to avoid wasting resources; Dynamically adapt to the on-site environment and workflow. The task type may be switched at any time at the power grid maintenance site, from inspection to fault diagnosis. By adjusting the task demand weight, the system can quickly respond and adjust the image compression strategy to ensure the continuity and efficiency of the maintenance work. It has strong scalability and is easy to combine with other features. The task requirement weight can be combined with factors such as texture complexity and target area ratio to form a comprehensive evaluation model to achieve a more accurate and intelligent image compression solution.

[0061] S3, compressing each superpixel region of the maintenance site image separately based on the image adjustment value to obtain an adjusted image, and traversing each maintenance site image as a picture frame to form a maintenance site video; The method compresses each superpixel region of the maintenance site image based on the image adjustment value to obtain an adjusted image, and traverses each maintenance site image as a picture frame to form a maintenance site video, specifically: Preset the image standard value of the maintenance site image, and compare it with the image adjustment value of each super-pixel area of ​​the maintenance site image to obtain the adjustment ratio of each super-pixel area; Based on the adjustment ratio, each superpixel area of ​​the maintenance site image is respectively adjusted in a corresponding ratio to obtain an adjusted image; Each maintenance site image is traversed and adjusted as a picture frame to form a maintenance site video.

[0062] S4, updates the maintenance site video according to the elimination rules and uploads it to the AR end to assist substation maintenance.

[0063] The elimination rules include clarity elimination, specifically: Obtain the maintenance site video frame image and calculate the Laplacian variance of the frame image according to the Laplacian transformation as the clarity evaluation value, remove the frame image corresponding to the clarity evaluation value less than the preset first threshold, and update the maintenance site video.

[0064] It should be noted that the first threshold is determined by testing a large number of maintenance site video samples, statistically analyzing the distribution of Laplacian variance, and determining the appropriate threshold through subjective observation; It should be noted that when the clarity evaluation value is less than the preset first threshold, it means that the clarity of the frame image is very low and the availability is low; when the clarity evaluation value is greater than or equal to the preset first threshold, it means that the clarity of the frame image is relatively clear and the availability is high, and it is the image data required by the maintenance AR end; It should be noted that in the above steps, the following is an overview of the calculation: The specific calculation formula of Laplacian transformation is:

[0065] The specific calculation formula of the clarity evaluation value is:

[0066] In the formula, is the pixel gradient value, that is, the Laplacian value, is the pixel intensity value of the frame image, is the spatial coordinate of the frame image, is the second-order derivative of the frame image in the x direction, is the second-order derivative of the frame image in the y direction, Frame pictures The clarity evaluation value of is the width of the frame image, is the height of the frame image, is the mean of the Laplacian values ​​of all pixels in the image; It should be noted that and They are the second-order derivatives of the frame image in the x-direction and the y-direction, respectively, indicating the rate at which the pixel intensity changes with space, and are used to detect the edge and texture changes of the image; It should be noted that the mean of the Laplacian value is calculated by the Laplacian values ​​of all pixels in the image, that is, ; It should be noted that the clarity elimination constraints are: ,in, is the first threshold, Frame pictures The clarity evaluation value.

[0067] The elimination rule includes adaptive time interval elimination, specifically: The actual time interval of the video frame images of the maintenance site after the update is obtained as the adaptive time interval; the clarity change of adjacent frame images is calculated by Laplacian transformation as the weight of the adaptive time interval, and the adaptive time interval is adjusted and updated, and the clarity change is the difference between the clarity evaluation values ​​of adjacent frame images; When the actual time interval is less than the adaptive time interval, the definition change between the frame image and the previous frame image is calculated; If the change in clarity is less than the set second threshold, the frame image corresponding to the actual time interval is eliminated and the maintenance site video is updated.

[0068] It should be noted that the clarity change is the difference between the clarity evaluation values ​​of adjacent frame images; It should be noted that when the definition changes between adjacent frames are large, the time interval is reduced and the sampling rate is increased; when the definition changes are small, the time interval is increased and the sampling rate is reduced; It should be noted that in the above steps, the following is an overview of the calculation: The specific calculation formula for the actual time interval is:

[0069] The specific calculation formula for adjusting the update adaptive time interval is:

[0070] In the formula, is the time interval between adjacent frames, that is, the actual time interval, is the picture of the i-th frame The corresponding time, is the picture of the i-th frame, is the actual time interval, which refers to the initial adaptive time interval. is the adaptive time interval after update, is the adaptive coefficient, is the clarity evaluation value, is the clarity change; It should be noted that is the adaptive coefficient, usually set to 0.05-0.2, used to control the amplitude of the time interval change; It should be noted that the constraints for adaptive time interval removal are: , where is the time interval between adjacent frames, that is, the actual time interval, is the clarity change, is the actual time interval, is the second threshold.

[0071] It should be noted that the second threshold is obtained by adopting a supervised learning method, using a large number of video samples and elimination results as training data, training a deep learning model, and predicting the optimal threshold as the second threshold.

[0072] Figure 2 The system of the digitalized individual maintenance control method of the present invention includes: An image segmentation processing module is used to obtain an image of the maintenance site, and segment and preprocess the image to obtain a first superpixel region set; An adjustment value acquisition module, used to acquire first data of a first superpixel region set, and establish an image compression evaluation model based on a convolutional neural network to generate an image adjustment value for each superpixel region; The image adjustment and video conversion module is used to compress each superpixel area of ​​the maintenance site image separately based on the image adjustment value to obtain an adjusted image, and traverse each maintenance site image as a picture frame to form a maintenance site video; The video update upload module is used to update the maintenance site video according to the elimination rules and upload it to the AR end to assist substation maintenance.

[0073] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0074] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product.

[0075] Those of ordinary skill in the art will appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0076] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0077] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0078] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A digitalized individual maintenance and control method, characterized in that: The steps include: Acquire a maintenance site image, and segment and preprocess the image to obtain a first superpixel region set; Acquire first data of a first superpixel region set, and establish an image compression evaluation model based on a convolutional neural network to generate an image adjustment value for each superpixel region; Based on the image adjustment value, each superpixel area of ​​the maintenance site image is compressed separately to obtain an adjusted image, and each maintenance site image is traversed as a picture frame to form a maintenance site video; The maintenance site video is updated according to the elimination rules and uploaded to the AR terminal to assist in substation maintenance.

2. The digitalized individual maintenance and control method according to claim 1 is characterized in that: The image is segmented and preprocessed to obtain a first superpixel region set, specifically: The first superpixel region includes a plurality of superpixel regions; Segment the maintenance site image based on the superpixel segmentation technology to obtain a second superpixel region set; Performing HSV color conversion on each superpixel region of the second superpixel region set to obtain corresponding color features, and performing data analysis based on the color features to mark the dirty areas; Based on the dirty area and Otsu threshold segmentation method, the dirty area in the second superpixel area set is eliminated, and the first superpixel area set is updated.

3. The digitalized individual maintenance and control method according to claim 2 is characterized in that: The first data includes texture complex data, and the specific acquisition method thereof is as follows: Get the pixel points of each superpixel area, process the pixel points through LBP, obtain the variance of the LBP value of the pixel point, and use the variance of the LBP value in each superpixel area as the texture complexity data.

4. The digitalized individual maintenance and control method according to claim 3 is characterized in that: The first data includes target area proportion data, and the specific acquisition method is as follows: Acquire pixel data of key devices, and determine the pixel ratio of key device pixels in each superpixel area based on a pixel statistics method to obtain a first ratio; The preset pixel ratio threshold is compared with the first ratio to determine whether each superpixel area is a target area, and the ratio of the target area to the superpixel area is used as the target area ratio data.

5. The digitalized individual maintenance and control method according to claim 4 is characterized in that: The first data includes equipment failure rate weight compression rate data, and the specific acquisition method is as follows: Based on the target detection network, the equipment type of each superpixel area in the maintenance site image is identified and combined with the equipment history database to obtain equipment information, and the failure rate weight of each equipment is calculated; A device failure rate weight compression data ratio model is constructed based on the device failure rate weight and a preset image compression rate, and the device failure rate weight compression rate data is output.

6. The digitalized individual maintenance and control method according to claim 5 is characterized in that: The image compression evaluation model is established based on the convolutional neural network to generate the image adjustment value of each superpixel area, specifically: According to the first data, a multidimensional image compression evaluation model is established based on a convolutional neural network, and a picture adjustment value of each superpixel area is obtained based on the multidimensional image compression evaluation model.

7. The digitalized individual maintenance and control method according to claim 6 is characterized in that: The method compresses each superpixel region of the maintenance site image based on the image adjustment value to obtain an adjusted image, and traverses each maintenance site image as a picture frame to form a maintenance site video, specifically: Preset the image standard value of the maintenance site image, and compare it with the image adjustment value of each super-pixel area of ​​the maintenance site image to obtain the adjustment ratio of each super-pixel area; Based on the adjustment ratio, each superpixel area of ​​the maintenance site image is respectively adjusted in a corresponding ratio to obtain an adjusted image; Each maintenance site image is traversed and adjusted as a picture frame to form a maintenance site video.

8. The digitalized individual maintenance and control method according to claim 7 is characterized in that: The elimination rules include clarity elimination, specifically: Obtain the maintenance site video frame image and calculate the Laplacian variance of the frame image according to the Laplacian transformation as the clarity evaluation value, remove the frame image corresponding to the clarity evaluation value less than the preset first threshold, and update the maintenance site video.

9. The digitalized individual maintenance and control method according to claim 8 is characterized in that: The elimination rule includes adaptive time interval elimination, specifically: The actual time interval of the video frame images of the maintenance site after the update is obtained as the adaptive time interval; the definition change of adjacent frame images is calculated by Laplacian transformation as the weight of the adaptive time interval, and the adaptive time interval is adjusted and updated; When the actual time interval is less than the adaptive time interval, the clarity change between the frame image and the previous frame image is calculated; if the clarity change is less than the set second threshold, the frame image corresponding to the actual time interval is eliminated and the maintenance site video is updated.

10. A system using the digitalized individual maintenance and control method according to any one of claims 1 to 9, characterized in that: include: An image segmentation processing module is used to obtain an image of the maintenance site, and segment and preprocess the image to obtain a first superpixel region set; An adjustment value acquisition module, used to acquire first data of a first superpixel region set, and establish an image compression evaluation model based on a convolutional neural network to generate an image adjustment value for each superpixel region; The image adjustment and video conversion module is used to compress each superpixel area of ​​the maintenance site image separately based on the image adjustment value to obtain an adjusted image, and traverse each maintenance site image as a picture frame to form a maintenance site video; The video update upload module is used to update the maintenance site video according to the elimination rules and upload it to the AR end to assist substation maintenance.

Images