Maintenance management and control method based on artificial intelligence

Abstract

The invention discloses a maintenance management and control method based on artificial intelligence, and the method comprises the steps: obtaining corresponding maintenance work information through analyzing a maintenance work ticket, then collecting an entrance monitoring video of a maintenance work region, carrying out the recognition of a person through the combination of the maintenance work information, and carrying out the maintenance management and control of the person. Judging whether a person going to enter the operation area is an authorized person or not, and if yes, opening an access control; then, after the personnel enter the operation area, the operation video of the personnel and the operation data of the personnel on the operation equipment are collected in real time; and finally, according to the maintenance operation information and the collected operation video and operation data, operation violation identification is carried out, and when violation operation of the personnel is identified, blocking operation is carried out and an alarm is given. Therefore, according to the method, digitalization and automation of maintenance management and control are realized, and compared with manual monitoring, the real-time performance is high, the efficiency is high, and the reliability is high; therefore, the method is very suitable for large-scale application and popularization in the field of operation and maintenance management.

Description

Technical Field

[0001] The present invention belongs to the field of operation and maintenance management technology, and specifically relates to an inspection and maintenance control method based on artificial intelligence. Background Art

[0002] With the digital transformation and development of power grid companies, the number of their information systems and equipment is growing rapidly. To ensure the reliable operation of each device, it is usually necessary to regularly inspect and maintain the various information devices in the enterprise. Currently, in actual use, most equipment maintenance work is manually initiated and then approved by superiors. After the approval is completed, the maintenance personnel enter the maintenance operation area, log in to the maintenance equipment, and remotely inspect the equipment in different areas.

[0003] Among them, enterprises need to monitor the maintenance process in real time to ensure the compliance of maintenance (such as whether the maintenance is carried out on time, whether there are any illegal operations, etc.). However, due to the lack of a complete operation and maintenance management system, enterprises need to assign dedicated personnel to carry out maintenance monitoring, which is not only inefficient and unreliable, but also has low real-time performance. As a result, enterprises cannot monitor the entire maintenance process, resulting in inadequate maintenance supervision, many illegal operations, and the inability to effectively manage maintenance operations. Based on this, how to carry out digital maintenance management to ensure the compliance of maintenance has become an urgent problem to be solved. Summary of the Invention

[0004] The purpose of the present invention is to provide an artificial intelligence-based maintenance control method to solve the problems of poor reliability, low efficiency and poor real-time performance of manual monitoring in the existing technology.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] In the first aspect, a maintenance control method based on artificial intelligence is provided, comprising:

[0007] Obtain the issued maintenance work ticket and parse the maintenance work ticket to obtain the maintenance personnel information, maintenance operation information and maintenance time information recorded in the maintenance work ticket;

[0008] Upon receiving an application for entry into the maintenance work area, obtaining a surveillance video of the entrance to the maintenance work area, and performing facial recognition processing on the personnel in the surveillance video to obtain a facial recognition result;

[0009] Based on the face recognition result, the maintenance time information and the maintenance personnel information, determining whether the person in the entrance surveillance video is an authorized person;

[0010] If yes, a door opening command is sent to the access control system of the maintenance operation area;

[0011] When the authorized person is detected to have entered the maintenance operation area, the operation video in the maintenance operation area is acquired in real time, and the equipment operation data of the operating equipment logged in by the authorized person is collected in real time;

[0012] Determine whether the maintenance work performed by the authorized personnel is in compliance with regulations based on the operation video, equipment operation data, and maintenance operation information;

[0013] If so, the device operation data is transmitted to the bastion host cluster so that the device operation data is sent to the device to be repaired through the bastion host cluster; otherwise, the device operation data is blocked and a violation alarm is issued.

[0014] Based on the above disclosed content, the present invention first obtains the issued maintenance work ticket, and then parses the maintenance personnel information, maintenance operation information and maintenance time information recorded therein; then, based on the above information, the entire maintenance process can be monitored, that is, when an application to enter the maintenance operation area is received, the entrance surveillance video of the maintenance operation area is obtained, and then, the person in the entrance surveillance video is subjected to facial recognition to obtain a facial recognition result; then, based on the facial recognition result, and in combination with the above maintenance personnel information and time information, it is determined whether the person who is about to work is an authorized person; if so, the passage is opened, that is, the access control of the maintenance operation area is opened; and when the authorized personnel enters the maintenance operation area, the present invention collects their operation video and the operation data of the login equipment in real time, so as to judge whether the authorized personnel have violated the regulations based on the maintenance operation information and the collected operation video and operation data. If so, their operation is blocked and an alarm is issued. Otherwise, their operation is allowed, thereby realizing the maintenance of the equipment to be maintained.

[0015] Through the above design, the present invention provides a digital maintenance control method, which can realize full-process monitoring of maintenance operations from personnel entry and verification to personnel operations during maintenance operations, and can block operations and issue alarms when illegal operations occur; thus, compared with manual monitoring, the present invention is not only highly real-time and efficient, but also highly reliable; therefore, it is very suitable for large-scale application and promotion in the field of operation and maintenance management.

[0016] In one possible design, face recognition processing is performed on the person in the entrance surveillance video to obtain a face recognition result, including:

[0017] Extracting a surveillance image containing a person's face from the entrance surveillance video;

[0018] Performing denoising processing on the monitoring image to obtain a denoised image;

[0019] Performing shadow detection processing on the denoised image to obtain shadow pixels in the denoised image;

[0020] Based on the shadow pixels in the denoised image, performing image enhancement processing on the denoised image to obtain an enhanced image;

[0021] Performing face recognition processing on the enhanced image to obtain the face recognition result.

[0022] In one possible design, performing denoising on the surveillance image to obtain a denoised image includes:

[0023] Performing wavelet decomposition processing on the monitoring image to obtain a low-frequency component image and a high-frequency component image;

[0024] Using an improved filtering algorithm to filter the low-frequency component image to obtain a filtered low-frequency component image;

[0025] Determining a wavelet threshold corresponding to the high-frequency component image, and performing denoising processing on the high-frequency component image based on the wavelet threshold to obtain a denoised high-frequency component image;

[0026] An image fusion process is performed on the filtered low-frequency component image and the denoised high-frequency component image to obtain the denoised image after the image fusion process.

[0027] In one possible design, an improved filtering algorithm is used to filter the low-frequency component image to obtain a filtered low-frequency component image, including:

[0028] For any pixel point in the low-frequency component image, obtaining a filtering window of the any pixel point;

[0029] Using the grayscale values ​​of each pixel in the filtering window to form a grayscale set, and removing the maximum grayscale value and the minimum grayscale value from the grayscale set to obtain an updated grayscale set;

[0030] Determining a noise threshold based on the updated grayscale set;

[0031] Determining whether any pixel is a noise point according to the noise threshold;

[0032] If so, the grayscale value of any pixel is updated to the noise threshold, and after all pixels in the low-frequency component image are polled, the filtered low-frequency component image is obtained.

[0033] In a possible design, determining the wavelet threshold corresponding to the high-frequency component image includes:

[0034] Obtain the decomposition scale of each high-frequency wavelet coefficient corresponding to the high-frequency component image, as well as the noise variance of the high-frequency component image;

[0035] Calculating a first threshold value according to the noise variance;

[0036] Calculating a second threshold at each decomposition scale based on the first threshold and the decomposition scale of each high-frequency wavelet coefficient;

[0037] The wavelet threshold is composed by using the second threshold at each decomposition scale.

[0038] In one possible design, performing shadow detection processing on the denoised image to obtain shadow pixels in the denoised image includes:

[0039] Convert the denoised image into an HSV image to obtain the brightness and saturation of each pixel in the denoised image;

[0040] Calculate the shadow feature value of each pixel based on the brightness and saturation of each pixel;

[0041] Dividing the denoised image into a background image and a foreground image;

[0042] Calculating a shadow feature threshold based on the background image, the foreground image, and the denoised image;

[0043] The shadow pixel points in the denoised image are determined using the shadow feature threshold and the shadow feature value of each pixel point.

[0044] In one possible design, performing image enhancement processing on the denoised image based on shadow pixels in the denoised image to obtain an enhanced image includes:

[0045] Calculating a first pixel variance and a pixel average of all shadow pixels in the denoised image, and a second pixel variance of all non-shadow pixels in the denoised image;

[0046] Get the neighborhood area of ​​each shadow pixel;

[0047] Each shadow pixel is enhanced based on the second pixel variance, the neighborhood area of ​​each shadow pixel, the first pixel variance, and the pixel average value, so as to obtain the enhanced image after the enhancement process.

[0048] In one possible design, determining whether the person in the entrance surveillance video is an authorized person based on the face recognition result, the maintenance time information, and the maintenance personnel information includes:

[0049] Obtaining the application time for entering the maintenance operation area, and using the application time as the personnel arrival time;

[0050] Determine whether the personnel arrival time matches the maintenance time information;

[0051] If so, determining whether the face recognition result matches the maintenance personnel information;

[0052] If so, it is determined that the person in the entrance surveillance video is an authorized person; otherwise, it is determined that the person in the entrance surveillance video is an illegal person.

[0053] In one possible design, the maintenance operation information includes: maintenance operation data corresponding to the illegal behavior and legal operation authority. The determination of whether the maintenance operation performed by the authorized personnel is compliant is performed based on the operation video, equipment operation data, and maintenance operation information, including:

[0054] Performing behavior recognition processing on the authorized personnel in the operation video to obtain the inspection behavior of the authorized personnel;

[0055] Determine whether the inspection behavior is a violation;

[0056] If not, determining whether the equipment operation data is maintenance operation data corresponding to the legal operation authority;

[0057] If not, the maintenance work performed by the authorized personnel is deemed to be an illegal work.

[0058] In one possible design, before obtaining the issued maintenance work ticket, the method further includes:

[0059] Obtaining an initial work ticket, wherein the initial work ticket contains maintenance personnel information and a maintenance information summary;

[0060] Setting the maintenance personnel in the maintenance personnel information as an authorized personnel, and allocating a legal login account for the operating equipment in the maintenance operation area to the authorized personnel;

[0061] Generating maintenance time information and maintenance operation information according to the maintenance information summary;

[0062] Recording the legal login account, the maintenance time information, and the maintenance operation information into the initial filled-in work ticket, so as to obtain the issued maintenance work ticket after recording;

[0063] Accordingly, after completing the maintenance work of the equipment to be repaired, the method further includes:

[0064] The login authority of the legal login account is canceled, and the device to be repaired is restored to clear all data during this repair process.

[0065] Secondly, an artificial intelligence-based maintenance management and control system is provided, including:

[0066] A parsing unit, configured to obtain an issued maintenance work ticket and parse the maintenance work ticket to obtain maintenance personnel information, maintenance operation information, and maintenance time information recorded in the maintenance work ticket;

[0067] A personnel verification unit is used to obtain a surveillance video of the entrance to the maintenance operation area upon receiving an application for entry to the maintenance operation area, and perform facial recognition processing on the personnel in the surveillance video to obtain a facial recognition result;

[0068] A personnel verification unit, configured to determine whether the person in the entrance surveillance video is an authorized person based on the face recognition result, the maintenance time information, and the maintenance personnel information;

[0069] The personnel verification unit is further configured to send a door opening instruction to the access control system of the maintenance operation area when it is determined that the person in the entrance surveillance video is an authorized person;

[0070] An operation monitoring unit is used to obtain the operation video in the maintenance operation area in real time when detecting that the authorized personnel enters the maintenance operation area, and to collect the equipment operation data of the operating equipment logged in by the authorized personnel in real time;

[0071] The operation monitoring unit is also used to determine whether the maintenance operation of the authorized personnel is compliant based on the operation video, equipment operation data and maintenance operation information;

[0072] The operation monitoring unit is also used to transmit the equipment operation data to the bastion host cluster when it is determined that the maintenance operation of the authorized personnel is compliant, so that the equipment operation data can be sent to the equipment to be repaired through the bastion host cluster; and when it is determined that the maintenance operation of the authorized personnel is illegal, block the equipment operation data and issue a violation alarm.

[0073] In the third aspect, another maintenance and control device based on artificial intelligence is provided. Taking the device as an electronic device as an example, it includes a memory, a processor and a transceiver that are communicatively connected in sequence, wherein the memory is used to store computer programs, the transceiver is used to send and receive messages, and the processor is used to read the computer program and execute the maintenance and control method based on artificial intelligence as described in the first aspect or any possible design of the first aspect.

[0074] In a fourth aspect, a storage medium is provided, on which instructions are stored. When the instructions are run on a computer, the artificial intelligence-based maintenance and control method as described in the first aspect or any possible design of the first aspect is executed.

[0075] In a fifth aspect, a computer program product comprising instructions is provided, which, when executed on a computer, causes the computer to execute the artificial intelligence-based maintenance and control method as described in the first aspect or any possible design of the first aspect.

[0076] Beneficial effects:

[0077] (1) The present invention provides a digital maintenance control method, which can monitor the entire maintenance process from personnel entry and verification to personnel operation during maintenance operations, and can block operations and issue alarms when illegal operations occur. In this way, compared with manual monitoring, the present invention is not only highly real-time and efficient, but also highly reliable; therefore, it is very suitable for large-scale application and promotion in the field of operation and maintenance management. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] Figure 1 A schematic diagram of the steps of the maintenance control method based on artificial intelligence provided by an embodiment of the present invention;

[0079] Figure 2 A schematic diagram of the structure of an artificial intelligence-based maintenance management and control system provided in an embodiment of the present invention;

[0080] Figure 3 A schematic structural diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0081] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the present invention will be briefly introduced below in conjunction with the drawings and the description of the embodiments or the prior art. Obviously, the following description of the structure of the drawings is only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work. It should be noted that the description of these embodiments is used to help understand the present invention, but does not constitute a limitation of the present invention.

[0082] It should be understood that although the terms "first," "second," etc. may be used herein to describe various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another. For example, a first element can be referred to as a second element, and similarly, a second element can be referred to as a first element without departing from the scope of the exemplary embodiments of the present invention.

[0083] It should be understood that the term "and / or" that may appear in this document is merely a description of the association relationship between associated objects, indicating that three relationships may exist. For example, A and / or B may indicate three situations: A exists alone, B exists alone, and A and B exist at the same time. The term " / and" that may appear in this document describes another type of association object relationship, indicating that two relationships may exist. For example, A / and B may indicate two situations: A exists alone, and A and B exist alone. In addition, the character " / " that may appear in this document generally indicates that the previous and subsequent associated objects are in an "or" relationship.

[0084] Example:

[0085] See also Figure 1 As shown, the maintenance control method based on artificial intelligence provided by this embodiment can obtain the corresponding maintenance work information by parsing the maintenance work ticket, and then collect the entrance surveillance video of the maintenance operation area, and combine it with the aforementioned maintenance work information to perform personnel identification, that is, determine whether the person about to enter the operation area is an authorized person. If so, open the access control; then, after the person enters the operation area, the person's operation video and the person's operation data on the operation equipment are collected in real time; finally, the operation violation can be identified based on the aforementioned maintenance operation information, as well as the collected operation video and operation data, and when it is identified that the person is operating in violation of the regulations, the blocking operation is performed and an alarm is issued; thus, this method realizes the digitization and automation of maintenance control. Compared with manual monitoring, it is not only real-time and efficient, but also reliable; thus, this method is very suitable for large-scale application and promotion in the field of operation and maintenance management; among them, for example, this method can be, but is not limited to, running on the operation and maintenance monitoring platform side. It can be understood that the aforementioned execution subject does not constitute a limitation on the embodiment of the present application. Accordingly, the operation steps of this method can be, but are not limited to, as shown in the following steps S1 to S7.

[0086] S1. Obtain the issued maintenance work ticket and parse the maintenance work ticket to obtain the maintenance personnel information, maintenance operation information and maintenance time information recorded in the maintenance work ticket; in the specific implementation, for example, but not limited to, the following steps S01 to S04 can be used to obtain the issued maintenance work ticket.

[0087] S01. Obtain an initial work ticket, wherein the initial work ticket contains maintenance personnel information and a summary of maintenance information; in this embodiment, the initial work ticket is applied for and filled out by the maintenance personnel, wherein the maintenance personnel information includes the name, work number, organizational structure, and facial image of the maintenance personnel, and the summary of maintenance information contains an information overview of this maintenance operation, such as the maintenance object (i.e., the equipment to be repaired), maintenance matters (i.e., the specific content of the maintenance, description of equipment problems fed back by the front end, etc.), etc.; in this way, after obtaining the aforementioned initial work ticket, personnel authorization, operating equipment account allocation, and maintenance strategy (i.e., maintenance time and operation data) can be issued, so that after completing the aforementioned work, the initial work ticket can be issued and an issued maintenance work ticket can be obtained; wherein the aforementioned process can be, but is not limited to, as shown in the following steps S02 to S04.

[0088] S02. Set the maintenance personnel in the maintenance personnel information as an authorized personnel, and assign a legal login account for the operating equipment in the maintenance operation area to the authorized personnel; in this embodiment, the maintenance personnel is determined based on the name and work number in the maintenance personnel information, and is thus set as an authorized personnel; at the same time, a temporary login account for the operating equipment can be assigned to the aforementioned authorized personnel to serve as their legal login account during maintenance operations; further, the legal login account can also be bound to the operating equipment in the maintenance operation area, that is, the operating equipment used for this maintenance operation is determined, and then the legal login account is used as the only legal account for the aforementioned operating equipment during this maintenance process; specifically, the process of obtaining the operating equipment is explained below.

[0089] After the personnel authorization is completed, a maintenance strategy can be generated, and the process is shown in the following step S03.

[0090] S03. Generate maintenance time information and maintenance operation information based on the maintenance information summary; in this embodiment, it has been explained above that the maintenance information summary contains maintenance objects and maintenance matters, so the idle time of the maintenance object can be obtained to generate maintenance time information based on the idle time (that is, spatial time is used as maintenance time); then, according to the specific content of the maintenance and the problems fed back by the front end, the corresponding maintenance strategy, the legal operation authority corresponding to the maintenance strategy, and the maintenance operation data corresponding to the legal operation authority can be matched in the maintenance plan library; at the same time, it is also necessary to obtain prohibited behaviors in the operating area as violations, and then use The maintenance operation information is composed of the maintenance operation data corresponding to the violation, maintenance strategy and legal operation authority; further, after obtaining the maintenance time, the equipment that is idle during the maintenance time can be screened out from all the equipment in the maintenance operation area, and thus used as the operating equipment for this maintenance; finally, the aforementioned legal login account is bound to the operating equipment for this maintenance, so that a one-to-one correspondence between the account and the equipment can be achieved; and after obtaining the aforementioned maintenance time information and maintenance operation information, they can be recorded together with the legal login account in the initial work ticket, so that the issuance of the work ticket can be completed, and the process is shown in the following step S04.

[0091] S04. Record the legal login account, the maintenance time information and the maintenance operation information into the initially filled-in work ticket, so as to obtain the issued maintenance work ticket after recording.

[0092] In this way, through the aforementioned steps S01 to S04, the issuance of the initial work ticket can be completed, and a maintenance work ticket containing maintenance personnel information, maintenance operation information and maintenance time information can be obtained; then, the maintenance work ticket can be parsed to obtain the aforementioned information, so as to complete the entire process control of the maintenance work based on the aforementioned information.

[0093] Among them, this embodiment realizes the whole process control of the maintenance work from two aspects: personnel verification and personnel operation monitoring; wherein, the personnel verification process can be but is not limited to the following steps S2 and S3.

[0094] S2. When an application for entry into the maintenance work area is received, the entrance surveillance video of the maintenance work area is obtained, and facial recognition processing is performed on the personnel in the entrance surveillance video to obtain facial recognition results; in specific applications, in maintenance scenarios, maintenance personnel usually enter the maintenance area (i.e., the maintenance work area) to use the internal operating equipment to remotely perform maintenance operations on the maintenance objects (such as system maintenance, equipment firewall settings, equipment virus removal, etc.); therefore, according to this scenario, this embodiment sets up access control at the entrance of the maintenance area, and combines facial recognition and the maintenance personnel information in the issued maintenance work ticket to determine whether the personnel entering the maintenance work area are authorized personnel.

[0095] Specifically, in this embodiment, an entry application button is provided at the access control of the maintenance work area. When a person presses the application button, an entry application to the maintenance work area is generated. At this time, the operation and maintenance monitoring platform can obtain the video taken by the camera at the access control, that is, the entrance monitoring video; then, the face recognition of the personnel in the entrance monitoring video can be performed to realize personnel verification.

[0096] Optionally, an implementation method of the following public face recognition may be, but is not limited to, steps S21 to S25 as shown below.

[0097] S21. Extracting a surveillance image containing a person's face from the entrance surveillance video; in this embodiment, the entrance surveillance video may be first framed, and then a target detection algorithm may be used to perform target detection on each framed image, thereby obtaining a surveillance image containing a person's face; and after obtaining the surveillance image containing the person's face, denoising processing may be performed, and the process is shown in the following step S22.

[0098] S22. De-noising the surveillance image to obtain a denoised image. In this embodiment, since the surveillance video may be interfered with by the environment and equipment during transmission, the surveillance video may contain significant interference noise. Therefore, denoising is required before performing face recognition to reduce the interference of noise on face recognition.

[0099] This embodiment discloses an improved denoising method, the process of which is shown in the following steps S22a to S22d.

[0100] S22a. Perform wavelet decomposition processing on the monitoring image to obtain a low-frequency component image and a high-frequency component image; in this embodiment, the monitoring image is first grayscale transformed, and then the wavelet basis function of sym4 is used to perform three-layer wavelet decomposition on the monitoring image, thereby obtaining low-frequency and high-frequency wavelet coefficients at different decomposition scales, and then, the signal subband corresponding to the low-frequency wavelet coefficient is reconstructed to obtain the low-frequency component image; similarly, the signal subband corresponding to the high-frequency wavelet coefficient is reconstructed to obtain the high-frequency component image; in this way, after obtaining the low-frequency and high-frequency component images, different denoising methods can be adopted for component images of different frequencies, and the process can be but is not limited to the following steps S22b to S22c.

[0101] S22b. Use the improved filtering algorithm to filter the low-frequency component image to obtain a filtered low-frequency component image; in specific applications, this embodiment constructs a filtering window and uses the grayscale value within the window to complete the denoising of the low-frequency component image. The process can be but is not limited to the following steps S22b1 to S22b5.

[0102] S22b1. For any pixel point in the low-frequency component image, obtain a filtering window of the any pixel point; in this embodiment, the filtering window of the any pixel point is a window with a size of n×n centered on the any pixel point, and n is an odd number; wherein, for example, n is preferably taken as 3; of course, it can be specifically set according to actual use and is not limited to the above example; and after obtaining the filtering window of the any pixel point, the noise threshold can be determined according to the grayscale value of the pixel point in the filtering window, and the process is shown in the following steps S22b2 and S22b3.

[0103] S22b2. Use the grayscale values ​​of each pixel point in the filtering window to form a grayscale set, and remove the maximum grayscale value and the minimum grayscale value from the grayscale set to obtain an updated grayscale set; in this embodiment, the filtering window is actually the neighborhood area of ​​any pixel point; therefore, it is equivalent to screening out the pixel points with the maximum grayscale value and the minimum grayscale value from each pixel point in the neighborhood area, and then using the grayscale values ​​of the remaining pixel points to form the aforementioned updated grayscale set; and after obtaining the updated grayscale set, the noise threshold can be calculated based on this, so that the noise threshold can be used to determine whether any pixel point is a noise point; wherein, the calculation process of the noise threshold can be but is not limited to the following step S22b3.

[0104] S22b3. Based on the updated grayscale set, a noise threshold is determined; in this embodiment, the grayscale values ​​in the updated grayscale set can be sorted in ascending order to obtain a sorted sequence; then, the grayscale values ​​at the target positions in the sorted sequence are taken (there are 2 target positions); then, the average of the grayscale values ​​at the target positions is calculated as the noise threshold; further, when the number of grayscale values ​​in the sorted sequence is an odd number, the target position is the middle position in the sorted sequence, and the position before or after the middle position, and when the number of grayscale values ​​in the sorted sequence is an even number, the target position is the middle position.

[0105] Thus, after obtaining the noise threshold based on the aforementioned step S22b3, it is possible to determine whether any pixel point is a noise point based on the noise threshold. The process may be, but is not limited to, as shown in the following step S22b4.

[0106] S22b4. According to the noise threshold, determine whether any pixel point is a noise point; in this embodiment, for example, but not limited to, first calculate the difference between the noise threshold and the grayscale value of any pixel point; then, take the absolute value of the difference; then, calculate the ratio between the noise threshold and 5; finally, determine whether the absolute value of the difference is less than or equal to the aforementioned ratio; wherein, if so, determine that any pixel point is a noise point, otherwise, determine that any pixel point is a non-noise point.

[0107] In this way, after determining that any pixel point is a noise point, filtering processing can be performed, and the process is shown in the following step S22b5.

[0108] S22b5. If so, the grayscale value of any pixel point is updated to the noise threshold, and after all the pixel points in the low-frequency component image are polled, the filtered low-frequency component image is obtained; in this embodiment, if any pixel point is a noise point, the grayscale value is replaced, that is, its grayscale value is updated to the noise threshold; on the contrary, if any pixel point is not a noise point, no processing is performed, that is, the pixel point is moved, and the next pixel point is judged, that is, the aforementioned steps S22b1 to S22b5 are repeated until all the pixel points in the low-frequency component image are polled, and the filtered low-frequency component image can be obtained.

[0109] Thus, through the aforementioned steps S22b1 to S22b5, this embodiment adopts an adaptive noise determination method to implement filtering processing of the low-frequency component image, thereby improving the filtering effect.

[0110] After the denoising process of the low-frequency component image is completed, the denoising process of the high-frequency component image may be performed. The process may be, but is not limited to, as shown in the following step S22c.

[0111] S22c. Determine the wavelet threshold corresponding to the high-frequency component image, and based on the wavelet threshold, denoise the high-frequency component image to obtain a denoised high-frequency component image; in specific applications, for example, but not limited to, the following steps S22c1 to S22c4 can be used to calculate the wavelet threshold.

[0112] S22c1. Obtain the decomposition scale of each high-frequency wavelet coefficient corresponding to the high-frequency component image, as well as the noise variance of the high-frequency component image; in specific implementation, the noise variance can be, but is not limited to, calculated through each high-frequency wavelet coefficient, that is, sorting each high-frequency wavelet coefficient in ascending order, and then taking the median of the sort; then, dividing the median by 0.6475, so that the noise variance can be obtained; of course, using high-frequency wavelet coefficients to estimate the noise variance is a common technique in wavelet denoising, and its principle will not be repeated here.

[0113] After obtaining the noise variance and the decomposition scale of each high-frequency wavelet coefficient, the first threshold value can be calculated based on the noise variance. The process can be, but is not limited to, as shown in the following step S22c2.

[0114] S22c2. According to the noise variance, a first threshold is calculated; in specific applications, it is equivalent to calculating the classic wavelet threshold based on the noise variance, that is, multiplying Thus, a first threshold is obtained, where X and Y represent the width and height of the high-frequency component image.

[0115] In this way, after the first threshold is calculated, the second threshold at each decomposition scale can be obtained according to the decomposition scale. The calculation process can be, but is not limited to, as shown in the following step S22c3.

[0116] S22c3. Based on the first threshold and the decomposition scale of each high-frequency wavelet coefficient, calculate the second threshold at each decomposition scale; in specific implementation, for example, the following formula (1) can be used to calculate the second threshold.

[0117]

[0118] In the above formula (1), T j represents the second threshold value under the jth decomposition scale, j represents the decomposition scale, σ represents the noise variance, that is, Represents the first threshold, and the second threshold is based on the first threshold, multiplied by 2 -j .

[0119] After calculating the second threshold value at each decomposition scale based on the aforementioned formula (1), a wavelet threshold value can be formed based on the second threshold value. The process is shown in the following step S22c4.

[0120] S22c4. Use the second threshold at each decomposition scale to form the wavelet threshold.

[0121] In this way, through the aforementioned steps S22c1 to S22c4, an adaptive wavelet threshold can be calculated based on the decomposition scale; and then, wavelet denoising of the high-frequency component image can be performed based on the wavelet threshold.

[0122] For the high-frequency wavelet coefficients at the j-th decomposition scale in the high-frequency component image, the updated high-frequency wavelet coefficients can be obtained by, for example but not limited to, using the following formula (2).

[0123]

[0124] In the above formula (2), w j represents the high-frequency wavelet coefficient at the jth decomposition scale, w′ j Represents the updated high-frequency wavelet coefficients, and sgn() represents the sign function.

[0125] In this way, each high-frequency wavelet coefficient can be updated through the above formula (2); then, by combining the signal subbands of each updated high-frequency wavelet coefficient, the high-frequency component image can be denoised to obtain a denoised high-frequency component image; that is, each updated high-frequency wavelet coefficient is used as a new wavelet coefficient of each high-frequency signal subband to obtain a denoised high-frequency signal subband; finally, the signal of the denoised high-frequency signal subband is reconstructed to obtain a denoised high-frequency component image.

[0126] After obtaining the denoised high-frequency component image through the aforementioned step S22c, the denoised image can be generated by combining it with the aforementioned filtered low-frequency component image. The process can be, but is not limited to, as shown in the following step S22d.

[0127] S22d. Perform image fusion processing on the filtered low-frequency component image and the denoised high-frequency component image to obtain the denoised image after the image fusion processing.

[0128] Thus, through the aforementioned steps S22a to S22d, the image denoising process can be completed; at the same time, since the maintenance work area is usually indoors with strong lighting, more obvious shadows will be produced in the actual monitoring video; and the shadows will interfere with the subsequent face recognition. Based on this, in order to improve the accuracy of face recognition, this embodiment also requires shadow removal processing, and the process can be but is not limited to the following step S23.

[0129] S23. Perform shadow detection processing on the denoised image to obtain shadow pixels in the denoised image. In a specific application, this embodiment converts the image into the HSV space and then determines the shadow pixels based on brightness and saturation. The process may be, but is not limited to, steps S23a to S23e as shown below.

[0130] S23a. Convert the denoised image to an HSV image to obtain the brightness and saturation of each pixel in the denoised image. In a specific implementation, the denoised image is first restored to obtain a color image. Then, the R, G, and B values ​​of each pixel in the color image are normalized, that is, the R, G, and B values ​​of each pixel are divided by 255 to obtain the normalized R value, normalized G value, and normalized B value. Then, for any pixel in the color image, the brightness is taken as the maximum value of the normalized R value, normalized G value, and normalized B value. The saturation is calculated as follows:

[0131] First, determine whether the maximum value among the normalized R value, normalized G value, and normalized B value is 0; if so, the saturation of any pixel point is set to 0; otherwise, calculate the difference between the maximum and minimum values ​​among the normalized R value, normalized G value, and normalized B value; then, divide the difference by the aforementioned maximum value to take the result as the saturation of any pixel point.

[0132] In this way, based on the above method, after obtaining the brightness and saturation of each pixel in the denoised image, the shadow feature value of each pixel in the denoised image can be calculated based on this. The process can be but is not limited to the following step S23b.

[0133] S23b according to the brightness and saturation of each pixel, calculate the shadow feature value of each pixel; in the specific implementation, or to any pixel in the denoised image as an example, the shadow feature value is calculated as follows:

[0134]

[0135] In the above formula (3), y represents the shadow feature value of any pixel in the denoised image, and S and V represent the brightness and saturation of any pixel in the denoised image.

[0136] Thus, after calculating the shadow feature value of each pixel in the denoised image based on the aforementioned formula (3), it is also necessary to calculate the shadow feature threshold so that the shadow pixels can be identified based on it later; wherein, the calculation process of the shadow feature threshold can be, but is not limited to, as shown in the following steps S23c and S23d.

[0137] S23c. Divide the denoised image into a background image and a foreground image; in this embodiment, it is equivalent to separating the background and the target to obtain a background image and a foreground image; then, based on this, the shadow feature threshold can be calculated, and the process can be but is not limited to the following step S23d.

[0138] S23d. Calculate the shadow feature threshold based on the background image, the foreground image and the denoised image; in this embodiment, first calculate the first grayscale parameter based on the denoised image; then, calculate the second grayscale parameter based on the background image; then, calculate the third grayscale parameter based on the foreground image; finally, calculate the shadow feature threshold based on the first grayscale parameter, the second grayscale parameter and the third grayscale parameter.

[0139] Optionally, the first grayscale parameter can be calculated by, for example but not limited to, using the following formula (4).

[0140]

[0141] In the above formula (4), υ1 represents the first grayscale parameter, P m represents the probability of a pixel with gray level m in the denoised image, where the number of pixels with gray level m in the denoised image is first counted (gray level refers to gray value, such as the number of pixels with gray value 1), and then the number of pixels with gray level m is divided by the total number of pixels in the denoised image to get P m .

[0142] Furthermore, for example, but not limited to, the following formula (5) may be used to calculate the second grayscale parameter.

[0143]

[0144] In the above formula (5), υ2 represents the first grayscale parameter, P m represents the probability of a pixel with gray level m in the background image, L represents the maximum gray level in the background gray image, and L is less than 255.

[0145] Similarly, the third grayscale parameter can be calculated by, for example but not limited to, using the following formula (6).

[0146]

[0147] In the above formula (6), υ3 represents the third grayscale parameter.

[0148] Thus, after the three grayscale parameters are calculated by the aforementioned formulas (4)-(6), the shadow feature threshold can be calculated based on the three grayscale parameters; wherein, for example, but not limited to, the shadow feature threshold can be calculated using the following formula (7).

[0149]

[0150] In the above formula (7), κ represents the shadow feature threshold.

[0151] After calculating the shadow feature threshold based on the above formula, the shadow feature values ​​of each pixel can be combined to determine the shadow pixels in the denoised image. The process can be, but is not limited to, as shown in the following step S23e.

[0152] S23e. Determine the shadow pixel points in the denoised image using the shadow feature threshold and the shadow feature value of each pixel point. In this embodiment, if the shadow feature value of any pixel point in the denoised image is less than the shadow feature threshold, then the pixel point is determined to be a shadow pixel point; otherwise, the pixel point is not a shadow pixel point.

[0153] Thus, through the aforementioned steps S23a to S23e, the shadow pixels in the denoised image can be identified, and then the real information of the shadow area can be restored. The process can be but is not limited to the following step S24.

[0154] S24. Based on the shadow pixels in the denoised image, image enhancement processing is performed on the denoised image to obtain an enhanced image; in this embodiment, this is equivalent to performing image enhancement on the shadow pixels to restore their true information; among them, for example, but not limited to, the following steps S24a to S24c can be used to perform image enhancement on the shadow pixels to obtain an enhanced image.

[0155] S24a. Calculate the first pixel variance and pixel average of all shadow pixels in the denoised image, and the second pixel variance of all non-shadow pixels in the denoised image. In this embodiment, since the denoised image is a grayscale value, it is equivalent to calculating the grayscale variance and grayscale average.

[0156] After calculating the pixel variance of the shadow area and the pixel variance of the non-shadow area, the neighborhood area of ​​each shadow pixel point can be obtained so as to combine the neighborhood area of ​​each shadow pixel point to perform enhancement processing on each shadow pixel point. The process can be, but is not limited to, as shown in the following steps S24b and S24c.

[0157] S24b. Obtain the neighborhood area of ​​each shadow pixel point; in specific application, any shadow pixel point is taken as an example for explanation, and the example can be, but is not limited to, taking the area with a size of 3 by 3 centered on the any shadow pixel point as its neighborhood area; of course, the size of the neighborhood area can be specifically set according to actual use, and is not limited to the aforementioned example.

[0158] After obtaining the neighborhood area of ​​each pixel point, the aforementioned pixel variance and average value may be combined to perform enhancement processing on each shadow pixel point. The process may be, but is not limited to, as shown in the following step S24c.

[0159] S24c. Perform enhancement processing on each shadow pixel point based on the second pixel variance, the neighborhood area of ​​each shadow pixel point, the first pixel variance and the pixel average value, so as to obtain the enhanced image after the enhancement processing; in the specific implementation, any shadow pixel point is taken as an example for explanation, and the enhancement process is shown in the first to fourth steps below.

[0160] Step 1: Calculate the product of the first pixel variance and the second pixel variance, and calculate the ratio between the product and the pixel average value to obtain the enhancement weight factor.

[0161] Step 2: For any shadow pixel, divide the grayscale value of each pixel in the neighborhood area corresponding to the shadow pixel by the enhancement weight factor to obtain several new grayscale values.

[0162] Step 3: Calculate the average value of several grayscale values ​​to obtain the enhancement parameters.

[0163] Step 4: Based on the grayscale value of any shadow pixel, the enhancement weight factor and the enhancement parameter, the following formula (8) is used to obtain the enhanced shadow pixel.

[0164]

[0165] In the above formula (8), r′ represents any shadow pixel after enhancement (referring to the enhanced grayscale value), r represents the grayscale value of any shadow pixel, δ represents the enhancement weight factor, τ represents the enhancement parameter, and μ represents a positive constant.

[0166] Through the aforementioned steps S24a to S24c, the enhancement of each shadow pixel can be completed, thereby restoring its actual information; based on this, an enhanced image can be obtained; then, face recognition can be performed on the enhanced image to obtain a face recognition result, and the process is shown in the following step S25.

[0167] S25. Perform face recognition processing on the enhanced image to obtain the face recognition result. In this embodiment, the face recognition can be performed by, but is not limited to, using a Y0LO neural network. Of course, other neural networks can also be used, and this is not limited to the aforementioned examples.

[0168] In this way, through the aforementioned steps S21 to S25, the facial recognition of the personnel at the entrance of the maintenance work area is completed, and then personnel verification can be performed to determine whether the personnel about to enter the maintenance work area are the same as the authorized personnel on the maintenance work ticket. The personnel verification process can be but is not limited to the following step S3.

[0169] S3. Based on the face recognition result, the maintenance time information and the maintenance personnel information, determine whether the person in the entrance surveillance video is an authorized person; in specific applications, personnel verification is performed based on both the entry time and the person himself, and the process can be but is not limited to the following steps S31 to S34.

[0170] S31. Obtain the application time for entering the maintenance work area, and use the application time as the arrival time of the personnel; in this embodiment, it has been explained above that an application button is provided at the access control, so the time when the personnel presses the application button is their arrival time; then, the arrival time of the personnel can be matched with the maintenance time in the maintenance work ticket to determine whether the maintenance personnel has arrived at the maintenance work area as required; wherein, the judgment process is shown in the following step S32.

[0171] S32. Determine whether the arrival time of the personnel matches the maintenance time information; in this embodiment, if the arrival time of the personnel matches the maintenance time in the maintenance time information, it means that the personnel arrived at the maintenance work area on time according to the maintenance time issued in the maintenance work ticket; then, personnel verification can be performed, and the process is shown in the following step S33.

[0172] S33. If so, determine whether the face recognition result matches the maintenance personnel information; in this embodiment, when the personnel arrival time does not match the maintenance time information, it means that the personnel has not arrived at the maintenance work area as required. At this time, an alarm is issued and a violation information is sent to the maintenance personnel and their superior terminal.

[0173] Similarly, when the arrival time of the personnel matches the maintenance time, and the face recognition result matches the maintenance personnel information, authorization to enter is performed, and the process is shown in the following step S34.

[0174] S34. If so, it is determined that the person in the entrance surveillance video is an authorized person, otherwise, it is determined that the person in the entrance surveillance video is an illegal person; in specific applications, if the personnel information in the personnel identification result matches the maintenance personnel information, it means that the person who is about to enter the maintenance work area is an authorized person (such as the person identified by face is Zhang San, and the authorized person is also Zhang San, and the work number is also the same, then he is determined to be an authorized person); at this time, the door can be opened, that is, the following step S4 is executed; at the same time, if the face recognition result does not match the maintenance personnel information, it means that the person who is about to enter the maintenance work area is an illegal person. At this time, an alarm prompt is given, and the access control is kept in a closed state.

[0175] Thus, through the aforementioned steps S31 to S34, the personnel verification during the maintenance work can be completed, and then, the door opening command can be issued to monitor the maintenance work. The process can be but is not limited to the following step S4.

[0176] S4. If yes, send a door opening command to the access control system of the maintenance operation area; after issuing the door opening command and allowing authorized personnel to enter the maintenance operation area, maintenance operation monitoring can be carried out, and the process is shown in the following steps S5 to S7.

[0177] S5. When it is detected that the authorized personnel enters the maintenance operation area, the operation video in the maintenance operation area is obtained in real time, and the equipment operation data of the operation equipment logged in by the authorized personnel is collected in real time; in this embodiment, it has been explained above that the maintenance work ticket contains the operation equipment used for this maintenance, and its corresponding legal login account; therefore, when the corresponding legal login account is detected and the operation equipment used for this maintenance is logged in, the monitoring video of the camera in the maintenance operation area can be obtained, thereby obtaining the operation video; at the same time, it is also necessary to collect the equipment operation data on the operation equipment used for this maintenance in real time, so as to subsequently identify maintenance operation violations based on the aforementioned collected video and operation data; wherein, the violation identification process can be but is not limited to the following step S6.

[0178] S6. Determine whether the maintenance operation of the authorized personnel is in compliance with regulations based on the operation video, equipment operation data and maintenance operation information; in specific implementation, the authorized personnel in the operation video are first subjected to behavior recognition processing to obtain the maintenance behavior of the authorized personnel; then, determine whether the maintenance behavior is a violation; wherein, the violation can be specifically set according to the actual maintenance scenario, which is not limited here. This embodiment lists smoking behavior, playing with mobile phones, etc.; wherein, after determining that the maintenance behavior is a violation, an alarm can be issued, and the broadcast in the maintenance operation area can be used to prevent the violation. Interrupt the behavior; at the same time, if the above-mentioned maintenance behavior does not constitute a violation, it is necessary to determine whether the equipment operation data is the maintenance operation data corresponding to the legal operation authority; that is, to determine whether the authorized personnel have performed an operation that exceeds the authority. If so, the maintenance operation of the authorized personnel is generated as a violation operation. At this time, it is necessary to block the equipment operation data and issue a violation alarm. Otherwise, the equipment operation data is sent to the bastion host cluster and forwarded to the equipment to be repaired based on the bastion host cluster, so as to realize the maintenance of the equipment to be repaired; wherein, the data forwarding and blocking operation process is shown in the following step S7.

[0179] S7. If yes, the device operation data is transmitted to the bastion host cluster, so that the device operation data is sent to the equipment to be repaired through the bastion host cluster; otherwise, the device operation data is blocked and a violation alarm is issued; in this embodiment, the operating equipment in the maintenance operation area is not directly connected to the equipment to be repaired, but is transferred through the operation and maintenance monitoring platform and the bastion host, so that the security of the data during maintenance can be improved; at the same time, for example, this embodiment also records the illegal equipment operation data and behavior, based on which, a violation database can be formed, and subsequent violation data statistics and analysis can be facilitated.

[0180] Therefore, through the maintenance control method based on artificial intelligence described in detail in the aforementioned steps S1 to S7, the present invention obtains the corresponding maintenance work information by parsing the maintenance work ticket, and then collects the entrance monitoring video of the maintenance operation area, and combines it with the aforementioned maintenance work information to perform personnel identification, that is, to determine whether the person who is about to enter the operation area is an authorized person. If so, the access control is opened; then, after the person enters the operation area, the person's operation video and the person's operation data on the operation equipment are collected in real time; finally, the operation violation identification can be performed based on the aforementioned maintenance operation information, as well as the collected operation video and operation data, and when it is identified that the person is operating in violation of the regulations, the blocking operation is performed and an alarm is issued; thus, the present invention realizes the digitization and automation of maintenance control, which is not only highly real-time and efficient, but also highly reliable compared to manual monitoring; thus, the method is very suitable for large-scale application and promotion in the field of operation and maintenance management.

[0181] like Figure 2 As shown, the second aspect of this embodiment provides a hardware system for implementing the artificial intelligence-based maintenance control method described in the first aspect of the embodiment, including:

[0182] The parsing unit is used to obtain the issued maintenance work ticket and parse the maintenance work ticket to obtain the maintenance personnel information, maintenance operation information and maintenance time information recorded in the maintenance work ticket.

[0183] The personnel verification unit is used to obtain the entrance surveillance video of the maintenance operation area when receiving the entry application of the maintenance operation area, and perform face recognition processing on the personnel in the entrance surveillance video to obtain the face recognition result.

[0184] A personnel verification unit is used to determine whether the person in the entrance monitoring video is an authorized person based on the face recognition result, the maintenance time information and the maintenance personnel information.

[0185] The personnel verification unit is also used to send a door opening instruction to the access control system of the maintenance operation area when it is determined that the person in the entrance monitoring video is an authorized person.

[0186] The operation monitoring unit is used to obtain the operation video in the maintenance operation area in real time when detecting that the authorized personnel enters the maintenance operation area, and to collect the equipment operation data of the operation equipment logged in by the authorized personnel in real time.

[0187] The operation monitoring unit is also used to determine whether the maintenance operation of the authorized personnel is compliant based on the operation video, equipment operation data and maintenance operation information.

[0188] The operation monitoring unit is also used to transmit the equipment operation data to the bastion host cluster when it is determined that the maintenance operation of the authorized personnel is compliant, so that the equipment operation data can be sent to the equipment to be repaired through the bastion host cluster; and when it is determined that the maintenance operation of the authorized personnel is illegal, block the equipment operation data and issue a violation alarm.

[0189] The working process, working details and technical effects of the system provided in this embodiment can be found in the second aspect of the embodiment and will not be described in detail here.

[0190] like Figure 3 As shown, the third aspect of this embodiment provides an artificial intelligence-based maintenance and control device. Taking the device as an electronic device as an example, it includes: a memory, a processor and a transceiver that are communicatively connected in sequence, wherein the memory is used to store computer programs, the transceiver is used to send and receive messages, and the processor is used to read the computer program and execute the artificial intelligence-based maintenance and control method as described in the first aspect of the embodiment.

[0191] For example, the memory may include, but is not limited to, random access memory (RAM), read-only memory (ROM), flash memory, first-in first-out memory (FIFO), and / or first-in last-out memory (FILO); specifically, the processor may include one or more processing cores, such as a quad-core processor, an octa-core processor, etc. The processor may be implemented in at least one hardware form selected from DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), and PLA (Programmable Logic Array). Furthermore, the processor may include a main processor and a coprocessor. The main processor is a processor for processing data in an awake state, also known as a CPU (Central Processing Unit); and the coprocessor is a low-power processor for processing data in a standby state.

[0192] In some embodiments, the processor may be integrated with a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content to be displayed on the display screen. For example, the processor may be, but is not limited to, a microprocessor of the STM32F105 series, a reduced instruction set computer (RISC) microprocessor, an X86 architecture processor, or a processor with an integrated embedded neural network processing unit (NPU); the transceiver may be, but is not limited to, a wireless fidelity (WIFI) wireless transceiver, a Bluetooth wireless transceiver, a general packet radio service technology (GPRS) wireless transceiver, a ZigBee protocol (a low-power local area network protocol based on the IEEE802.15.4 standard, ZigBee) wireless transceiver, a 3G transceiver, a 4G transceiver, and / or a 5G transceiver. In addition, the device may also include, but is not limited to, a power module, a display screen, and other necessary components.

[0193] The working process, working details and technical effects of the electronic device provided in this embodiment can be found in the first aspect of the embodiment and will not be described in detail here.

[0194] The fourth aspect of this embodiment provides a storage medium that stores instructions containing the artificial intelligence-based maintenance and control method described in the first aspect of the embodiment, that is, the storage medium stores instructions, and when the instructions are run on a computer, the artificial intelligence-based maintenance and control method described in the first aspect of the embodiment is executed.

[0195] The storage medium refers to a carrier for storing data, which may include but is not limited to a floppy disk, an optical disk, a hard disk, a flash memory, a USB flash drive and / or a memory stick, and the computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device.

[0196] The working process, working details and technical effects of the storage medium provided in this embodiment can be found in the first aspect of the embodiment and will not be described in detail here.

[0197] The fifth aspect of this embodiment provides a computer program product containing instructions, which, when executed on a computer, enables the computer to execute the artificial intelligence-based maintenance and control method as described in the first aspect of the embodiment, wherein the computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device.

[0198] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention shall be included in the scope of protection of the present invention.

Claims

1. A maintenance control method based on artificial intelligence, characterized in that: include: Obtain the issued maintenance work ticket and parse the maintenance work ticket to obtain the maintenance personnel information, maintenance operation information and maintenance time information recorded in the maintenance work ticket; Upon receiving an application for entry into the maintenance work area, obtaining a surveillance video of the entrance to the maintenance work area, and performing facial recognition processing on the personnel in the surveillance video to obtain a facial recognition result; Based on the face recognition result, the maintenance time information and the maintenance personnel information, determining whether the person in the entrance surveillance video is an authorized person; If yes, a door opening command is sent to the access control system of the maintenance operation area; When the authorized person is detected to have entered the maintenance operation area, the operation video in the maintenance operation area is acquired in real time, and the equipment operation data of the operating equipment logged in by the authorized person is collected in real time; Determine whether the maintenance work performed by the authorized personnel is in compliance with regulations based on the operation video, equipment operation data, and maintenance operation information; If so, the device operation data is transmitted to the bastion host cluster so that the device operation data is sent to the device to be repaired through the bastion host cluster; otherwise, the device operation data is blocked and a violation alarm is issued.

2. The method according to claim 1, characterized in that Performing facial recognition processing on the person in the entrance surveillance video to obtain a facial recognition result includes: Extracting a surveillance image containing a person's face from the entrance surveillance video; Performing denoising processing on the monitoring image to obtain a denoised image; Performing shadow detection processing on the denoised image to obtain shadow pixels in the denoised image; Based on the shadow pixels in the denoised image, performing image enhancement processing on the denoised image to obtain an enhanced image; Performing face recognition processing on the enhanced image to obtain the face recognition result.

3. The method according to claim 2, characterized in that Performing denoising on the surveillance image to obtain a denoised image includes: Performing wavelet decomposition processing on the monitoring image to obtain a low-frequency component image and a high-frequency component image; Using an improved filtering algorithm to filter the low-frequency component image to obtain a filtered low-frequency component image; Determining a wavelet threshold corresponding to the high-frequency component image, and performing denoising processing on the high-frequency component image based on the wavelet threshold to obtain a denoised high-frequency component image; An image fusion process is performed on the filtered low-frequency component image and the denoised high-frequency component image to obtain the denoised image after the image fusion process.

4. The method according to claim 3, characterized in that The improved filtering algorithm is used to filter the low-frequency component image to obtain a filtered low-frequency component image, including: For any pixel point in the low-frequency component image, obtaining a filtering window of the any pixel point; Using the grayscale values ​​of each pixel in the filtering window to form a grayscale set, and removing the maximum grayscale value and the minimum grayscale value from the grayscale set to obtain an updated grayscale set; Determining a noise threshold based on the updated grayscale set; Determining whether any pixel is a noise point according to the noise threshold; If so, the grayscale value of any pixel is updated to the noise threshold, and after all pixels in the low-frequency component image are polled, the filtered low-frequency component image is obtained.

5. The method according to claim 3, characterized in that Determine the wavelet threshold corresponding to the high-frequency component image, including: Obtain the decomposition scale of each high-frequency wavelet coefficient corresponding to the high-frequency component image, as well as the noise variance of the high-frequency component image; Calculating a first threshold value according to the noise variance; Calculating a second threshold at each decomposition scale based on the first threshold and the decomposition scale of each high-frequency wavelet coefficient; The wavelet threshold is composed by using the second threshold at each decomposition scale.

6. The method according to claim 2, characterized in that Performing shadow detection processing on the denoised image to obtain shadow pixels in the denoised image includes: Convert the denoised image into an HSV image to obtain the brightness and saturation of each pixel in the denoised image; Calculate the shadow feature value of each pixel based on the brightness and saturation of each pixel; Dividing the denoised image into a background image and a foreground image; Calculating a shadow feature threshold based on the background image, the foreground image, and the denoised image; The shadow pixel points in the denoised image are determined using the shadow feature threshold and the shadow feature value of each pixel point.

7. The method according to claim 2, characterized in that Based on the shadow pixels in the denoised image, image enhancement processing is performed on the denoised image to obtain an enhanced image, including: Calculating a first pixel variance and a pixel average of all shadow pixels in the denoised image, and a second pixel variance of all non-shadow pixels in the denoised image; Get the neighborhood area of ​​each shadow pixel; Each shadow pixel is enhanced based on the second pixel variance, the neighborhood area of ​​each shadow pixel, the first pixel variance, and the pixel average value, so as to obtain the enhanced image after the enhancement process.

8. The method according to claim 1, characterized in that Determining whether the person in the entrance surveillance video is an authorized person based on the face recognition result, the maintenance time information, and the maintenance personnel information includes: Obtaining the application time for entering the maintenance operation area, and using the application time as the personnel arrival time; Determine whether the personnel arrival time matches the maintenance time information; If so, determining whether the face recognition result matches the maintenance personnel information; If so, it is determined that the person in the entrance surveillance video is an authorized person; otherwise, it is determined that the person in the entrance surveillance video is an illegal person.

9. The method according to claim 1, characterized in that The maintenance operation information includes: maintenance operation data corresponding to illegal behaviors and legal operation permissions, wherein, based on the operation video, equipment operation data and maintenance operation information, judging whether the maintenance operation of the authorized personnel is compliant includes: Performing behavior recognition processing on the authorized personnel in the operation video to obtain the inspection behavior of the authorized personnel; Determine whether the inspection behavior is a violation; If not, determining whether the equipment operation data is maintenance operation data corresponding to the legal operation authority; If not, the maintenance work performed by the authorized personnel is deemed to be an illegal work.

10. The method according to claim 1, characterized in that Before obtaining the issued maintenance work ticket, the method further includes: Obtaining an initial work ticket, wherein the initial work ticket contains maintenance personnel information and a maintenance information summary; Setting the maintenance personnel in the maintenance personnel information as an authorized personnel, and allocating a legal login account for the operating equipment in the maintenance operation area to the authorized personnel; Generating maintenance time information and maintenance operation information according to the maintenance information summary; Recording the legal login account, the maintenance time information, and the maintenance operation information into the initial filled-in work ticket, so as to obtain the issued maintenance work ticket after recording; Accordingly, after completing the maintenance work of the equipment to be repaired, the method further includes: The login authority of the legal login account is canceled, and the device to be repaired is restored to clear all data during this repair process.

Images