Operation and maintenance early warning management method and system based on Internet of Things

Abstract

The invention relates to the field of the Internet of Things, in particular to an operation and maintenance early warning management method and system based on the Internet of Things, and the method comprises the steps: monitoring an early warning region through a camera, and obtaining a figure image; segmenting five-sense organ images in the figure image by adopting a first deep learning model; segmenting the five-sense organ image by using a second deep learning model, and performing character recognition; performing data entry and registration on the identified figure, obtaining an in-out time interval and times, and similarly, obtaining abnormal in-out time and times corresponding to the figure; identifying the monitoring picture information to obtain a blurred image, and analyzing the picture orientation, the camera connection state and the picture shielding information to obtain a monitoring picture abnormal index; if yes, a door lock opening instruction is generated, the door lock is controlled to be opened through the door lock opening instruction, and if not, early warning is conducted in time. The method has the advantages that the personnel identification accuracy can be improved, and early warning can be carried out on unauthorized personnel in time.

Description

Technical Field

[0001] The present invention relates to the field of the Internet of Things, and in particular to an Internet of Things-based operation and maintenance warning management method and system.

Background Art

[0002] In some fields where strict supervision is required, camera monitoring can only provide real-time monitoring. The traditional manual supervision method can no longer meet the requirements and cannot achieve the functions of timely warning and alarm, so economic losses are likely to occur.

[0003] The existing operation and maintenance warning management system can only send people to patrol and take pictures by cameras, cannot give warnings in time, has a low accuracy rate of personnel identification, and threatens the safety of items and data privacy.

[0004] Therefore, an Internet of Things-based operation and maintenance warning management method and system are proposed, which can improve the accuracy rate of personnel identification and give timely warnings to unauthorized personnel.

Summary of the Invention

[0005] The technical problem to be solved by the present invention is to overcome the deficiencies of the prior art and provide an Internet of Things-based operation and maintenance warning management method and system, which can improve the accuracy rate of personnel identification and give timely warnings to unauthorized personnel.

[0006] To solve the above technical problems, the present invention adopts the following technical solutions:

[0007] An Internet of Things-based operation and maintenance warning management method includes the following steps:

[0008] S1: Monitor the warning area through a camera to obtain a human image;

[0009] S2: Use a first deep learning model to segment the facial organ images in the human image;

[0010] S3: Use a second deep learning model to segment the facial organ images for personnel identification;

[0011] S4: Enter and register the identified personnel into the system, and obtain the entry and exit time intervals and frequencies. Similarly, obtain the abnormal entry and exit times and frequencies corresponding to the personnel;

[0012] S5: Identify the monitoring screen information to obtain a blurred image. The blurred image includes the screen orientation, camera connection status, and screen occlusion information. Analyze the screen orientation, camera connection status, and screen occlusion information to obtain a monitoring screen anomaly index, and analyze the screen orientation, camera connection status, and screen occlusion information to obtain a monitoring screen anomaly index;

[0013] S6: Determine whether the person image is on the authorized list. If it exists, update the door lock opening password stored in the storage module with the new door lock opening password. After the update is completed, generate a door lock opening instruction and control the door lock to open with the door lock opening instruction. If it does not exist, give an alarm in time.

[0014] Preferably, the facial feature images obtained by splitting the second deep learning model are split out from the person image to obtain the person image after the first split; use the third deep learning model to split the person image after the first split; multiply the split results obtained by the second deep learning model and the split results obtained by the third deep learning model with the original person image to obtain the final facial feature image split result.

[0015] Preferably, S4 is specifically as follows: Mark the user access records corresponding to the abnormal access addresses and abnormal access times as key operation records. The user access records include query, modification, and deletion. Set the system content to multiple content ranges, and each content range corresponds to a set of information. Match the user access records of each user with the multiple content ranges. When the user access record is different from the set information corresponding to the content range, mark it as an abnormal user operation, and count the number of abnormal user accesses corresponding to each user and record it as the total number of abnormal accesses.

[0016] Preferably, S5 is specifically as follows: Identify the connection status of each camera. When a connection abnormality occurs in the connection of the camera, mark the corresponding camera as an abnormal camera, count the number of abnormal cameras, and calculate the ratio of abnormal cameras by comparing the number of abnormal cameras with the total number of cameras;

[0017] Extract the high-quality picture area corresponding to each camera according to the monitoring area of each camera, use the area of the high-quality picture area of each camera as the corresponding effective picture area, obtain the picture occlusion area of each camera according to the picture occlusion information of each camera, and calculate the picture occlusion ratio by comparing the picture occlusion area with the effective picture area.

[0018] An Internet of Things operation and maintenance warning management system includes

[0019] An acquisition module that monitors the warning area through a camera and obtains a person image;

[0020] A refinement module that uses a first deep learning model to split out facial feature organ images from the person image;

[0021] An identification module that uses a second deep learning model to split the facial feature organ images for person identification;

[0022] Registration module, which enters the recognized person's data for registration, obtains the entry and exit time intervals and frequencies, and similarly, obtains the abnormal entry and exit times and frequencies corresponding to the person;

[0023] Abnormal index module, which recognizes the monitoring screen information to obtain a blurred image. The blurred image includes the screen orientation, camera connection status, and screen occlusion information. Analyze the screen orientation, camera connection status, and screen occlusion information to obtain the monitoring screen abnormal index, and analyze the screen orientation, camera connection status, and screen occlusion information to obtain the monitoring screen abnormal index;

[0024] Activation module, which determines whether the person image is on the authorized list. If so, update the door lock opening password stored in the storage module with the new door lock opening password. After the update is completed, generate a door lock opening instruction and control the door lock to open with the door lock opening instruction.

[0025] Preferably, it further includes at least one camera system, and each camera system is deployed in a communication base station for real-time monitoring of the power parameters and environmental parameters of the communication base station; the power parameters at least include: voltage value, current value, power value, power supply status, and battery power, and the environmental parameters at least include: temperature value, humidity value, water immersion status quantity, and smoke detection status quantity.

[0026] Preferably, each communication base station is used to collect the power parameters and environmental parameters monitored by all camera systems within the area where the base station is located, determine whether the power parameters and / or environmental parameters are abnormal, and after the power parameters and / or environmental parameters are abnormal, analyze the abnormal power parameters and / or environmental parameters, generate an abnormal report and a warning message, and upload the abnormal report and warning message to the main station.

[0027] Preferably, the main station is used for visualizing and storing the management of the abnormal reports and warning messages uploaded by each base station, and generating an inspection plan according to the abnormal reports and warning messages;

[0028] The inspection plan includes:

[0029] The number, location coordinates, and remaining maintenance duration of the communication base stations that need to be maintained;

[0030] The main station is also used for:

[0031] Obtain the current location coordinates of the mobile terminal of the maintenance personnel communicating with the main station, and generate the best maintenance route according to the current location coordinates of the mobile terminal, the number, location coordinates, remaining maintenance duration, and estimated maintenance duration of the communication base stations that need to be maintained.

[0032] Preferably, it further includes an early warning and feedback module. The early warning and feedback module is used to receive personnel abnormal feedback information, abnormal data packets, and system abnormal feedback information, generate corresponding personnel early warning information and system early warning information according to the personnel abnormal feedback information and the system abnormal feedback information, and isolate the abnormal data packets corresponding to the system abnormal feedback information.

[0033] A computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, it implements the steps of the above method.

[0034] Beneficial effects of adopting the present invention:

[0035] 1. The deep learning model can adopt the UNET model. The UNET model consists of an encoding sub-network, a decoding sub-network, and a skip connection layer. The network structure is similar to the shape of the letter 'U', so it is called the UNET network. The UNET network contains a total of 23 convolutional layers. There are 4 encoding operations and 4 decoding operations in the network. During the encoding process, two 3×3 convolutional operations and one 2×2 pooling operation are involved. This operation reduces the size of the pre-processed full-jaw digital panoramic tomogram image to half of the original image, while the number of features is twice that of the original.

[0036] During the decoding process, only two 3×3 convolutional operations are involved, doubling the size of the image reduced during the encoding process, making the final image size the same as the original image size and reducing the number of features by half. Finally, a 1×1 convolutional layer is added to the UNET network. The purpose of this is to map the features extracted during the encoding and decoding processes to the corresponding classes. The decoding process extracts high-dimensional features of the input image through continuous convolutional and pooling operations. The mutual mapping relationship between the encoding network and the decoding network is the characteristic of the UNET network. During decoding, the missing edge information is repaired by fusing the features of the encoding layer mapped to it, thereby improving the accuracy of predicting the edge information. The long connection located in the middle of the encoding and decoding of the UNET network is mainly used to copy the information of the input image and transfer it to downsampling, promoting the network to recover the information loss caused by downsampling.

[0037] Among them, the convolutional operation can be represented by a specific mathematical expression

[0038] Among them, y ij represents the convolution value of the i-th row and j-th column of the feature map image, U and V represent the height and width of the convolution kernel matrix, u and v are the coordinates of the convolution kernel matrix change, k uvIt represents the weight value corresponding to the \(u\)-th row and \(v\)-th column of the convolutional kernel, and \(i + u - 1, j + v - 1\) represents the pixel value corresponding to the \((i + u - 1)\)-th row and \((j + v - 1)\)-th column of the feature image.

[0039] 2. S4: Enter the recognized person into the data registration, and obtain the entry and exit time intervals and the number of times. Similarly, obtain the abnormal entry and exit times and the number of times corresponding to the person.

[0040] Specifically, S4 marks the user entry and exit records corresponding to the abnormal entry and exit addresses and abnormal entry and exit times as key operation records. The user entry and exit records include query, modification, and deletion. Set the system content to multiple content ranges, and each content range corresponds to a set information. Match the user entry and exit records of each user with multiple content ranges. When the user entry and exit record is different from the set information corresponding to the content range, mark it as an abnormal user operation, and count the number of abnormal user entry and exits corresponding to each user and record it as the total number of abnormal entries and exits.

[0041] 3. S5: Identify the monitoring screen information to obtain a blurred image. The blurred image includes the screen orientation, camera connection status, and screen occlusion information. Analyze the screen orientation, camera connection status, and screen occlusion information to obtain the monitoring screen anomaly index, and calculate the monitoring screen anomaly index. The formula is where \(kcs\), \(tkz\), and \(wgr\) represent the behavior influence value, the number influence value, and the abnormal stay value respectively; \(r1\), \(r2\), \(r3\) are all preset weight factors, and their values are 3.171, 2.237, and 1.979 respectively.

[0042] Specifically, S5 identifies the connection status of each camera. When the connection of the camera shows a connection anomaly, mark the corresponding camera as an abnormal camera, count the number of abnormal cameras, and calculate the ratio of abnormal cameras to the total number of cameras by comparison.

[0043] Extract the high-quality picture area corresponding to each camera according to the monitoring area of each camera, take the area of the high-quality picture area of each camera as the corresponding effective picture area, obtain the picture occlusion area of each camera according to the picture occlusion information of each camera, compare the picture occlusion area with the effective picture area to obtain the picture occlusion ratio, and perform normalization processing, expressed as where \(n\) represents the total number of operations, \(S\) n represents the picture occlusion area, \(q\) n represents the effective picture area, \(S\) n-1 represents the picture occlusion area of the previous frame, \(q\) n-1 represents the effective picture area of the previous frame.

[0044] These features and advantages of the present invention will be disclosed in detail in the following specific embodiments and the accompanying drawings.

Description of the Drawings

[0045] The present invention will be further described below in conjunction with the accompanying drawings:

[0046] Figure 1 It is a schematic flowchart of the first embodiment of the present invention.

Specific Embodiments

[0047] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts shall fall within the scope of protection of the present application.

[0048] The concepts involved in the present application will be described below first in conjunction with the accompanying drawings. It should be noted here that the descriptions of the following concepts are only for making the content of the present application easier to understand and do not represent a limitation on the protection scope of the present application.

[0049] Embodiment 1:

[0050] An Internet of Things operation and maintenance warning management method, as Figure 1 shown, includes the following steps:

[0051] S1: Monitor the warning area through a camera to obtain a human image;

[0052] S2: Use a first deep learning model to segment the facial organ images in the human image;

[0053] S3: Use a second deep learning model to segment the facial organ images for human recognition;

[0054] The deep learning model can adopt the UNET model. The UNET model consists of an encoding sub-network, a decoding sub-network and a skip connection layer. The network structure is similar to the shape of the letter U, so it is called the UNET network. This UNET network contains a total of 23 convolutional layers. There are 4 encoding operations and 4 decoding operations in the network. During the encoding process, two 3×3 convolutional operations and one 2×2 pooling operation are involved. This operation reduces the size of the preprocessed panoramic digital pantomograph image to half of the original image, while the number of features is twice that of the original.

[0055] During the decoding process, only two 3×3 convolutional operations are involved, which magnify the image reduced during the encoding process by two times, making the size of the finally obtained image the same as that of the original image and reducing the number of features by half. Finally, a 1×1 convolutional layer is added to the UNET network. The purpose of this is to map the features extracted during the encoding and decoding processes to the corresponding classes. The decoding process extracts the high-dimensional features of the input image through continuous convolutional operations and pooling operations. The mutual mapping relationship between the encoding network and the decoding network is the characteristic of the UNET network. During decoding, the missing edge information is repaired by fusing the features of the encoding layer mapped to it, thereby improving the accuracy of predicting the edge information. The long connection located in the middle of the encoding and decoding of the UNET network is mainly used to copy the information of the input image and transfer it to downsampling, promoting the network to recover the information loss caused by downsampling.

[0056] Among them, the convolutional operation can be represented by a specific mathematical expression

[0057] Among them, y ij represents the convolution value of the i-th row and j-th column of the feature map image, U and V represent the height and width of the convolution kernel matrix, u and v are the coordinates of the convolution kernel matrix change, and k uv represents the weight value corresponding to the u-th row and v-th column of the convolution kernel, and i+u-1,j+v-1 represents the pixel value corresponding to the (i+u-1)-th row and (j+v-1)-th column of the feature image.

[0058] S4: Enter the recognized person into the data registration, and obtain the entry and exit time intervals and times. Similarly, obtain the abnormal entry and exit times and times corresponding to the person;

[0059] Specifically, S4 marks the user entry and exit records corresponding to the abnormal entry and exit addresses and abnormal entry and exit times as key operation records. The user entry and exit records include query, modification, and deletion. The system content is set to multiple content ranges, and each content range corresponds to a set information. Match the user entry and exit records of each user with multiple content ranges. When the user entry and exit record is different from the set information corresponding to the content range, mark it as an abnormal user operation, and count the number of abnormal user entry and exits corresponding to each user and record it as the total number of abnormal entries and exits.

[0060] S5: Identify the monitoring screen information to obtain a blurred image. The blurred image includes the screen orientation, camera connection status, and screen occlusion information. Analyze the screen orientation, camera connection status, and screen occlusion information to obtain the monitoring screen anomaly index, analyze the screen orientation, camera connection status, and screen occlusion information to obtain the monitoring screen anomaly index, and calculate the monitoring screen anomaly index. The formula is Among them, kcs, tkz, and wgr respectively represent the behavior influence value, the frequency influence value, and the abnormal stay value; r1, r2, and r3 are all preset weight factors, and their values are 3.1, 2.2, and 1.9 respectively.

[0061] Specifically, S5 is to identify the connection status of each camera. When the connection of a camera is abnormal, mark the corresponding camera as an abnormal camera, count the number of abnormal cameras, and calculate the ratio of abnormal cameras by comparing the number of abnormal cameras with the total number of cameras.

[0062] Extract the high-quality picture area corresponding to each camera according to the monitoring area of each camera, use the area of the high-quality picture area of each camera as the corresponding effective picture area, obtain the picture occlusion area of each camera according to the picture occlusion information of each camera, compare the picture occlusion area with the effective picture area to obtain the picture occlusion ratio, and perform normalization processing, which is expressed as Among them, n represents the total number of operations, S n represents the picture occlusion area, q n represents the effective picture area, S n-1 represents the picture occlusion area of the previous frame, q n-1 represents the effective picture area of the previous frame.

[0063] S6: Determine whether the human image is in the authorized list. If it exists, update the door lock opening password stored in the storage module with a new door lock opening password. After the update is completed, generate a door lock opening instruction to control the door lock to open with the door lock opening instruction.

[0064] Each communication base station is provided with an access control system, and the access control system includes:

[0065] A door lock, an access control controller, a storage module, and a Bluetooth module. The Bluetooth module and the storage module are both electrically connected to the access control controller. The access control controller is used to control the opening or closing of the door lock. The storage module of the access control system of each communication base station stores a unique door lock opening password, and the main station stores the door lock opening password corresponding to each access control system.

[0066] After the maintenance personnel arrive at the communication base station that needs to be maintained according to the best maintenance route, the mobile terminal of the maintenance personnel is used for:

[0067] Generate a request to obtain the door lock opening password, and upload the request to obtain the door lock opening password by the mobile terminal of the maintenance personnel to the main station.

[0068] After the main station receives the request from the mobile terminal of the maintenance personnel to obtain the door lock opening password, the main station is also used for:

[0069] Based on the request for obtaining the door lock opening password, a new door lock opening password is randomly generated, and the door lock opening password stored locally at the main station is updated with the new door lock opening password;

[0070] The new door lock opening password and the door lock opening password stored locally at the main station before the update are packaged to obtain a password package;

[0071] The password package is sent to the mobile terminal of the maintenance personnel;

[0072] After receiving the password package, the mobile terminal is further configured to:

[0073] Establish a communication connection with the access control controller of the access control system of the communication base station through the Bluetooth module, and the mobile terminal sends the password package to the access control controller;

[0074] After receiving the password package, the access control controller is further configured to:

[0075] Parse the password package to obtain the new door lock opening password and the door lock opening password stored locally at the main station before the update.

[0076] S7: Receive personnel abnormal feedback information, abnormal data packets and system abnormal feedback information, generate corresponding personnel early warning information and system early warning information according to the personnel abnormal feedback information and the system abnormal feedback information, and isolate the abnormal data packets corresponding to the system abnormal feedback information.

[0077] The facial feature images obtained by splitting the second deep learning model are split out from the person image to obtain the person image after the first split; the person image after the first split is split using the third deep learning model; the split result obtained by the second deep learning model and the split result obtained by the third deep learning model are multiplied by the original person image to obtain the final facial feature image split result.

[0078] The second deep learning model specifically uses two 4*4 convolutional layers and a 4*4 pooling layer with a stride of 4 as the basic network structure, and the third deep learning model specifically uses two 5*5 convolutional layers and a 5*5 pooling layer with a stride of 5 as the basic network structure. A residual network is used as the network information feature extraction module, and an encoder-decoder structure is used to extract features and increase the resolution. Among them, the encoder consists of a downsampling module and an SE module composed of a residual network, and the decoder consists of an upsampling module composed of transposed convolution and a residual network and an SE module, so that the different scale information obtained by the encoder can be transmitted to the decoder to retain the features extracted by the encoders of different scales to the greatest extent, and the image clarity is higher.

[0079] An Internet of Things operation and maintenance early warning management system, including

[0080] An acquisition module that monitors a warning area through a camera and obtains a human image;

[0081] A refinement module that uses a first deep learning model to segment facial feature images from the human image;

[0082] An identification module that uses a second deep learning model to segment the facial feature images for human identification;

[0083] A registration module that enters and registers the identified person's data, obtains the entry and exit time intervals and frequencies, and similarly obtains the abnormal entry and exit times and frequencies corresponding to the person;

[0084] An abnormal index module that identifies the monitoring screen information to obtain a blurred image, where the blurred image includes the screen orientation, camera connection status, and screen occlusion information, analyzes the screen orientation, camera connection status, and screen occlusion information to obtain a monitoring screen abnormal index, and analyzes the screen orientation, camera connection status, and screen occlusion information to obtain a monitoring screen abnormal index;

[0085] An activation module that determines whether the human image is in the authorized list. If it exists, updates the door lock opening password stored in the storage module with a new door lock opening password. After the update is completed, generates a door lock opening instruction to control the door lock to open with the door lock opening instruction.

[0086] It further includes at least one camera system, and each camera system is deployed in a communication base station for real-time monitoring of the power parameters and environmental parameters of the communication base station; the power parameters at least include: voltage value, current value, power value, power supply status, and battery power, and the environmental parameters at least include: temperature value, humidity value, water immersion status quantity, and smoke detection status quantity.

[0087] Each communication base station is used to collect the power parameters and environmental parameters monitored by all camera systems within the area where the base station is located, determine whether the power parameters and / or environmental parameters are abnormal, and after the power parameters and / or environmental parameters are abnormal, analyze the abnormal power parameters and / or environmental parameters, generate an abnormal report and a warning message, and upload the abnormal report and warning message to the main station.

[0088] The main station is used to visualize and store and manage the abnormal reports and warning messages uploaded by each base station, and generate an inspection plan according to the abnormal reports and warning messages;

[0089] The inspection plan includes:

[0090] The number, location coordinates, and remaining maintenance duration of the communication base stations that need to be maintained;

[0091] The main station is further used for:

[0092] Obtain the current position coordinates of the mobile terminal of the maintenance personnel communicating with the main station, and generate an optimal maintenance route based on the current position coordinates of the mobile terminal, the number, position coordinates, remaining maintenance duration, and estimated maintenance duration of the communication base stations to be maintained.

[0093] It further includes an early warning and feedback module. The early warning and feedback module is used to receive personnel abnormal feedback information, abnormal data packets, and system abnormal feedback information, generate corresponding personnel early warning information and system early warning information according to the personnel abnormal feedback information and system abnormal feedback information, and isolate the abnormal data packets corresponding to the system abnormal feedback information.

[0094] A computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the steps of the above method.

[0095] The above is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Those skilled in the art should understand that the present invention includes but is not limited to the content described in the drawings and the above specific implementation manner. Any modification that does not deviate from the functional and structural principles of the present invention will be included in the scope of the claims.

[0096] Those of ordinary skill in the art can realize that the units (or modules, steps, etc., the same below) of each example described in combination with the embodiments disclosed in this article can be implemented by electronic hardware, computer software, or a combination of the two. To clearly illustrate the interchangeability of hardware and software, the components of each example have been generally described according to their functions in the above description. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0097] In the embodiments provided by the present invention, it should be understood that the division of units is only a logical function division, and there may be other division methods in actual implementation. For example, multiple units can be combined into one unit, one unit can be split into multiple units, or some features can be ignored, etc.

[0098] In addition, the functional units in each embodiment of the present invention can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above integrated units can be implemented in the form of hardware or in the form of software functional units.

[0099] When the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes: various media such as USB flash drives, read-only memories (ROMs), random access memories (RAMs), mobile hard disks, magnetic disks, or optical discs that can store program codes.

[0100] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the various embodiments of the present invention, and they should all be covered within the scope of the claims and the description of the present invention.

Claims

1. A method for early warning management of operation and maintenance based on the Internet of Things, characterized in that: The following steps are involved: S1: Monitor the warning area through the camera and obtain the image of the person; S2: Use the first deep learning model to segment facial organs from the person image; S3: Segmenting the facial organs image using a second deep learning model to perform person recognition; S4: Enter and register the data of the identified person, and obtain the entry and exit time interval and number of times. Similarly, obtain the abnormal entry and exit time and number of times corresponding to the person; S5: Identify the monitoring screen information to obtain a fuzzy image, wherein the fuzzy image includes screen orientation, camera connection status, and screen occlusion information, and analyze the screen orientation, camera connection status, and screen occlusion information to obtain a monitoring screen abnormality index, wherein the screen orientation, camera connection status, and screen occlusion information are analyzed to obtain a monitoring screen abnormality index; S6: Determine whether the person image is in the authorized list. If so, update the door lock opening password stored in the storage module with the new door lock opening password. After the update is completed, generate a door lock opening instruction, and control the door lock opening with the door lock opening instruction.

2. The method for early warning management of operation and maintenance based on the Internet of Things as claimed in claim 1, characterized in that: The facial features image obtained by the second deep learning model is segmented out of the character image to obtain the character image after the first segmentation; the character image after the first segmentation is segmented using the third deep learning model; the segmentation result obtained by the second deep learning model and the segmentation result obtained by the third deep learning model are multiplied with the original character image to obtain the final facial features image segmentation result.

3. The method for early warning management of operation and maintenance based on the Internet of Things as claimed in claim 1, characterized in that: S4 specifically includes marking the user entry and exit records corresponding to the abnormal entry and exit addresses and abnormal entry and exit times as key operation records. The user entry and exit records include query, modification and deletion. The system content is set to multiple content ranges, each content range corresponds to a setting information, and the user entry and exit records of each user are matched with the multiple content ranges. When the user entry and exit records are different from the setting information corresponding to the content range, it is marked as an abnormal user operation, and the number of abnormal user entry and exit times corresponding to each user is counted and recorded as the total number of abnormal entries and exits.

4. The method for early warning management of operation and maintenance based on the Internet of Things as claimed in claim 1, characterized in that: S5 specifically includes identifying the connection status of each camera, marking the corresponding camera as an abnormal camera when the connection of the camera is abnormal, counting the number of abnormal cameras, and comparing the number of abnormal cameras with the total number of cameras to calculate the abnormal camera ratio; The high-quality picture area corresponding to each camera is extracted according to the monitoring area of ​​each camera, and the area of ​​the high-quality picture area of ​​each camera is used as the corresponding effective picture area. The picture occlusion area of ​​each camera is obtained according to the picture occlusion information of each camera, and the picture occlusion area is compared with the effective picture area to obtain the picture occlusion ratio.

5. An IoT-based operation and maintenance early warning management system, characterized in that: include The acquisition module monitors the warning area through the camera and obtains the image of the person; The refinement module uses the first deep learning model to segment the facial organs in the human image; A recognition module, which uses a second deep learning model to segment the facial organs image and perform person recognition; The registration module records the data of the identified person and obtains the entry and exit time interval and number of times. Similarly, it obtains the abnormal entry and exit time and number of times corresponding to the person. An abnormality index module identifies the monitoring screen information to obtain a fuzzy image, wherein the fuzzy image includes screen orientation, camera connection status, and screen occlusion information, and analyzes the screen orientation, camera connection status, and screen occlusion information to obtain a monitoring screen abnormality index; wherein the screen orientation, camera connection status, and screen occlusion information are analyzed to obtain a monitoring screen abnormality index; The opening module determines whether the person image is in the authorized list. If so, the door lock opening password stored in the storage module is updated with the new door lock opening password. After the update is completed, a door lock opening instruction is generated, and the door lock opening instruction is used to control the door lock to open.

6. The IoT-based operation and maintenance early warning management system as claimed in claim 5, characterized in that: It also includes at least one camera system, each camera system is deployed in a communication base station and is used to monitor the power parameters and environmental parameters of the communication base station in real time; The power parameters include at least: voltage value, current value, power value, power supply state and battery power; the environmental parameters include at least: temperature value, humidity value, water immersion state quantity and smoke detection state quantity.

7. The IoT-based operation and maintenance early warning management system as claimed in claim 6, characterized in that: Each communication base station is used to collect the power parameters and environmental parameters monitored by all camera systems within the area where the base station is located, determine whether the power parameters and / or environmental parameters are abnormal, and if the power parameters and / or environmental parameters are abnormal, analyze the abnormal power parameters and / or environmental parameters, generate abnormal reports and warning information, and upload the abnormal reports and warning information to the main station.

8. The IoT-based operation and maintenance early warning management system as claimed in claim 7, characterized in that: The master station is used to visualize and store the abnormal reports and warning information uploaded by each base station, and generate an inspection plan based on the abnormal reports and warning information; The inspection plan includes: The number, location coordinates and remaining maintenance time of the communication base stations that need maintenance; The master station is also used for: The current location coordinates of the maintenance personnel's mobile terminal connected to the main station are obtained, and the optimal maintenance route is generated based on the current location coordinates of the mobile terminal and the number, location coordinates, remaining maintenance time and estimated maintenance time of the communication base stations that need to be maintained.

9. The IoT-based operation and maintenance early warning management system as claimed in claim 5, characterized in that: It also includes an early warning and feedback module, which is used to receive personnel abnormal feedback information, abnormal data packets and system abnormal feedback information, generate corresponding personnel early warning information and system early warning information according to the personnel abnormal feedback information and the system abnormal feedback information, and isolate and process the abnormal data packets corresponding to the system abnormal feedback information.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.

Images