Full-life-cycle intelligent supervision platform of camera device

Abstract

The invention discloses a full-life-cycle intelligent supervision platform of a camera device, relates to the technical field of intelligent management, and solves the problems that the running state of the camera device is not comprehensively monitored and the troubleshooting efficiency is low due to the fact that the full-life-cycle management of monitoring equipment is not coordinated. The full-life-cycle intelligent supervision platform comprises an equipment management module, a data management module, a monitoring analysis module, a remote cooperation module and a user service module. The hardware configuration information, the software log information and the deployment environment data of the camera device are specially processed through the data management module, the processing speed of different information data is increased, and the processing efficiency of mass information data is improved; the whole life cycle of each camera device is supervised and predicted through the intelligent diagnosis model, the whole life cycle supervision and prediction map is obtained, the troubleshooting efficiency of the camera device is improved, and the running state monitoring range of the camera device is expanded.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent management, and particularly to an intelligent supervision platform for the whole life cycle of a camera device. Background Art

[0002] In today's society, as the core component of a video surveillance system, camera devices have been widely used in multiple fields such as public security, traffic management, enterprise security, and smart home. With the popularization of camera devices and the continuous progress of technology, how to efficiently manage a large number and various types of camera devices to ensure the stable operation and performance optimization of their whole life cycle has become an urgent problem to be solved. In this context, the intelligent supervision platform for the whole life cycle of camera devices has emerged.

[0003] With the development of information technology, the functions of camera devices are becoming more and more powerful and the intelligent level is constantly improving, but this also makes the operation and maintenance management of camera devices more complex. The traditional management methods can no longer meet the efficient operation and maintenance needs of modern camera devices, and there is an urgent need for a new management means to improve the management level.

[0004] Patent No. CN2019112022663 discloses an information monitoring system based on the whole life cycle of a device, including a sensing system, an information management system, and at least one service processing system. By pre-configuring a message server, the sensing system, the information management system, and each service processing system are communicatively connected to each other through the message server to realize data information synchronization between the information management system and each service processing system, and to realize the unified management of the information of the whole life cycle of the device in the Internet of Things by the information management system. The above system can enable the information management system to integrate and manage the information of a series of operation nodes in the whole life cycle of the device, including factory, installation, alarm, maintenance, and uninstallation, so that all information of the device at the current time node can be obtained traceably during the process of device monitoring, and the efficiency of users in monitoring and managing the device is improved.

[0005] Patent No. CN2022113690796 discloses a device full - life - cycle management and monitoring system, including: a pre - operation management subsystem of the device, configured to supervise information related to the device during the first cycle period; an in - operation management subsystem of the device, configured to monitor information related to the device during the second cycle period; a device maintenance management subsystem, respectively connected to the pre - operation management subsystem and the in - operation management subsystem of the device, configured to supervise operation and maintenance information related to the device during the third cycle period, and cooperate with the pre - operation management subsystem and the in - operation management subsystem of the device to achieve closed - loop control of the full life cycle of the device. Through the technical solution of the above - mentioned invention, each subsystem in the system is used to supervise different periods of the entire life cycle of the device, realizing closed - loop control of the full life cycle of the device and meeting the actual use requirements.

[0006] However, although the above patent solves the problem of full - life - cycle management of monitoring devices, there is a lack of unified management and coordination in the procurement, installation, maintenance, upgrade, and scrapping of camera devices, resulting in resource waste and management chaos; moreover, the monitoring of the operating status and fault troubleshooting of camera devices mainly rely on manual work, with low efficiency and easy to make mistakes. Summary of the Invention

[0007] The purpose of the present invention is to provide an intelligent supervision platform for the full life cycle of camera devices, which can use artificial intelligence technology to specially process the hardware configuration information, software log information, and deployment environment data of camera devices through a data management module, improve the processing speed of different information data, and enhance the processing efficiency of massive information data; use an intelligent diagnosis model to supervise and predict the full life cycle of each camera device, obtain a full - life - cycle supervision and prediction map, improve the fault troubleshooting efficiency of camera devices, and expand the monitoring range of the operating status of camera devices.

[0008] The present invention uses the following technical solutions:

[0009] An intelligent supervision platform for the full life cycle of camera devices, including a device management module, a data management module, a monitoring and analysis module, a remote collaboration module, and a user service module; among them,

[0010] The device management module is used to manage the operation and maintenance process of camera devices, and divide all the addresses where camera devices are installed into several device management areas and allocate several maintenance personnel according to the contract order; the operation and maintenance process includes procurement, installation, configuration, upgrade, and maintenance;

[0011] The data management module is used to block, clean, classify, associate, encrypt, and store the hardware configuration information, software log information, and deployment environment data of several camera devices in each device management area, and obtain encrypted information data blocks;

[0012] The monitoring and analysis module is used to supervise and predict the entire life cycle of each camera device based on the encrypted information data block, hardware configuration information, software log information, and deployment environment data through a threshold detection algorithm or an intelligent diagnosis model, and obtain an entire life cycle supervision and prediction map;

[0013] The remote collaboration module is used to monitor the operating status and conduct networked collaborative maintenance on several camera devices within each device management area according to the entire life cycle supervision and prediction map;

[0014] The operation and maintenance service module is used to provide a supervision and feedback channel for the operation and maintenance process and progress for customers, administrators, and maintainers within each device management area during the entire life cycle of the camera device.

[0015] Preferably, the device management module calculates the procurement quantity of camera devices based on the inventory redundancy quantity, contract order quantity, and iterative replacement quantity, and selects a manufacturing manufacturer for procurement according to the procurement quantity; the device management module assigns the number of contract orders to be fulfilled and the number of camera devices to be installed to each maintainer; the content of the contract order includes customer information, project overview, project funds, construction period, business requirements, installation and commissioning, and after-sales service; the customer information includes the customer name code and the customer area code; after the maintainer installs and connects the camera device to the network and runs it, the maintainer configures the functions of the camera device according to the business requirements; the business requirements include remote monitoring, video storage, playback alarm, high-definition night vision, motion detection, and / or privacy protection; the device management module uses the SSH protocol to obtain the device information of the camera device that has completed the function configuration; the device information includes the device serial number, device model, hardware configuration, firmware version, software version, and physical address; the device management module regularly queries the firmware version and software version of each camera device. If a new version appears in the firmware version and / or software version, the remote differential algorithm is used to upgrade and maintain the camera device; the device management module uses the clustering algorithm to perform a primary clustering on all camera devices according to the customer name code to form a customer list; then, the customer list is clustered again according to the customer area code to construct a device management area and assign several maintainers.

[0016] Preferably, the data management module collects the hardware configuration information of the imaging device by using a remote programming interface or network scanning technology; the hardware configuration information includes cameras, lenses, pan-tilt heads, listeners, alarm detectors, multi-functional decoders, protective covers, server performance, and fan speed; the data management module extracts the software log information of the imaging device by using the SSL / TLS protocol; the software log information includes operation logs, system logs, error logs, and screenshots; the data management module collects the deployment environment data of the imaging device by using the built-in sensors or external sensors of the imaging device; the deployment environment data includes weather, temperature, humidity, light, and wind speed; the data management module first divides the hardware configuration information, software log information, and deployment environment data into several simulated data blocks according to the data collection time by using a data chunking algorithm, and then recombines the information data within each simulated data block according to the customer name code to obtain a recombined information data block.

[0017] Preferably, the data management module determines the data collection interval of all the information data in the recombined information data block: if the collection time interval between a piece of information data and the adjacent information data is less than the preset time difference, and this piece of information data is the same as any adjacent information data, then this piece of information data is determined to be duplicate data, and the duplicate data is cleaned by using a data deduplication algorithm to remove redundant data;

[0018] If the collection time interval between a piece of information data and the adjacent information data is less than the preset time difference, but this piece of information data is different from any adjacent information data, then this piece of information data is determined to be noise data, and the noise data is cleaned according to the business requirements by using an information filtering algorithm to smooth the data fluctuation;

[0019] If the collection time interval between a piece of information data and the adjacent information data is greater than the preset time difference, and this piece of information data is different from any adjacent information data, then this piece of information data is determined to be a normal information data fluctuation, and the recombined information data block is checked for consistency by using a data consistency algorithm to obtain several simplified information data blocks;

[0020] If the collection time interval between a piece of information data and the adjacent information data is greater than the preset time difference, but this piece of information data is different from any adjacent information data, then this piece of information data is determined to be a normal information data change, and the recombined information data block is checked for consistency by using a data consistency algorithm to obtain several simplified information data blocks;

[0021] The data management module re-classifies the information data within the abbreviated information data block according to the customer address code using a data classification algorithm to obtain a hardware aggregation block, a software aggregation block, and an environment aggregation block; the data management module associates the hardware aggregation block, the software aggregation block, and the environment aggregation block of each camera device using a data association algorithm to obtain a number of associated information data blocks; the data management module performs nested-level encryption on the associated information data blocks according to the information sensitivity using a data encryption algorithm to obtain encrypted information data blocks; and stores the encrypted information data blocks, the hardware configuration information, the software log information, and the deployment environment data using a meta-database.

[0022] Preferably, the monitoring and analysis module uses the information decomposition layer of the intelligent diagnosis model to perform dimensionality reduction and desensitization on the stored encrypted information data blocks to obtain a non-sensitive hardware information block, a non-sensitive software information block, and a non-sensitive environment data block; at the same time, the extraction and verification layer of the intelligent diagnosis model verifies and aggregates the stored hardware configuration information, software log information, and deployment environment data to obtain a hardware verification set, a software verification set, and an environment verification set; then the feature extraction layer of the intelligent diagnosis model uses the bacterial foraging optimization algorithm to extract features from the non-sensitive hardware information block, the non-sensitive software information block, and the non-sensitive environment data block to obtain a group of key feature values; subsequently, the iterative training layer of the intelligent diagnosis model uses the firefly algorithm to perform several iterations on the group of key feature values in combination with the hardware verification set, the software verification set, and the environment verification set to obtain a group feature weight matrix; finally, the recognition and construction layer of the intelligent diagnosis model uses a graph construction algorithm to update and optimize the group feature weight matrix and construct a full-life cycle supervision and prediction graph.

[0023] Preferably, the information decomposition layer first performs dimensionality reduction decomposition on the stored encrypted information data blocks to obtain a dimensionality-reduced hardware information block, a dimensionality-reduced software information block, and a dimensionality-reduced environment data block:

[0024]

[0025] Among them, KD represents the encrypted information data block, MDS -1 represents the inverse multi-dimensional decomposition function, PCA -1 represents the inverse principal component analysis function, α 1 represents the hardware information screening factor, Q 1 represents the dimensionality-reduced hardware information block, represents the multiple vector sum, NMF -1 represents the inverse non-negative matrix decomposition function, α 2 represents the software information screening factor, Q 2 represents the dimensionality-reduced software information block, LLE -1 represents the inverse local linear embedding function, α 3 represents the environment data screening factor, Q 3 represents the dimensionality-reduced environment data block;

[0026] The information decomposition layer then performs desensitization processing on the dimension-reduced hardware information block, dimension-reduced software information block, and dimension-reduced environmental data block formed by dimension reduction decomposition respectively to obtain a non-sensitive hardware information block, a non-sensitive software information block, and a non-sensitive environmental data block:

[0027] [NH, NS, NE] = [Hash(Q 1 ), Fifo(Q 2 ), Rrip(Q 3 )] (2)

[0028] Among them, NH represents the non-sensitive hardware information block, NS represents the non-sensitive software information block, NE represents the non-sensitive environmental data block, Hash represents the hash algorithm, Fifo represents the first-in first-out algorithm, and Rrip represents the interval replacement algorithm.

[0029] Preferably, the feature extraction layer first initializes the bacterial population and randomly generates several groups of sliding windows with different scales; then, through 2 3×3 convolutional layers with a stride of 2, 1 5×5 convolutional layer with a stride of 1, and 2 batch normalization layers, the non-sensitive hardware information block, non-sensitive software information block, and non-sensitive environmental data block are used to extract information with sliding windows of different scales to obtain multi-level feature information values; subsequently, the multi-level feature information values are compared with a preset information threshold: if the multi-level feature information value is greater than or equal to the preset information threshold, the running mode of this sliding window is swimming, and reproduction, aggregation, and migration are performed within the non-sensitive hardware information block, non-sensitive software information block, and non-sensitive environmental data block until the information extraction is completed, and then the population key feature value is obtained; if the multi-level feature information value is less than the preset information threshold, the running mode of this sliding window is tumbling, and the information extracted by this sliding window in each information block is aggregated and eliminated, and an information extraction influence factor is generated. The next group of sliding windows with different scales is combined with the information extraction influence factor to extract information from each information block to obtain the population key feature value.

[0030] Preferably, the iterative training layer first randomly generates several groups of fireflies, and each group of fireflies represents a group of candidate weight matrices; then each group of candidate weight matrices is used as the initial training weights of two parallel 3×3 convolutional layers, two CBS blocks, two residual blocks, one inverted residual block, two InceptionV3 blocks, two batch normalization layers, and a Silu activation function; the CBS block includes three 3×3 convolutional layers with a stride of 2, two batch normalization layers, and one Softmax activation function; then, the initial training weights are used to iterate the population key feature values several times, and further evaluate the real-time fitness of each group of initial training weights; compare the real-time fitness with the preset fitness: if the real-time fitness is less than the preset fitness, calculate the ratio of the real-time fitness to the preset fitness, and calculate it with the next group of candidate weight matrices to obtain a corrected weight matrix, and then use the corrected weight matrix to iterate the network structure again until the last group of candidate weight matrices; if the real-time fitness is greater than or equal to the preset fitness, adaptively update the learning rate of the current initial training weights, and perform an update calculation on all candidate weight matrices to obtain a population feature weight matrix; finally, use the hardware test set, software test set, and environment test set to update and optimize the population feature weight matrix.

[0031] Preferably, the operation and maintenance service module includes a user side, a management side, and an execution side; customers can view the operation and maintenance nodes, repair and replacement progress, and the operating status of each camera device in the operation and maintenance service through the user side; administrators can allocate operation and maintenance service orders to maintainers according to the device management area through the management side, and monitor the operation and maintenance progress of the maintainers; maintainers receive the operation and maintenance service orders through the execution side and go to the deployment site of the camera device to perform operation and maintenance services. The maintainers feed back the start time of operation and maintenance, the progress of maintenance and repair, and the end time to the management side in the form of images or texts for storage.

[0032] Preferably, the operation and maintenance service module further includes a re-inspection side; the re-inspection side uses the rule decision algorithm and the image recognition algorithm to monitor the operation and maintenance processes of all camera devices in each device management area.

[0033] The present invention specially processes the hardware configuration information, software log information, and deployment environment data of the camera device through the data management module, improving the processing speed of different information data and the processing efficiency of massive information data; through the intelligent diagnosis model, it monitors and predicts the entire life cycle of each camera device to obtain an entire life cycle monitoring and prediction map, improving the fault troubleshooting efficiency of the camera device and expanding the monitoring range of the operating status of the camera device. Description of the Drawings

[0034] To more clearly illustrate the technical solutions in the embodiments of the present invention or related technologies, the following will briefly introduce the drawings required for use in the description of the embodiments or related technologies. Obviously, the drawings in the following description are only the embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on the provided drawings.

[0035] Figure 1 It is a schematic diagram of the principle of the full - life - cycle intelligent supervision platform;

[0036] Figure 2 It is a schematic diagram of the principle of the intelligent diagnosis model. Detailed implementation manners

[0037] The present invention will be described in detail below in conjunction with the drawings and embodiments:

[0038] As Figure 1 - Figure 2 shown, a full - life - cycle intelligent supervision platform for a camera device according to the present invention includes a device management module, a data management module, a monitoring and analysis module, a remote collaboration module, and a user service module; among them,

[0039] The device management module is used to manage the operation and maintenance process of the camera device, and divide all the addresses where the camera devices are installed into several device management areas and allocate several maintenance personnel according to the contract order; the operation and maintenance process includes procurement, installation, configuration, upgrade, and maintenance;

[0040] The data management module is used to block, clean, classify, associate, encrypt, and store the hardware configuration information, software log information, and deployment environment data of several camera devices in each device management area, and obtain encrypted information data blocks;

[0041] The monitoring and analysis module is used to supervise and predict the full life cycle of each camera device through a threshold detection algorithm or an intelligent diagnosis model according to the encrypted information data blocks, hardware configuration information, software log information, and deployment environment data, and obtain a full - life - cycle supervision and prediction map;

[0042] The remote collaboration module is used to monitor the running status and perform networked collaborative maintenance on several camera devices in each device management area according to the full - life - cycle supervision and prediction map;

[0043] The operation and maintenance service module is used to provide a way to monitor and feedback the operation and maintenance process and progress to customers, administrators, and maintenance personnel in each device management area during the full life cycle of the camera device;

[0044] In the present invention, the device management module calculates the procurement quantity of the camera device according to the inventory redundancy quantity, the contract order quantity, and the iterative replacement quantity, and selects a manufacturer for procurement based on the procurement quantity; the device management module assigns the quantity of contract orders to be fulfilled and the number of camera devices to be installed to each maintenance staff; the content of the contract order includes customer information, project overview, project payment, construction period, business requirements, installation and commissioning, and after-sales service; the customer information includes the customer name code and the customer area code; after the maintenance staff installs and network-connects the camera device and runs it, the function of the camera device is configured according to the business requirements; the business requirements include remote monitoring, video storage, playback alarm, high-definition night vision, motion detection, and / or privacy protection; the device management module uses the SSH protocol to obtain the device information of the camera device that has completed the function configuration; the device information includes the device serial number, device model, hardware configuration, firmware version, software version, and physical address; the device management module regularly queries and updates the firmware version and software version of each camera device. If a new version appears in the firmware version and / or software version, the remote differential algorithm is used to upgrade and maintain the camera device; the device management module uses the clustering algorithm to perform a primary clustering of all camera devices according to the customer name code to form a customer list; then, according to the customer area code, the customer list is clustered again to construct a device management area and assign several maintenance staff.

[0045] In the present invention, the data management module collects the hardware configuration information of the camera device by using a remote programming interface or network scanning technology; the hardware configuration information includes cameras, lenses, pan-tilt heads, listeners, alarm detectors, multi-functional decoders, protective covers, server performance, and fan speed; the data management module extracts the software log information of the camera device by using the SSL / TLS protocol; the software log information includes operation logs, system logs, error logs, and screenshots; the data management module collects the deployment environment data of the camera device by using the built-in sensors or external sensors of the camera device; the deployment environment data includes weather, temperature, humidity, light, and wind speed; the data management module uses the data chunking algorithm to first divide the hardware configuration information, software log information, and deployment environment data into several simulated data chunks according to the data collection time, and then recombines the information data including the hardware configuration information, software log information, and deployment environment data in each simulated data chunk according to the customer name code to obtain the recombined information data;

[0046] The data management module determines the data collection interval of all information data in the recombined information data block: if the collection time interval between a certain information data and the adjacent information data is less than the preset time difference, and this information data is the same as any adjacent information data, then this information data is determined to be duplicate data, and the data deduplication algorithm is used to clean the duplicate data and remove the redundant data;

[0047] If the acquisition time interval between a piece of information data and adjacent information data is less than a preset time difference, and this information data is different from any adjacent information data, then this information data is determined to be noise data, and the information filtering algorithm is used to clean the noise data according to business requirements to smooth the data fluctuation;

[0048] In this embodiment, different business requirements have different requirements for noise data: For remote monitoring, the requirements for noise data mainly focus on a low noise level to ensure the clarity and accuracy of the monitoring screen. For example, monitoring devices need to have good anti-interference capabilities to avoid interference from environmental noises (such as wind noise, vehicle noise, etc.) on the video signal.

[0049] Video storage needs to ensure the high fidelity and integrity of noise data. During data acquisition and transmission, the loss or distortion of noise data should be avoided to ensure the traceability and reliability of the video content.

[0050] For playback alarm, the real-time and accuracy requirements for noise data are relatively high. The system needs to be able to quickly identify and analyze noise data to ensure that the alarm is triggered timely and accurately.

[0051] For high-definition night vision, the requirements for noise data are mainly reflected in the image noise reduction ability. Due to insufficient light in the night vision environment, noises (such as sensor noise, thermal noise, etc.) will significantly affect the image quality;

[0052] For motion detection, the requirement for noise data is a low false alarm rate. The system needs to be able to accurately distinguish environmental noise and target movement signals to avoid false alarms caused by noise interference;

[0053] For privacy protection, the requirements for noise data are reflected in data desensitization and de-identification. By adding noise or perturbing data (such as Laplace noise, Gaussian noise), privacy can be protected without revealing sensitive information.

[0054] If the acquisition time interval between a piece of information data and adjacent information data is greater than a preset time difference, and this information data is different from any adjacent information data, then this information data is determined to be a normal information data fluctuation, and the data consistency algorithm is used to perform a consistency check on the reorganized information data block to obtain several simplified information data blocks;

[0055] If the acquisition time interval between a piece of information data and adjacent information data is greater than a preset time difference, but this information data is different from any adjacent information data, then this information data is determined to be a normal information data change, and the data consistency algorithm is used to perform a consistency check on the reorganized information data block to obtain several simplified information data blocks;

[0056] The data management module re-classifies the information data in the abbreviated information data block according to the customer address code by using the data classification algorithm, and obtains a hardware aggregation block, a software aggregation block, and an environment aggregation block; the data management module associates the hardware aggregation block, the software aggregation block, and the environment aggregation block of each camera device by using the data association algorithm, and obtains several associated information data blocks; the data management module uses the data encryption algorithm to perform nested-level encryption on the associated information data blocks according to the information sensitivity, and obtains encrypted information data blocks; and stores the encrypted information data blocks, the hardware configuration information, the software log information, and the deployment environment data by using the meta-database.

[0057] In this embodiment, the information sensitivity is the potential harm degree that information leakage or illegal use may cause to an individual or an organization;

[0058] In the present invention, the monitoring and analysis module uses the information decomposition layer of the intelligent diagnosis model to perform dimensionality reduction and desensitization on the stored encrypted information data blocks, and obtains a non-sensitive hardware information block, a non-sensitive software information block, and a non-sensitive environment data block;

[0059] In this embodiment, the information decomposition layer first performs dimensionality reduction decomposition on the stored encrypted information data blocks, and obtains a dimensionality-reduced hardware information block, a dimensionality-reduced software information block, and a dimensionality-reduced environment data block:

[0060]

[0061] Among them, KD represents the encrypted information data block, MDS -1 represents the inverse multi-dimensional decomposition function, PCA -1 represents the inverse principal component analysis function, α 1 represents the hardware information screening factor, Q 1 represents the dimensionality-reduced hardware information block, represents the multiple vector sum, NMF -1 represents the inverse non-negative matrix decomposition function, α 2 represents the software information screening factor, Q 2 represents the dimensionality-reduced software information block, LLE -1 represents the inverse local linear embedding function, α 3 represents the environment data screening factor, Q 3 represents the dimensionality-reduced environment data block;

[0062] The information decomposition layer then performs desensitization processing on the dimensionality-reduced hardware information block, the dimensionality-reduced software information block, and the dimensionality-reduced environment data block formed by the dimensionality reduction decomposition respectively, and obtains a non-sensitive hardware information block, a non-sensitive software information block, and a non-sensitive environment data block:

[0063] [NH,NS,NE]=[Hash(Q 1 ),Fifo(Q 2 ),Rrip(Q3 )] (2)

[0064] Among them, NH represents the non-sensitive hardware information block, NS represents the non-sensitive software information block, NE represents the non-sensitive environment data block, Hash represents the hash algorithm, Fifo represents the first-in-first-out algorithm, and Rrip represents the interval replacement algorithm;

[0065] In this embodiment, the principle of the hash algorithm: sensitive data (such as ID card numbers, phone numbers, etc.) is converted into a hash value through a hash function; the hash function is one-way, that is, the original data cannot be deduced from the hash value;

[0066] The principle of the first-in-first-out algorithm: sensitive data is stored in a queue in the order of entry; when sensitive data needs to be accessed, the earliest entered data is taken out of the queue for desensitization processing; the desensitized data can be used for display or analysis, while the original sensitive data is hidden;

[0067] The principle of the interval replacement algorithm: define an interval value, which represents the interval of characters or numbers to be replaced in sensitive data; traverse the sensitive data, and replace the characters or numbers at a certain interval; the replaced characters or numbers can be randomly generated or preset.

[0068] At the same time, the extraction and verification layer of the intelligent diagnosis model verifies and aggregates the stored hardware configuration information, software log information, and deployment environment data to obtain a hardware verification set, a software verification set, and an environment verification set;

[0069] Then, the feature extraction layer of the intelligent diagnosis model uses the bacterial foraging optimization algorithm to extract features from the non-sensitive hardware information block, the non-sensitive software information block, and the non-sensitive environment data block to obtain the group key feature values;

[0070] In this embodiment, the feature extraction layer first initializes the bacterial population and randomly generates several groups of sliding windows with different scales. Then, through two 3×3 convolutional layers with a stride of 2, one 5×5 convolutional layer with a stride of 1, and two batch normalization layers, information extraction is performed on the insensitive hardware information block, insensitive software information block, and insensitive environmental data block using sliding windows with different scales to obtain multi-level feature information values. Subsequently, the multi-level feature information values are compared with a preset information threshold: if the multi-level feature information value is greater than or equal to the preset information threshold, the running mode of this sliding window is swimming, and reproduction, aggregation, and migration are performed within the insensitive hardware information block, insensitive software information block, and insensitive environmental data block until the information extraction is completed, thereby obtaining the population key feature value; if the multi-level feature information value is less than the preset information threshold, the running mode of this sliding window is tumbling, and the information extracted by this sliding window in each information block is aggregated and eliminated, and an information extraction influence factor is generated. The next group of sliding windows with different scales is used in combination with the information extraction influence factor to perform information extraction on each information block to obtain the population key feature value;

[0071] Subsequently, the iterative training layer of the intelligent diagnosis model uses the firefly algorithm to perform several iterations on the population key feature values in combination with the hardware test set, software test set, and environmental test set to obtain the population feature weight matrix;

[0072] In this embodiment, the iterative training layer first randomly generates several groups of fireflies, and each group of fireflies represents a group of candidate weight matrices. Then, each group of candidate weight matrices is used as the initial training weights of two parallel 3×3 convolutional layers, two CBS blocks, two residual blocks, one inverted residual block, two InceptionV3 blocks, two batch normalization layers, and the Silu activation function. The CBS block includes three 3×3 convolutional layers with a stride of 2, two batch normalization layers, and one Softmax activation function. Then, several iterations are performed on the population key feature values using the initial training weights to evaluate the real-time fitness of each group of initial training weights. The real-time fitness is compared with the preset fitness: if the real-time fitness is less than the preset fitness, the ratio value of the real-time fitness to the preset fitness is calculated, and the ratio value is calculated with the next group of candidate weight matrices to obtain the corrected weight matrix, and then the network structure is iterated again using the corrected weight matrix until the last group of candidate weight matrices; if the real-time fitness is greater than or equal to the preset fitness, the learning rate of the current initial training weight is adaptively updated, and all candidate weight matrices are updated and calculated to obtain the population feature weight matrix; finally, the population feature weight matrix is updated and optimized using the hardware test set, software test set, and environmental test set;

[0073] Finally, the recognition and construction layer of the intelligent diagnosis model uses the atlas construction algorithm to update and optimize the population feature weight matrix and construct the full-life cycle supervision and prediction atlas.

[0074] In the present invention, the operation and maintenance service module includes a user side, a management side, an execution side, and a re-inspection side; the customer can view the operation and maintenance nodes, repair and replacement progress, and the operation status of each camera device in the operation and maintenance service through the user side; the administrator assigns operation and maintenance service orders to the maintainers according to the device management area through the management side, and monitors the operation and maintenance progress of the maintainers; the maintainer receives the operation and maintenance service order through the execution side and goes to the deployment site of the camera device to perform the operation and maintenance service, and the maintainer feeds back the operation start time, maintenance progress, and end time to the management side in the form of images or texts for storage; the re-inspection side uses rule decision algorithms and image recognition algorithms to monitor the operation and maintenance processes of all camera devices in each device management area.

[0075] Embodiment:

[0076] The device management module calculates the procurement quantity of camera devices according to the inventory redundancy quantity, contract order quantity, and iterative replacement quantity, and selects a manufacturing manufacturer for procurement based on the procurement quantity; the device management module assigns the number of contract orders to be fulfilled and the number of camera devices to be installed to each maintainer; the content of the contract order includes customer information, project overview, project payment, construction period, business requirements, installation and commissioning, and after-sales service; the customer information includes customer name code and customer area code.

[0077] After the maintainer installs and connects the camera device to the network and runs it, the maintainer configures the functions of the camera device according to the business requirements; the business requirements include remote monitoring, video storage, playback alarm, high-definition night vision, motion detection, and / or privacy protection.

[0078] The device management module uses the SSH protocol to obtain the device information of the camera device that has completed function configuration; the device information includes device serial number, device model, hardware configuration, firmware version, software version, and physical address.

[0079] The device management module regularly queries the firmware version and software version of each camera device. If a new version appears in the firmware version and / or software version, the remote differential algorithm is used to upgrade and maintain the camera device.

[0080] The device management module uses the clustering algorithm to perform a primary clustering of all camera devices according to the customer name code to form a customer list; then, according to the customer area code, the customer list is clustered again to construct a device management area and assign several maintainers.

[0081] At the same time, the data management module uses the remote programming interface or network scanning technology to collect the hardware configuration information of the camera, lens, pan-tilt, listener, alarm detector, multi-functional decoder, protective cover, server performance, and fan speed of the camera device.

[0082] The data management module extracts software log information such as operation logs, system logs, error logs, and screenshots of the camera device using the SSL / TLS protocol; the data management module collects deployment environment data such as weather, temperature, humidity, light, and wind speed of the camera device using the built-in sensors or external sensors of the camera device;

[0083] The data management module first divides the hardware configuration information, software log information, and deployment environment data into several simulated data blocks according to the data collection time using the data chunking algorithm, and then recombines the information data within each simulated data block according to the customer name code to obtain the recombined information data blocks;

[0084] The data management module judges the data collection interval of all information data within the recombined information data blocks: if the collection time interval between a piece of information data and the adjacent information data is less than the preset time difference, and this piece of information data is the same as any adjacent information data, then this piece of information data is determined to be duplicate data, and the data deduplication algorithm is used to clean the duplicate data and remove redundant data;

[0085] If the collection time interval between a piece of information data and the adjacent information data is less than the preset time difference, but this piece of information data is different from any adjacent information data, then this piece of information data is determined to be noise data, and the information filtering algorithm is used to clean the noise data according to business requirements and smooth the data fluctuations;

[0086] If the collection time interval between a piece of information data and the adjacent information data is greater than the preset time difference, and this piece of information data is different from any adjacent information data, then this piece of information data is determined to be normal information data fluctuations, and the data consistency algorithm is used to perform consistency checks on the recombined information data blocks to obtain several simplified information data blocks;

[0087] If the collection time interval between a piece of information data and the adjacent information data is greater than the preset time difference, but this piece of information data is different from any adjacent information data, then this piece of information data is determined to be normal information data changes, and the data consistency algorithm is used to perform consistency checks on the recombined information data blocks to obtain several simplified information data blocks;

[0088] The data management module re-classifies the information data within the simplified information data blocks according to the customer address code using the data classification algorithm to obtain the hardware aggregation block, software aggregation block, and environment aggregation block; the data management module uses the data association algorithm to associate the hardware aggregation block, software aggregation block, and environment aggregation block of each camera device to obtain several associated information data blocks; the data management module uses the data encryption algorithm to perform nested-level encryption on the associated information data blocks according to the information sensitivity to obtain the encrypted information data blocks; and uses the meta-database to store the encrypted information data blocks, hardware configuration information, software log information, and deployment environment data;

[0089] The monitoring and analysis module uses the information decomposition layer of the intelligent diagnosis model to perform dimensionality reduction and desensitization on the stored encrypted information data blocks, and obtains non-sensitive hardware information blocks, non-sensitive software information blocks, and non-sensitive environment data blocks. At the same time, the extraction and verification layer of the intelligent diagnosis model verifies and aggregates the stored hardware configuration information, software log information, and deployment environment data to obtain a hardware verification set, a software verification set, and an environment verification set. Then, the feature extraction layer of the intelligent diagnosis model uses the bacterial foraging optimization algorithm to extract features from the non-sensitive hardware information blocks, non-sensitive software information blocks, and non-sensitive environment data blocks to obtain the group key feature values. Subsequently, the iterative training layer of the intelligent diagnosis model uses the firefly algorithm to perform several iterations on the group key feature values in combination with the hardware verification set, the software verification set, and the environment verification set to obtain the group feature weight matrix. Finally, the recognition and construction layer of the intelligent diagnosis model uses the graph construction algorithm to update and optimize the group feature weight matrix and construct a full-life cycle supervision and prediction graph.

[0090] The remote collaboration module monitors the operating status and conducts networked collaborative maintenance on several camera devices in each device management area according to the full-life cycle supervision and prediction graph.

[0091] The operation and maintenance service module includes a user side, a management side, an execution side, and a re-inspection side. Customers can view the operation and maintenance nodes, repair and replacement progress, and the operating status of each camera device in the operation and maintenance service through the user side. The administrator assigns operation and maintenance service orders to maintenance personnel according to the device management area through the management side and monitors the operation and maintenance progress of the maintenance personnel. The maintenance personnel receive the operation and maintenance service orders through the execution side and go to the deployment site of the camera device to perform operation and maintenance services. The maintenance personnel feedback the start time of the operation and maintenance, the progress of the maintenance and repair, and the end time to the management side in the form of images or text for storage. The re-inspection side uses the rule decision algorithm and the image recognition algorithm to monitor the operation and maintenance processes of all camera devices in each device management area.

Claims

1. A full life cycle intelligent monitoring platform for camera devices, characterized by: It includes equipment management module, data management module, monitoring and analysis module, remote collaboration module and user service module; among them, The equipment management module is used to manage the operation and maintenance process of the camera device and divide all the addresses where the camera device is installed into several equipment management areas and assign several maintenance personnel according to the contract order; the operation and maintenance process includes procurement, installation, configuration, upgrade and maintenance; A data management module is used to block, clean, classify, associate, encrypt and store hardware configuration information, software log information and deployment environment data of several camera devices in each device management area, and obtain encrypted information data blocks; The monitoring and analysis module is used to perform supervision and prediction on the entire life cycle of each camera device based on the encrypted information data block, hardware configuration information, software log information and deployment environment data through a threshold detection algorithm or an intelligent diagnosis model, and obtain a full life cycle supervision prediction map; Remote collaboration module, used to monitor the operating status and conduct online collaborative maintenance of several camera devices in each equipment management area based on the full life cycle supervision prediction map; The operation and maintenance service module is used to provide customers, administrators and maintainers in each equipment management area with a monitoring and feedback channel for the operation and maintenance process and progress throughout the life cycle of the camera device.

2. The full life cycle intelligent monitoring platform for camera devices according to claim 1 is characterized by: The equipment management module calculates the purchase quantity of the camera devices according to the inventory redundancy quantity, the contract order quantity and the iterative replacement quantity, and selects the manufacturer for purchase according to the purchase quantity; the equipment management module allocates the number of contract orders to be fulfilled and the number of camera devices to be installed to each maintenance worker; after the maintenance worker installs and connects the camera device to the network, he configures the function of the camera device according to business needs; the equipment management module obtains the equipment information of the camera device that has completed the function configuration; the equipment management module regularly updates and queries the firmware version and software version of each camera device, and if a new version of the firmware version and / or software version appears, the camera device is upgraded and maintained; The equipment management module clusters all the cameras according to the customer name code to form a customer list; it clusters the customer list again according to the customer area code, constructs the equipment management area and assigns a number of maintenance personnel.

3. The full life cycle intelligent monitoring platform of the camera device according to claim 1 is characterized by: The data management module uses a remote program programming interface or a network scanning technology to collect hardware configuration information of the camera device; the data management module uses the SSL / TLS protocol to extract software log information of the camera device; the data management module uses the built-in sensor or external sensor of the camera device to collect deployment environment data of the camera device; The data management module uses a data segmentation algorithm to first divide the hardware configuration information, software log information and deployment environment data into several simulation data blocks according to the data collection time, and then recombine the information data in each simulation data block according to the customer name code to obtain the reorganized information data block.

4. The full life cycle intelligent monitoring platform for camera devices according to claim 3 is characterized by: The data management module determines the data collection interval of all information data in the reorganized information data block; the data management module reclassifies the information data in the simplified information data block according to the customer address code to obtain a hardware aggregation block, a software aggregation block and an environment aggregation block; the data management module associates the hardware aggregation block, the software aggregation block and the environment aggregation block of each camera device to obtain a number of associated information data blocks; The data management module performs nested-level encryption on the associated information data blocks according to the information sensitivity to obtain the encrypted information data blocks; and uses the metadata database to store the encrypted information data blocks, hardware configuration information, software log information and deployment environment data.

5. The full life cycle intelligent monitoring platform of the camera device according to claim 1, characterized in that: The monitoring and analysis module uses the information decomposition layer of the intelligent diagnosis model to reduce the dimension of the stored encrypted information data block and desensitize it, and obtains the insensitive hardware information block, insensitive software information block and insensitive environment data block; at the same time, the extraction and inspection layer of the intelligent diagnosis model verifies and aggregates the stored hardware configuration information, software log information and deployment environment data, and obtains the hardware inspection set, software inspection set and environment inspection set; then the feature extraction layer of the intelligent diagnosis model uses the bacterial foraging optimization algorithm to extract features from the insensitive hardware information block, insensitive software information block and insensitive environment data block, and obtains the key feature values ​​of the group; Then, the iterative training layer of the intelligent diagnosis model uses the firefly algorithm to perform several iterations on the key characteristic values ​​of the group in combination with the hardware test set, software test set, and environmental test set to obtain the group characteristic weight matrix; Finally, the recognition and construction layer of the intelligent diagnosis model uses a graph construction algorithm to update and optimize the population feature weight matrix and construct a full life cycle supervision prediction graph.

6. The full life cycle intelligent monitoring platform of the camera device according to claim 5, characterized in that: The information decomposition layer first performs dimensionality reduction decomposition on the stored encrypted information data blocks to obtain reduced-dimensionality hardware information blocks, reduced-dimensionality software information blocks and reduced-dimensionality environment data blocks: the information decomposition layer then desensitizes the reduced-dimensionality hardware information blocks, reduced-dimensionality software information blocks and reduced-dimensionality environment data blocks formed by the dimensionality reduction decomposition to obtain non-sensitive hardware information blocks, non-sensitive software information blocks and non-sensitive environment data blocks.

7. The full life cycle intelligent monitoring platform of the camera device according to claim 5, characterized in that: The feature extraction layer first initializes the bacterial population and randomly generates several groups of sliding windows of different scales; then, through two 3×3 convolutional layers with a step size of 2, one 5×5 convolutional layer with a step size of 1, and two batch standard layers, the sliding windows of different scales are used to extract information from the non-sensitive hardware information block, the non-sensitive software information block, and the non-sensitive environment data block to obtain multi-level feature information values; Then, the multi-level characteristic information value is compared with the preset information threshold: if the multi-level characteristic information value is greater than or equal to the preset information threshold, the operation mode of this sliding window is swimming, and reproduction, aggregation and migration are performed in the non-sensitive hardware information block, non-sensitive software information block and non-sensitive environment data block until the information extraction is completed, thereby obtaining the key characteristic value of the group; If the multi-level feature information value is less than the preset information threshold, the sliding window will operate in a rolling manner, and the information extracted by the sliding window in each information block will be aggregated and eliminated, and an information extraction influence factor will be generated. The next set of sliding windows of different scales will be used in combination with the information extraction influence factor to extract information from each information block and obtain the key feature values ​​of the group.

8. The full life cycle intelligent monitoring platform for camera devices according to claim 1, characterized in that: The iterative training layer first randomly generates several groups of fireflies, each group of fireflies represents a group of candidate weight matrices; then each group of candidate weight matrices is used as the initial training weights of 2 parallel 3×3 convolutional layers, 2 CBS blocks, 2 residual blocks, 1 inverted residual block, 2 InceptionV3 blocks, 2 batch standard layers and Silu activation function; the CBS block includes 3 3×3 convolutional layers with a step size of 2, 2 batch standard layers and 1 Softmax activation function; the initial training weights are then used to iterate the key feature values ​​of the group several times, and then the real-time fitness of each group of initial training weights is evaluated; the real-time fitness is compared with the preset fitness: if the real-time fitness is less than the preset fitness, the ratio of the real-time fitness to the preset fitness is calculated, and the ratio is calculated with the next group of candidate weight matrices to obtain the corrected weight matrix, and then the network structure is iterated again using the corrected weight matrix until the last group of candidate weight matrices; If the real-time fitness is greater than or equal to the preset fitness, the learning rate of the current initial training weight is adaptively updated, and all candidate weight matrices are updated and calculated to obtain the group feature weight matrix; Finally, the group feature weight matrix is ​​updated and optimized using the hardware test set, software test set and environment test set.

9. The full life cycle intelligent monitoring platform of the camera device according to claim 1, characterized in that: The operation and maintenance service module includes a user end, a management end and an execution end; the customer views the operation and maintenance nodes, repair and replacement progress and the operating status of each camera device in the operation and maintenance service through the user end; the administrator assigns an operation and maintenance service order to the maintenance personnel according to the equipment management area through the management end, and monitors the maintenance personnel's operation and maintenance progress; the maintenance personnel receive the operation and maintenance service order through the execution end and go to the deployment site of the camera device to perform operation and maintenance services, and the maintenance personnel feedback the operation and maintenance start time, maintenance and repair progress and end time to the management end in the form of images or text for storage.

10. The full life cycle intelligent monitoring platform of the camera device according to claim 9, characterized in that: The operation and maintenance service module also includes a re-inspection end; the re-inspection end uses a rule decision algorithm and an image recognition algorithm to monitor the operation and maintenance process of all camera devices in each equipment management area.

Images