Smart archive integrated safety management system

Abstract

The invention discloses an integrated safety management system for an intelligent archive, relates to the technical field of intelligent management, and solves the problems that the safety of the archive needs to be highly guaranteed, and once the safety management system is attacked or data is leaked, the archive is greatly influenced. The security management system comprises an intelligent collection module, an information processing module, an information storage module, an authority management module, a file destruction module and a security policy module. Different types of archive data are processed through an information processing module, and then an archive association graph and an archive node number grade table are constructed; a corresponding safety protection monitoring strategy is generated through a strategy derivation model according to the operation states of all the function modules and the safety monitoring equipment, and then all-directional safety protection linkage is carried out on the function modules and the safety monitoring equipment so as to protect all archive data of the archive. The intrusion attack protection capability of the security management system is improved, and the data security of digital data is improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent management technology, and in particular to an integrated security management system for smart archives. Background Art

[0002] With the development of society, the trend of social informatization and digitalization is becoming increasingly stronger, and archives are gradually realizing digital management. The integrated security management system of smart archives is born to meet the needs of digital management. The integrated security management system of smart archives is a security management system based on network and cloud computing technologies. It can realize the security management of archives and the storage, management and query of digital archives, and improve the management efficiency and security of archives.

[0003] Patent No. CN2016100083709 discloses a smart archive big data platform system, including an archive smart transfer subsystem, an archive smart sorting and archiving subsystem, an archive smart warehouse management system (including physical or virtual archives), an archive warehouse environment monitoring and control system, an archive warehouse safety management system, an archive smart utilization subsystem, an archive smart regional management subsystem, an archive information visualization subsystem, and an archive smart management integrated application platform subsystem. The data of each subsystem is stored in the data center, and information exchange is also completed through the data center. The various subsystems are integrated into a unified integrated application platform to provide unified management of business applications. It is a new technology for managing the entire life cycle of archives, including a complete set of software and hardware systems from file collection and sorting to management and application, and an integrated application platform system for several intelligent fully automatic devices. Unified management of archive files is achieved from generation to processing and sorting to utilization, and linkage and intelligent management are achieved between various systems.

[0004] Patent No. CN2020111321887 discloses a digital archive system, including an archive type management module, a classification table management module, a code table management module, a sorting rule management module, a report management module, an archive management module, an archive transfer module and an archive retrieval module: it can customize a variety of archive types with different sorting methods according to the archive type template, and adjust the field definition and entry layout, customize different classification tables, code tables, and sorting rules for different archives, meet the archive management needs of various units, and solve the problem of low work efficiency caused by insufficient archive funds and manpower; at the same time, it realizes the digital management of archives, and solves the problem of insufficient warehouse space through the information assistance of the system; through web page information capture and transfer, OCR optical character recognition, document online transfer browsing, full-text indexing and other technologies, it provides convenient and fast full-text retrieval and browsing functions, and solves the problem of low sharing and utilization of archive information resources.

[0005] Patent No. CN2021106403330 discloses a hierarchical service model of a smart archive based on cluster analysis, which can provide archival services at the data layer, information layer, intelligence layer, and smart layer. The specific implementation idea is: adopt the K-Medoids algorithm to design a clustering analysis method with configurable attribute weights. The service needs of different levels are met by configuring the attribute weights. The service utilization value of the archives is improved on the basis of ensuring the value of the archival vouchers. Based on the role behavior and needs of the digital archives and combined with the current status of technological development, a new digital archives system architecture is designed to uniformly maintain system performance and security to ensure that each step is carried out safely and reliably. Then, based on the functional and technical requirements of each step, specific functional modules are designed, and detailed technical analysis is carried out to realize the professionalization, automation, flexibility and intelligent evolution of digital archives.

[0006] However, the above patents cannot highly guarantee the security of digital data and archives. Once the system is attacked or the data is leaked, it will have a great impact on the archives and the data security of digital data cannot be guaranteed. Summary of the invention

[0007] The purpose of the present invention is to provide an integrated security management system for a smart archive, which can process different types of archival materials through an information processing module, and then construct an archive association map and an archive node numbering level table; generate corresponding security protection monitoring strategies according to the operating status of each functional module and security monitoring equipment through a policy derivation model, and then link the functional modules with the security monitoring equipment for all-round security protection to protect all archival materials in the archive; and improve the intrusion attack protection capability of the security management system, while improving the data security of digital data.

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

[0009] An integrated security management system for smart archives, including an intelligent collection module, an information processing module, an information storage module, a permission management module, an archive destruction module and a security policy module; wherein:

[0010] Intelligent collection module, used to collect and organize different types of archival materials according to the archival confidentiality level and the archival location, and obtain digital data; the archival confidentiality level includes core, top secret, confidential, secret and public;

[0011] The information processing module is used to clean, classify and associate the collected archival materials according to the archive types, and to construct an archive association map and an archive node numbering level table; the archive types include text archives, image archives and audio-visual archives;

[0012] The information storage module is used to encrypt and store the digitally processed archive data according to the archive confidentiality level;

[0013] The authority management module is used to control the authority and authenticate the identity of visitors according to the confidentiality level of the archives;

[0014] The file destruction module is used to destroy expired or useless file materials according to the destruction management procedures;

[0015] The security policy module is used to generate corresponding security protection monitoring strategies based on the operating status of each functional module and security monitoring equipment using an adaptive autonomous protection algorithm or a policy derivative model.

[0016] Preferably, the intelligent collection module uses a color scanner to scan the public-level text files and image files in sequence according to the file placement position to obtain digital public information; the intelligent collection module uses a confidential digital workstation to scan the core-level, top-secret-level, confidential-level and secret-level text files and graphic files in sequence according to the file placement position to obtain digital confidential information; the intelligent collection module uses wireless or wired transmission methods to collect and organize the public-level audio and video files, and uses confidential mobile storage media to collect and organize the core-level, top-secret-level, confidential-level and secret-level audio and video files.

[0017] Preferably, the information processing module uses a text normalization algorithm and a wavelet transform algorithm to perform text normalization operations and image denoising on the text data and image data in the digitized semi-public data and the digitized confidential data respectively; the text normalization operation includes character conversion and traditional-simplified conversion;

[0018] The information processing module uses a word segmentation tool to remove stop words from the standardized text data, and divides the text data without stop words into several independent phrases; the information processing module cleans the uncommon characters or wrong characters in all phrases according to the custom dictionary, and calculates the similarity of all the cleaned phrases, thereby forming a description summary of the text data;

[0019] The information processing module uses the semantic feature model to segment the element features of the denoised image data, obtains several image element features, and uses the text description model combined with the image background to summarize all image element features in text to obtain a semantic summary;

[0020] The information processing module uses a sound quality separator and pixel difference algorithm to filter noise, separate speakers and segment shots of audio and video data in the audio and video archives. It also uses a fixed window to downsample and extract key frames of the audio and video archives according to the time and space sequence, and uses a multimodal model to summarize keywords for all key frames to obtain audio and video synopsis.

[0021] The information processing module uses the information association algorithm to associate the description summary, semantic summary and audio-visual summary according to the time and space sequence to obtain the archive association map; and uses the information classification algorithm to classify and number all the nodes in the archive association map according to the archive confidentiality level, and uses the clustering algorithm to normalize and aggregate the node numbers of archival materials with the same archive confidentiality level to construct an archive node number level table.

[0022] Preferably, the information storage module uses a random number generator to generate an N-bit random number, and performs confidentiality level power operations and hash operations on each node number to obtain the power level node number and the hash node number; the information storage module uses a quantum encryption function to perform quantum superposition operations on the power level node number to obtain the power level superposition number; and performs quantum entanglement operations on the hash node number to obtain the hash entanglement number; the information storage module uses a stream encryption algorithm to concatenate the power level superposition number and the hash entanglement number and then performs real-time encryption operations to obtain a steganographic encrypted number series; the information storage module performs multiple encryption operations on the steganographic encrypted number series and the archival data stored in each node in the archive association map, thereby completing the encrypted storage of digitized data.

[0023] Preferably, the authority management module first utilizes the information separation layer to divide the archival materials in each node in the archive association map into several separated blocks according to the content of the materials; the content of the materials includes legislative archives, administrative archives, military archives, diplomatic archives, economic archives, scientific and technological archives, art archives and religious archives; then utilizes the authority granting layer to restrict the access rights of archival materials with different content according to the confidentiality level, and after the identity verification of the visitor, based on the search query content, displays to the visitor the placement and access dynamic code of all relevant archival materials within the access rights, and at the same time displays the verification documents required to lift the restrictions on the access rights of all relevant archival materials outside the access rights; finally, utilizes the intelligent monitoring layer to monitor the visitor's search query behavior and archive browsing behavior, and reminds and warns the visitor of the over-authority behavior.

[0024] Preferably, the archive destruction module regularly conducts social demand judgment on the data content, collection time, data source, content reliability, physical appearance and storage period of all archival materials in each node of the archive association map, obtains the preservation value of each archival material, and then marks the placement of archival materials with no preservation value to form a destruction list; the archive destruction module initiates a value review procedure for all archival materials in the destruction list, and the administrator prepares an appraisal report, briefly introduces the number and content of the archival materials to be destroyed, and puts forward destruction opinions and reasons for destruction; the archive destruction module submits the appraisal report and the destruction list to the superior person in charge for approval; after approval, the sales supervisor will count and check the entities of the archival materials to be destroyed and sign and seal them; after the physical archival materials are destroyed, the archive destruction module will automatically clear the corresponding digital data and permanently preserve the destruction list.

[0025] Preferably, the security policy module includes a program monitoring unit and a dynamic policy unit; the program monitoring unit uses a policy derivation model to parse and separate the operation process of each functional module using a parsing separation function to obtain the state data stream of each functional module; the functional module includes an intelligent collection module, an information processing module, an information storage module, a permission management module and an archive destruction module; the policy derivation model uses a dimensional compression algorithm to compress the dimension of the state data stream several times according to the archive confidentiality level:

[0026]

[0027] Among them, Z represents the potential dimension, W represents the weight, σ() represents the activation function, b represents the bias vector, Q E represents the CVAE encoding network, X represents the state data stream sample, Y represents the true label, s represents the file confidentiality level, s = 1, 2, 3, 4, 5;

[0028] Then the state data stream after dimension reduction is input into the feature extraction layer to extract the feature information of the state data stream; the feature extraction layer includes three 3×3 convolution layers, two 1×1 point convolution layers, one 3×3 maximum pooling layer and two normalization layers; then the extracted feature information is encoded using the Detec encoder; H i+1 =R((Z|H i )(A|W))(2)

[0029] Among them, H i represents the i-th convolutional layer, R() represents the RELU activation function, i represents the number of convolutional layers, and A represents the adjacency matrix;

[0030] Then, the state data stream after information compilation is input into the two network branches of the learning and training layer for iterative training to obtain the deep feature matrix; the first network branch includes 2 3×3 convolutional layers, 3 3×3 upsampling layers, 3 3×3 downsampling layers, 2 1×1 point convolutional layers and 1 SoftPlus activation function, and the second network branch includes 2 1×1 convolutional layers, 3 3×1 upsampling layers, 3 1×3 downsampling layers, 3 CBS blocks, 2 inverted residual blocks, 2 InceptionV3 blocks, 2 SigMoid activation functions and 2 MISH activation functions; the CBS block includes 3 5×5 convolutional layers with a step size of 2, 2 3×3 convolutional layers with a step size of 1, 1 7×7 hole convolutional layer with a step size of 2 and 2 Silu activation functions; at the same time, the deep feature values ​​output by the two network branches are spliced ​​and fused;

[0031] Then, the deep feature values ​​of the completed splicing and fusion are reconstructed using the reconstruction decoder for both structure and attribute

[0032] Reconstruction; S = P (H i ,A|W)+SE(H i H i T |W)(3)

[0033] Among them, S represents the reconstructed decoding value, P() represents the PRelu activation function, SE() represents the SELU activation function, and T represents transposition;

[0034] Finally, the reconstructed state data stream passes through the average pooling layer, the fully connected layer and the LRelu activation function in turn to divide the state data stream into benign data stream, malignant data stream and unknown suspicious data stream.

[0035] Preferably, the dynamic policy unit determines the expected benefit of the unknown suspicious data flow according to the probability of the unknown suspicious data flow successfully invading the security monitoring device through the fully heterogeneous layer of the policy derivation model:

[0036] B1(μ)=B(-C(t0)+α-t0) / C(α-t0) (4)

[0037] Among them, B1 represents the total benefit, B() represents the expected benefit function, μ represents the benefit, C() represents the cost, t0 represents the attack time, and α represents the rotation period;

[0038] At this point, the defense cycle of the security monitoring equipment is extended, making the expected benefit of the unknown suspicious data flow negative, and the unknown suspicious data flow loses its intrusion capability:

[0039] Among them, γ represents the rational factor, f() represents the invasion probability density function, and t represents time;

[0040] When passing through the finite heterogeneous layer of the policy-derived model, the unknown suspicious data flow uses the number of intrusions as accumulated experience. When the unknown suspicious data flow invades again, the expected benefit of the unknown suspicious data flow is:

[0041]

[0042] Among them, B2 represents limited benefits, k represents the number of invasions;

[0043] When the two levels of the strategy derivative model are run irregularly, the expected comprehensive return is:

[0044]

[0045] Among them, β represents the level transition probability;

[0046] Finally, according to the comprehensive benefit expectations, a dynamic defense strategy for security monitoring equipment is formulated and the combined architecture of each functional module of the security management system is adjusted to achieve security isolation of the security management system and generate corresponding security protection monitoring strategies; security monitoring equipment includes video surveillance equipment, access control equipment, network defense equipment, fire protection facilities, environmental monitoring equipment, intelligent lighting equipment, electronic patrol equipment and information release equipment.

[0047] The present invention processes different types of archival materials through an information processing module, and then constructs an archival association map and an archival node numbering level table; generates corresponding security protection monitoring strategies according to the operating status of each functional module and security monitoring equipment through a policy derivation model, and then links the functional modules with the security monitoring equipment for all-round security protection to ensure the safety of all archival materials in the archives; improves the intrusion attack protection capability of the security management system and enhances the data security of digitized data. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or related technologies, the drawings required for use in the embodiments or related technical descriptions are briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0049] Figure 1 This is the schematic diagram of the safety management system;

[0050] Figure 2 Schematic diagram of the strategy derivation model principle. DETAILED DESCRIPTION

[0051] The present invention is described in detail below with reference to the accompanying drawings and embodiments:

[0052] like Figure 1 to Figure 2 As shown, the integrated security management system for smart archives described in the present invention includes an intelligent collection module, an information processing module, an information storage module, a permission management module, an archive destruction module and a security policy module; wherein,

[0053] Intelligent collection module, used to collect and organize different types of archival materials according to the archival confidentiality level and the archival location, and obtain digital data; the archival confidentiality level includes core, top secret, confidential, secret and public;

[0054] The information processing module is used to perform information cleaning, information classification and information association on the digitized data in turn according to the archive type, and to construct an archive association map and an archive node numbering level table; the archive types include text archives, image archives and audio-visual archives;

[0055] The information storage module is used to encrypt and store the archive data that has completed the information association according to the archive confidentiality level;

[0056] The authority management module is used to control the authority and authenticate the identity of visitors according to the confidentiality level of the archives;

[0057] The file destruction module is used to destroy expired or useless file materials according to the destruction management procedures;

[0058] The security policy module is used to generate corresponding security protection monitoring strategies based on the operating status of each functional module and security monitoring equipment using an adaptive autonomous protection algorithm or a policy derivative model.

[0059] In the present invention, the intelligent collection module uses a color scanner to scan the text files and image files of the public level and the secret level in sequence according to the file placement position to obtain digital semi-public information; the intelligent collection module uses a confidential digital workstation to scan the text files and graphic files of the core level, the top secret level and the confidential level in sequence according to the file placement position to obtain digital confidential information; the intelligent collection module uses a wireless or wired transmission method to collect and organize the audio and video files of the public level and the secret level, and uses a confidential mobile storage medium to collect and organize the audio and video files of the core level, the top secret level and the confidential level;

[0060] In the present invention, the information processing module uses a text normalization algorithm and a wavelet transform algorithm to perform text normalization operations and image denoising on the text data and image data in the digitized semi-public data and the digitized confidential data respectively; the text normalization operation includes character conversion and traditional-simplified conversion;

[0061] The information processing module uses a word segmentation tool to remove stop words from the standardized text data, and divides the text data without stop words into several independent phrases; the information processing module cleans the uncommon characters or wrong characters in all phrases according to the custom dictionary, and calculates the similarity of all the cleaned phrases, thereby forming a description summary of the text data;

[0062] The information processing module uses the semantic feature model to segment the element features of the denoised image data, obtains several image element features, and uses the text description model combined with the image background to summarize all image element features in text to obtain a semantic summary;

[0063] The information processing module uses a sound quality separator and pixel difference algorithm to filter noise, separate speakers and segment shots of audio and video data in the audio and video archives. It also uses a fixed window to downsample and extract key frames of the audio and video archives according to the time and space sequence, and uses a multimodal model to summarize keywords for all key frames to obtain audio and video synopsis.

[0064] The information processing module uses the information association algorithm to associate the description summary, semantic summary and audio-visual summary according to the time-space sequence to obtain the archive association map; and uses the information classification algorithm to classify and number all nodes in the archive association map according to the archive confidentiality level, and uses the clustering algorithm to normalize and aggregate the node numbers of archive materials with the same archive confidentiality level to construct an archive node number level table;

[0065] In the present invention, the information storage module generates an N-bit random number using a random number generator, and performs a power operation and a hash operation on each node number to obtain a power node number and a hash node number; the information storage module performs a quantum superposition operation on the power node number using a quantum encryption function to obtain a power superposition number; and performs a quantum entanglement operation on the hash node number to obtain a hash entanglement number; the information storage module performs a real-time encryption operation after splicing the power superposition number and the hash entanglement number using a stream encryption algorithm to obtain a steganographic encrypted number series; the information storage module performs multiple encryption operations on the steganographic encrypted number series and the archive data stored in each node in the archive association map, thereby completing the encrypted storage of the digital data;

[0066] In the present invention, the authority management module first uses the information separation layer to divide the archival data in each node in the archive association map into several separation blocks according to the data content; the data content includes legislative archives, administrative archives, military archives, diplomatic archives, economic archives, scientific and technological archives, art archives and religious archives; then the authority granting layer is used to restrict the access rights of archival data with different data contents according to the confidentiality level, and after the identity verification of the visitor, the placement position and access dynamic code of all relevant archival data within the access right are displayed to the visitor based on the search query content, and the verification file required for lifting the restriction of the access right of all relevant archival data outside the access right is displayed; finally, the intelligent monitoring layer is used to monitor the visitor's search query behavior and archive browsing behavior, and remind and warn the visitor of the over-authority behavior;

[0067] In the present invention, the archive destruction module regularly judges the social demand of the data content, collection time, data source, content reliability, physical appearance and storage period of all the archive materials in each node of the archive association map, obtains the preservation value of each archive material, and then marks the placement position of the archive materials without preservation value to form a destruction list; the archive destruction module starts the value review procedure for all the archive materials in the destruction list, and the administrator prepares an appraisal report, briefly introduces the number and content of the archive materials to be destroyed, and puts forward destruction opinions and reasons for destruction; the archive destruction module submits the appraisal report and the destruction list to the superior person in charge for approval; after approval, the sales supervisor will count and check the entities of the archive materials to be destroyed and sign and seal them; after the physical archive materials are destroyed, the archive destruction module will automatically clear the corresponding digital data and permanently preserve the destruction list;

[0068] In the present invention, the security policy module includes a program monitoring unit and a dynamic policy unit; the program monitoring unit uses a policy derivation model to parse and separate the operation process of each functional module using an analytical separation function to obtain the state data stream of each functional module; the functional module includes an intelligent collection module, an information processing module, an information storage module, a permission management module and an archive destruction module; the policy derivation model uses a dimensional compression algorithm to compress the dimension of the state data stream several times according to the archive confidentiality level:

[0069]

[0070] Among them, Z represents the potential dimension, W represents the weight, σ() represents the activation function, b represents the bias vector, Q E represents the CVAE encoding network, X represents the state data stream sample, Y represents the true label, s represents the file confidentiality level, s = 1, 2, 3, 4, 5;

[0071] In this embodiment, the core level is 5, the top secret level is 4, the confidential level is 3, the secret level is 2, and the public level is 1; as the confidentiality level of the file increases, the number of compressions increases;

[0072] Then the state data stream after dimension reduction is input into the feature extraction layer to extract the feature information of the state data stream; the feature extraction layer includes three 3×3 convolution layers, two 1×1 point convolution layers, one 3×3 maximum pooling layer and two normalization layers; then the extracted feature information is encoded using the Detec encoder; H i+1 =R((Z|H i )(A|W))(2)

[0073] Among them, H i represents the i-th convolutional layer, R() represents the RELU activation function, i represents the number of convolutional layers, and A represents the adjacency matrix;

[0074] Then, the state data stream after information compilation is input into the two network branches of the learning and training layer for iterative training to obtain the deep feature matrix; the first network branch includes 2 3×3 convolutional layers, 3 3×3 upsampling layers, 3 3×3 downsampling layers, 2 1×1 point convolutional layers and 1 SoftPlus activation function, and the second network branch includes 2 1×1 convolutional layers, 3 3×1 upsampling layers, 3 1×3 downsampling layers, 3 CBS blocks, 2 inverted residual blocks, 2 InceptionV3 blocks, 2 SigMoid activation functions and 2 MISH activation functions; the CBS block includes 3 5×5 convolutional layers with a step size of 2, 2 3×3 convolutional layers with a step size of 1, 1 7×7 hole convolutional layer with a step size of 2 and 2 Silu activation functions; at the same time, the deep feature values ​​output by the two network branches are spliced ​​and fused;

[0075] Then, the deep feature values ​​of the completed splicing and fusion are reconstructed using the reconstruction decoder for both structure and attribute

[0076] Reconstruction; S = P (H i ,A|W)+SE(H i H i T |W)(3)

[0077] Among them, S represents the reconstructed decoding value, P() represents the PRelu activation function, SE() represents the SELU activation function, and T represents transposition;

[0078] Finally, the reconstructed state data stream is sequentially passed through the average pooling layer, the fully connected layer, and the LRelu activation function to divide the state data stream into benign data stream, malignant data stream, and unknown suspicious data stream;

[0079] In this embodiment, the state data stream of the intelligent collection module includes light source intensity, exposure, resolution and color mode; the state data stream of the information processing module includes text standardization rules, text encoding format, stop word list, part-of-speech tagging rules, phrase similarity, image denoising threshold, feature segmentation value, element description template, noise filtering threshold, timbre recognition, shot segmentation threshold, training learning rate and association logic rules; the state data stream of the information storage module includes power operation, hash operation, quantum superposition operation, quantum entanglement operation and multiple encryption operation; the state data stream of the authority management module includes information separation rules, authority granting level, identity verification code and intelligent monitoring frequency; the state data stream of the archive destruction module includes social demand judgment rules, destruction inventory records and automatic clearing process;

[0080] In the present invention, the dynamic policy unit determines the expected benefit of the unknown suspicious data flow according to the probability of the unknown suspicious data flow successfully invading the security monitoring device through the fully heterogeneous layer of the policy derivation model:

[0081] B1(μ)=B(-C(t0)+α-t0) / C(α-t0) (4)

[0082] Among them, B1 represents the total benefit, B() represents the expected benefit function, μ represents the benefit, C() represents the cost, t0 represents the attack time, and α represents the rotation period;

[0083] At this point, the defense cycle of the security monitoring equipment is extended, making the expected benefit of the unknown suspicious data flow negative, and the unknown suspicious data flow loses its intrusion capability:

[0084] Among them, γ represents the rational factor, f() represents the invasion probability density function, and t represents time;

[0085] When passing through the finite heterogeneous layer of the policy-derived model, the unknown suspicious data flow uses the number of intrusions as accumulated experience. When the unknown suspicious data flow invades again, the expected benefit of the unknown suspicious data flow is:

[0086]

[0087] Among them, B2 represents limited benefits, k represents the number of invasions;

[0088] When the two levels of the strategy derivative model are run irregularly, the expected comprehensive return is:

[0089]

[0090] Among them, β represents the level transition probability;

[0091] Finally, according to the comprehensive benefit expectations, a dynamic defense strategy for security monitoring equipment is formulated and the combined architecture of each functional module of the security management system is adjusted to achieve security isolation of the security management system and generate corresponding security protection monitoring strategies; security monitoring equipment includes video surveillance equipment, access control equipment, network defense equipment, fire protection facilities, environmental monitoring equipment, intelligent lighting equipment, electronic patrol equipment and information release equipment.

[0092] Example:

[0093] When the security management system is started, the intelligent collection module uses a color scanner to scan the text files and image files of the public and secret levels in sequence according to the file placement to obtain digital semi-public information; the intelligent collection module uses a confidential digital workstation to scan the text files and graphic files of the core, top secret and confidential levels in sequence according to the file placement to obtain digital confidential information; the intelligent collection module uses wireless or wired transmission methods to collect and organize the audio and video files of the public and secret levels, and uses confidential mobile storage media to collect and organize the audio and video files of the core, top secret and confidential levels.

[0094] The information processing module uses text normalization algorithm and wavelet transform algorithm to perform text normalization operation and image denoising on the text data and image data in the digital semi-public data and digital confidential data respectively; the text normalization operation includes character conversion and traditional-simplified conversion;

[0095] The information processing module uses a word segmentation tool to remove stop words from the standardized text data, and divides the text data without stop words into several independent phrases; the information processing module cleans the uncommon characters or wrong characters in all phrases according to the custom dictionary, and calculates the similarity of all the cleaned phrases, thereby forming a description summary of the text data;

[0096] The information processing module uses the semantic feature model to segment the element features of the denoised image data, obtains several image element features, and uses the text description model combined with the image background to summarize all image element features in text to obtain a semantic summary;

[0097] The information processing module uses a sound quality separator and pixel difference algorithm to filter noise, separate speakers and segment shots of audio and video data in the audio and video archives. It also uses a fixed window to downsample and extract key frames of the audio and video archives according to the time and space sequence, and uses a multimodal model to summarize keywords for all key frames to obtain audio and video synopsis.

[0098] The information processing module uses the information association algorithm to associate the description summary, semantic summary and audio-visual synopsis according to the time-space sequence to obtain the archive association map; and uses the information classification algorithm to classify and number all nodes in the archive association map according to the archive confidentiality level, and uses the clustering algorithm to normalize and aggregate the node numbers of archival materials with the same archive confidentiality level to construct an archive node number level table; the information storage module uses a random number generator to generate N-bit random numbers, and performs confidentiality level power operations and hash operations on each node number to obtain the power level node number and the hash node number; the information storage module uses the quantum encryption function to perform quantum superposition operations on the power level node number to obtain the power level superposition number; and performs quantum entanglement operations on the hash node number to obtain the hash entanglement number; the information storage module uses the stream encryption algorithm to splice the power level superposition number with the hash entanglement number and then performs real-time encryption operations to obtain the steganographic encrypted number series; the information storage module performs multiple encryption operations on the steganographic encrypted number series and the archival materials stored in each node in the archive association map, thereby completing the encrypted storage of digital data.

[0099] The authority management module first uses the information separation layer to divide the archival materials in each node in the archive association map into several separated blocks according to the content of the materials; the content of the materials includes legislative archives, administrative archives, military archives, diplomatic archives, economic archives, scientific and technological archives, art archives and religious archives; then the authority granting layer is used to restrict the access rights of archival materials with different content according to the confidentiality level, and after the identity verification of the visitor, the visitor is shown the placement and access dynamic code of all relevant archival materials within the access rights based on the search query content, and the verification documents required to lift the restrictions on the access rights of all relevant archival materials outside the access rights are displayed; finally, the intelligent monitoring layer is used to monitor the visitor's search query behavior and archive browsing behavior, and remind and warn the visitor of the over-authority behavior.

[0100] The archive destruction module regularly conducts social demand judgments on the content, collection time, source, content reliability, physical appearance and storage period of all archival materials in each node of the archive association map, obtains the preservation value of each archival material, and then marks the placement of archival materials with no preservation value to form a destruction list; the archive destruction module initiates a value review procedure for all archival materials in the destruction list, and the administrator prepares an appraisal report, briefly introduces the number and content of archival materials to be destroyed, and puts forward destruction opinions and reasons for destruction; the archive destruction module submits the appraisal report and destruction list to the superior person in charge for approval; after approval, the sales supervisor will count and check the entities of the archival materials to be destroyed and sign and seal them; after the physical archival materials are destroyed, the archive destruction module will automatically clear the corresponding digital materials and permanently preserve the destruction list.

[0101] The security policy module includes a program monitoring unit and a dynamic policy unit; the program monitoring unit uses a policy derivation model and a parsing separation function to parse and separate the operation processes of the intelligent collection module, information processing module, information storage module, authority management module and archive destruction module, and obtains the status data stream of each functional module: the status data stream of the intelligent collection module includes light source intensity, exposure, resolution and color mode; the status data stream of the information processing module includes text standardization rules, text encoding format, stop word list, part-of-speech tagging rules, phrase similarity, image denoising threshold, feature segmentation value, element description template, noise filtering threshold, timbre recognition, shot segmentation threshold, training learning rate and association logic rules; the status data stream of the information storage module includes power operation, hash operation, quantum superposition operation, quantum entanglement operation and multiple encryption operation.

[0102] The status data stream of the authority management module includes information separation rules, authority granting levels, identity verification codes and intelligent monitoring frequencies; the status data stream of the archive destruction module includes social demand judgment rules, destruction inventory records and automatic clearing processes; the policy derivation model uses a dimensionality compression algorithm to compress the dimensions of the status data stream several times according to the archive confidentiality level; then the state data stream that has undergone dimensionality reduction processing is input into the feature extraction layer to extract the feature information of the state data stream.

[0103] The feature extraction layer includes three 3×3 convolutional layers, two 1×1 point convolutional layers, one 3×3 maximum pooling layer and two normalization layers; the extracted feature information is then encoded using the Detec encoder; the state data stream after information encoding is then input into the two network branches of the learning and training layer for iterative training to obtain the deep feature matrix; the first network branch includes two 3×3 convolutional layers, three 3×3 upsampling layers, three 3×3 downsampling layers, two 1×1 point convolutional layers and one SoftPlus activation function, and the second network branch includes two 1×1 convolutional layers, three 3×1 upsampling layers, three 1×3 downsampling layers, three CBS blocks, two inverted Residual block, 2 InceptionV3 blocks, 2 SigMoid activation functions and 2 MISH activation functions; CBS block includes 3 5×5 convolution layers with a step size of 2, 2 3×3 convolution layers with a step size of 1, 1 7×7 hole convolution layer with a step size of 2 and 2 Silu activation functions; at the same time, the deep feature values ​​output by the two network branches are spliced ​​and fused; then the spliced ​​and fused deep feature values ​​are reconstructed in both structure and attribute using the reconstruction decoder; finally, the reconstructed state data stream passes through the average pooling layer, the fully connected layer and the LRelu activation function in turn, and the state data stream is divided into benign data stream, malignant data stream and unknown suspicious data stream;

[0104] The dynamic policy unit determines the expected benefit of the unknown suspicious data flow according to the probability of the unknown suspicious data flow successfully invading the security monitoring device through the fully heterogeneous layer of the policy derivative model; at this time, the defense period of the security monitoring device is extended to make the expected benefit of the unknown suspicious data flow negative, and the unknown suspicious data flow loses its invasion capability; when passing through the limited heterogeneous layer of the policy derivative model, the unknown suspicious data flow uses the number of invasions as accumulated experience, and calculates the expected benefit of the unknown suspicious data flow when the unknown suspicious data flow invades again; when the two levels of the policy derivative model operate irregularly, the comprehensive expected benefit is calculated; finally, according to the comprehensive expected benefit, a dynamic defense strategy for the security monitoring equipment is formulated and the combined architecture of each functional module of the security management system is adjusted to achieve security isolation of the security management system and generate corresponding security protection monitoring strategies; security monitoring equipment includes video surveillance equipment, access control equipment, network defense equipment, fire protection facilities, environmental monitoring equipment, intelligent lighting equipment, electronic patrol equipment and information release equipment.

Claims

1. An integrated security management system for smart archives, characterized by: It includes intelligent collection module, information processing module, information storage module, authority management module, file destruction module and security policy module; among them, Intelligent collection module, used to collect and organize different types of archival materials according to the archival confidentiality level and the archival location, and obtain digital data; the archival confidentiality level includes core, top secret, confidential, secret and public; The information processing module is used to perform information cleaning, information classification and information association on the digitized data in turn according to the archive type, and to construct an archive association map and an archive node numbering level table; the archive types include text archives, image archives and audio-visual archives; The information storage module is used to encrypt and store the archive data that has completed the information association according to the archive confidentiality level; The authority management module is used to control the authority and authenticate the identity of visitors according to the confidentiality level of the archives; The file destruction module is used to destroy expired or useless file materials according to the destruction management procedures; The security policy module is used to generate corresponding security protection monitoring strategies based on the operating status of each functional module and security monitoring equipment using an adaptive autonomous protection algorithm or a policy derivative model.

2. The integrated security management system for smart archives according to claim 1 is characterized by: The intelligent collection module uses a color scanner to scan the public-level text files and image files in order according to the file placement positions to obtain digital public data; The intelligent collection module uses the confidential digital workstation to scan the text and graphic files of the core level, top secret level, confidential level and secret level in sequence according to the file placement to obtain the digitized confidential information; The intelligent collection module collects and organizes the public-level audio and video archives using wireless or wired transmission methods, and collects and organizes the core-level, top-secret-level, confidential-level and secret-level audio and video archives using confidential mobile storage media.

3. The integrated security management system for smart archives according to claim 1 is characterized by: The information processing module uses a text normalization algorithm and a wavelet transform algorithm to perform text normalization operations and image denoising on the text data and image data in the digitized public data and the digitized confidential data respectively; the text normalization operation includes character conversion and traditional-simplified conversion; The information processing module uses a word segmentation tool to remove stop words from the standardized text data, and divides the text data without stop words into several independent phrases; the information processing module cleans the rare characters or incorrect characters in all phrases according to the custom dictionary, and calculates the similarity of all cleaned phrases, thereby forming a description summary of the text data.

4. The integrated security management system for smart archives according to claim 3 is characterized by: The information processing module uses the semantic feature model to perform element feature segmentation on the denoised image data to obtain a number of image element features, and uses the text description model combined with the image background to perform a text summary of all image element features to obtain a semantic summary; The information processing module uses a sound quality separator and pixel difference algorithm to filter noise, separate speakers and segment shots of audio and video data in the audio and video archives. It also uses a fixed window to downsample and extract key frames of the audio and video archives according to the time and space sequence, and uses a multimodal model to summarize keywords for all key frames to obtain audio and video synopsis. The information processing module uses the information association algorithm to associate the description summary, semantic summary and audio-visual summary according to the time and space sequence to obtain the archive association map; and uses the information classification algorithm to classify and number all the nodes in the archive association map according to the archive confidentiality level, and uses the clustering algorithm to normalize and aggregate the node numbers of archival materials with the same archive confidentiality level to construct an archive node number level table.

5. The integrated security management system for smart archives according to claim 1 is characterized by: The information storage module uses a random number generator to generate an N-bit random number, and performs a power operation and a hash operation on each node number to obtain a power-level node number and a hash node number; the information storage module uses a quantum encryption function to perform a quantum superposition operation on the power-level node number to obtain a power-level superposition number; and performs a quantum entanglement operation on the hash node number to obtain a hash entanglement number; the information storage module uses a stream encryption algorithm to splice the power-level superposition number and the hash entanglement number, and then performs a real-time encryption operation to obtain a steganographic encrypted number series; the information storage module performs multiple encryption operations on the steganographic encrypted number series and the archive data stored in each node in the archive association map, thereby completing the encrypted storage of digital data.

6. The integrated security management system for smart archives according to claim 1 is characterized by: The authority management module first uses the information separation layer to divide the archival data in each node in the archive association map into several separated blocks according to the data content; the data content includes legislative archives, administrative archives, military archives, diplomatic archives, economic archives, scientific and technological archives, art archives and religious archives; then uses the authority granting layer to limit the access rights of archival data with different data contents according to the confidentiality level, and after the visitor's identity verification, based on the search query content, the visitor is displayed with the placement and access dynamic code of all relevant archival data within the access permission, and at the same time, the verification documents required for lifting the restrictions on the access rights of all relevant archival data outside the access permission are displayed; finally, the intelligent monitoring layer is used to monitor the visitor's search query behavior and archive browsing behavior, and the visitor is reminded and warned of the over-authority behavior.

7. The integrated security management system for smart archives according to claim 1 is characterized by: The archive destruction module regularly conducts social demand judgment on the data content, collection time, data source, content reliability, physical appearance and storage period of all archival materials in each node of the archive association map, obtains the preservation value of each archival material, and then marks the placement of archival materials with no preservation value to form a destruction list; the archive destruction module initiates a value review procedure for all archival materials in the destruction list, and the administrator prepares an appraisal report, briefly introduces the number and content of the archival materials to be destroyed, and puts forward destruction opinions and reasons for destruction; the archive destruction module submits the appraisal report and the destruction list to the superior person in charge for approval; after approval, the sales supervisor will count and check the entities of the archival materials to be destroyed and sign and seal them; after the physical archival materials are destroyed, the archive destruction module will automatically clear the corresponding digital data and permanently preserve the destruction list.

8. The integrated security management system for smart archives according to claim 1 is characterized by: The security policy module includes a program monitoring unit and a dynamic policy unit; the program monitoring unit uses a policy derivation model to parse and separate the operation process of each functional module and obtain the state data stream of each functional module; the functional module includes an intelligent collection module, an information processing module, an information storage module, a permission management module and an archive destruction module; The policy derivation model uses the dimension compression algorithm to compress the dimension of the state data stream several times according to the confidentiality level of the archive; Then the state data stream after dimension reduction is input into the feature extraction layer to extract the feature information of the state data stream; then the extracted feature information is compiled using the Detec encoder; Then, the information-compiled state data stream is input into the two network branches of the learning and training layer for iterative training to obtain the deep feature matrix; at the same time Concatenate and fuse the deep feature values ​​output by the two network branches; The concatenated and fused deep feature values ​​are then reconstructed in terms of structure and attributes using a reconstruction decoder. Finally, the reconstructed state data stream is passed through an average pooling layer, a fully connected layer, and an LRelu activation function in sequence, thereby dividing the state data stream into benign data stream, malignant data stream, and unknown suspicious data stream.

9. The integrated security management system for smart archives according to claim 8 is characterized by: The dynamic policy unit determines the expected benefit of the unknown suspicious data flow according to the probability of the unknown suspicious data flow successfully invading the security monitoring device through the fully heterogeneous layer of the policy derivation model; At this time, the defense cycle of the security monitoring equipment is extended to make the expected benefit of the unknown suspicious data flow negative, and the unknown suspicious data flow loses its intrusion ability. When passing through the finite heterogeneous layer of the policy derivative model, the unknown suspicious data flow uses the number of intrusions as accumulated experience, and when the unknown suspicious data flow invades again, the expected benefit of the unknown suspicious data flow is calculated. Obtain the expected comprehensive return when the two levels of the strategy derivative model are operated irregularly; Finally, according to the comprehensive benefit expectations, a dynamic defense strategy for security monitoring equipment is formulated and the combined architecture of each functional module of the security management system is adjusted to achieve security isolation of the security management system and generate corresponding security protection monitoring strategies; security monitoring equipment includes video surveillance equipment, access control equipment, network defense equipment, fire protection facilities, environmental monitoring equipment, intelligent lighting equipment, electronic patrol equipment and information release equipment.

Images