Anonymous identity energy big data sharing security audit method based on data update

Abstract

The present application relates to the field of data security technology, and specifically to an anonymous identity energy big data sharing security audit method based on data update. The method includes: presetting a security factor, obtaining public parameters and a master private key to authorize the agent; obtaining energy elements and energy entity texts; obtaining the lexical level feature strength of energy vocabulary according to the distance between energy vocabulary in the energy entity text, based on this and combining the similarity between energy vocabulary to obtain the replacement vocabulary homomorphic hierarchy, and obtaining the lexical level value coefficient according to the lexical level feature strength and the replacement vocabulary homomorphic hierarchy; obtaining a representative score sequence according to the lexical level value coefficient, obtaining a specific vocabulary convergence index and a topic elimination strength according to the characteristics of the representative score sequence, and obtaining the element text distinction by the two and the security factor; obtaining an ordered identifier according to the element text distinction to complete encryption, and storing the encryption result to complete data update. The present application improves the security factor of encryption.

Description

Technical Field

[0001] The present application relates to the field of data security technology, and in particular to an anonymous identity energy big data sharing security audit method based on data update. Background Art

[0002] Energy big data sharing provides users with a convenient, flexible and efficient data management method in today's era of energy data explosion. Due to the sensitivity of energy data, the transmission and storage of data are all running in the public network, the complexity of the system and the frequency of attacks in the public network environment, the outsourced energy data and the user's identity privacy face severe security threats. Because of the outsourced storage of data, users lose physical control over the outsourced energy data and are vulnerable to online eavesdropping, malicious deletion or tampering by external adversaries. Therefore, how to use cryptographic security technology to ensure the integrity and confidentiality of outsourced energy data and identity anonymity is an important and hot research topic.

[0003] The original data owner sends the energy data he owns to the agent through a secure channel according to the provisions of the agent agreement. The agent divides the energy data file into multiple energy data blocks. The traditional division method is to select continuous fixed-length energy data from the energy data file as energy data blocks, which will lead to a large data correlation between energy data blocks and the problem of low security of the ciphertext information obtained by encrypting subsequent energy data blocks separately. Traditional clustering algorithms, such as the AGNES clustering algorithm, are simple and easy to implement, and can realize the classification of energy entity texts, but they often need to set the number of clusters in advance, and there is a problem that they cannot be adaptively adjusted according to the energy entity text and have large errors. Summary of the invention

[0004] In order to solve the technical problems of low security and large errors, this application provides an anonymous identity energy big data sharing security audit method based on data update. The technical solutions adopted are as follows:

[0005] This application proposes an anonymous identity energy big data sharing security audit method based on data update, which includes the following steps:

[0006] Obtain all energy elements of energy data and several energy entity texts corresponding to the energy elements;

[0007] Each energy entity text is divided into a preset number of text levels and energy vocabulary is obtained. The hierarchical vocabulary sequence and the vocabulary distance of each energy vocabulary are obtained according to the text level; the vocabulary level feature strength of the energy vocabulary is obtained according to the difference of the hierarchical vocabulary sequence, the vocabulary distance of the energy vocabulary and the number of energy vocabulary at the text level; the replacement vocabulary isomorphism level of the energy vocabulary is obtained according to the vocabulary level feature strength of the energy vocabulary and the similarity between the energy vocabulary; the vocabulary level value coefficient of the energy vocabulary is obtained according to the vocabulary level feature strength of the energy vocabulary and the replacement vocabulary isomorphism level;

[0008] Determine specific representative words according to the lexical level value coefficient of energy vocabulary, obtain specific vocabulary vectors according to the specific representative words, determine feature space according to the specific vocabulary vector to obtain representative score sequence; obtain specific vocabulary convergence index of energy elements according to the distance of data points in the feature space and the characteristics of representative score sequence; obtain the topic elimination strength of energy elements according to the lexical level value coefficients of different specific representative words and the number of adjacent data points of feature space data points; preset the safety factor, and obtain the element text distinctiveness of energy elements according to the topic elimination strength of energy elements, the specific vocabulary convergence index and the safety factor;

[0009] Obtain a unique ordered identifier according to the element text distinction of the energy element, and complete encryption according to the unique ordered identifier;

[0010] After the encryption results are stored, the audit results are fed back to the user through a third-party auditor to complete the data update.

[0011] Preferably, the method for obtaining the vocabulary distance between the hierarchical vocabulary sequence and each energy vocabulary according to the text hierarchy is:

[0012] The number of each type of energy vocabulary appearing at each text level in the same energy entity text is counted, recorded as the level vocabulary, and all the level vocabulary of the same energy vocabulary are organized into a level vocabulary sequence according to the order of text levels;

[0013] Any energy word that appears is recorded as the target energy word. In each text level, the energy word of the same type before the target energy word is recorded as the first previous energy word, and the energy word of the same type after the target energy word is recorded as the first subsequent energy word. The number of energy words between the target energy word and the first previous energy word is recorded as the first number, the number of energy words between the target energy word and the first subsequent energy word is recorded as the second number, and the sum of the first number and the second number is recorded as the lexical distance of the target energy word.

[0014] Preferably, the method for obtaining the vocabulary-level feature strength of energy vocabulary according to the difference of the hierarchical vocabulary sequence, the vocabulary distance of energy vocabulary and the number of energy vocabulary at the text level is:

[0015] Any energy vocabulary at each text level in the energy entity text is recorded as a central vocabulary, any text level is recorded as a central text level, and the ratio of the level vocabulary of the central vocabulary at the central text level to the number of all energy vocabulary at the central text level is recorded as the first quantity ratio;

[0016] The information entropy of the hierarchical vocabulary sequence of the central vocabulary is recorded as the first information entropy, and the information entropy of the hierarchical vocabulary sequence after the central vocabulary removes the central text level is recorded as the second information entropy; the absolute value of the difference between the first information entropy and the second information entropy is recorded as the information difference;

[0017] Obtain the lexical distance of each energy word of each central word, and record all lexical distances as the first cumulative distance; record the ratio of the information difference to the first cumulative distance as the first ratio, and record the product of the first ratio and the first quantity ratio as the lexical level feature strength of the central word at the central text level.

[0018] Preferably, the method for obtaining the replacement vocabulary homomorphic level of energy vocabulary according to the vocabulary level feature strength of energy vocabulary and the similarity between energy vocabulary is:

[0019] Record any energy vocabulary as the first energy vocabulary, calculate the normalized Google of the first energy vocabulary and another energy vocabulary, sort the normalized Google in ascending order, and select a preset number of energy vocabulary as the semantic replacement vocabulary of the first energy vocabulary;

[0020] The normalized Google of the first energy word and each semantic replacement word is recorded as the first distance; the dtw distance between the hierarchical word sequence of the first energy word and the hierarchical word sequence of each semantic replacement word is recorded as the second distance; the product of the ratio of the first distance and the second distance and the word-level feature strength of the first energy word at the text level is recorded as the first-level product, and the first-level products of the first energy word and all semantic replacement words are accumulated as the replacement word homomorphism level of the first energy word at the text level.

[0021] Preferably, the method for obtaining the vocabulary level value coefficient of energy vocabulary according to the vocabulary level feature strength of energy vocabulary and the replacement vocabulary homomorphic level is:

[0022] For any energy entity text, the lexical level feature strength of energy words in the energy entity text at the text level and the level of homomorphism of replacement words are recorded as the first level sum;

[0023] The first-level sum of the same energy vocabulary at the same text level in all energy text entities is accumulated as the first accumulated-level sum, and the ratio of the first-level sum to the first accumulated-level sum is recorded as the vocabulary-level value coefficient of the energy vocabulary at the text level in a certain energy entity text.

[0024] Preferably, the method of determining a specific representative word according to the vocabulary level value coefficient of the energy vocabulary, obtaining a specific vocabulary vector according to the specific representative word, and determining a feature space according to the specific vocabulary vector to obtain a representative score sequence is:

[0025] The energy words with the largest vocabulary level value coefficient in each text level in the energy entity text are recorded as the specific representative words;

[0026] Take all the special representative words as the input of the Word2Vec model and output the word embedding vector of each special representative word;

[0027] Use the PCA algorithm to embed the word vector to obtain the specific vocabulary vector;

[0028] Each component of the specific vocabulary vector is used as a dimension to construct a feature space, and each specific representative word corresponds to a projected data point in the feature space; all data points are used as inputs of the INFLO algorithm to obtain the INFLO score of each specific representative word;

[0029] The INFLO scores of the specific representative words of different energy entity texts at the same text level are sorted in ascending order according to the energy entity text number to form a representative score sequence.

[0030] Preferably, the method for obtaining the specific vocabulary convergence index of energy elements according to the distance of data points in the feature space and the features representing the score sequence is:

[0031] Calculate the Euclidean distances between the projected data points of all energy entity texts of the energy element at the same text level, and add up all the Euclidean distances to record them as the total Euclidean distance;

[0032] The ratio of the Herfindahl index representing the score sequence at the same text level to the total Euclidean distance at the same text level is taken as the specific vocabulary convergence index of energy elements.

[0033] Preferably, the method for obtaining the topic elimination intensity of energy elements according to the vocabulary level value coefficients of different specific representative words and the number of adjacent data points of feature space data points is:

[0034] If the nearest neighbor data point of any projected data point in the feature space and the projected data point are in the same energy entity text, the nearest neighbor data point in the same energy entity text is used as the specific word distance of the intended representative word, and the number of nearest neighbor data points is obtained by cross-validation method; the expression of topic elimination strength is:

[0035]

[0036] In the formula, G x,y (c) represents the lexical value coefficient of the unique representative word of the yth energy entity text of energy element x at the cth text level, ΔG represents the mean of the lexical value coefficients of the unique representative words of all energy entity texts of energy element x at each text level, θ x,y (c) represents the unique word distance of the yth energy entity text of energy element x at the cth text level, Y x represents the number of energy entity texts in energy element x, β1 represents the adjustment factor, α1 represents the number of text levels, exp[] represents the exponential function with natural constant as the base, τ x Represents the topic elimination intensity of energy element x.

[0037] Preferably, the method for obtaining the element text distinctiveness of the energy element according to the subject elimination intensity, the specific vocabulary convergence index and the safety factor of the energy element is:

[0038]

[0039] In the formula, w x (c) represents the specific vocabulary convergence index of energy element x at the cth text level, τ x represents the topic elimination strength of energy factor x, α1 represents the number of text levels, β1 represents the adjustment factor, β2 represents the second adjustment factor, round{} represents the rounding function, V x Indicates the element text distinctiveness of energy element x.

[0040] Preferably, the method for obtaining a unique ordered identifier according to the element text distinctiveness of the energy element is:

[0041] Clustering the energy entity text of each energy element to obtain similar text clusters; wherein the number of similar text clusters is set to be the element text distinctiveness of the energy element;

[0042] For each energy element number, an energy entity text is extracted from each similar text cluster of each energy element; the extracted energy entity text is used as an energy data block;

[0043] For each energy element, an energy data block is obtained; all energy data blocks are converted into binary data, and the binary data file composed of all energy data blocks is used as the input of the MD5 algorithm, and a unique ordered identifier is output.

[0044] The present application has the following beneficial effects: the present application realizes a security audit method for sharing energy big data with anonymous identity based on data update, obtains the vocabulary value coefficient of each energy vocabulary at different text levels of the energy entity text according to the structural hierarchical characteristics of the energy entity text, combined with word segmentation processing and normalized Google distance, reflects the degree of representation of the energy vocabulary to the text level part of the energy entity text, and its beneficial effect is to avoid the problem that different energy entity texts may have different idiomatic words under the same energy element, but the idiomatic words are semantically similar and can be replaced with each other, resulting in the energy vocabulary to The text level part of the energy entity text is actually low; based on the distinguishing features of the energy entity text under the same energy element, combined with the security factor λ, the element text distinction of each energy element is constructed, reflecting the value of the energy element after the energy data block is divided and encrypted, as the number of clustering clusters of the AGNES clustering algorithm, adaptively adjusted according to the energy entity text, accurately classified the energy data of the energy owner, reduced the data correlation between the energy data blocks, and improved the ciphertext security of the subsequent encryption of the energy data blocks, preventing the agent from being attacked, resulting in the problem of energy data leakage of the energy owner. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present application or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0046] Figure 1 A flow chart of a method for anonymous identity energy big data sharing security audit based on data update provided in one embodiment of the present application;

[0047] Figure 2 Schematic diagram of user registration and proxy authorization;

[0048] Figure 3 It is a diagram of energy element composition;

[0049] Figure 4 A flowchart for encrypting the data of the data owner;

[0050] Figure 5 The process of data encryption, uploading, security auditing and updating. DETAILED DESCRIPTION

[0051] In order to further explain the technical means and effects adopted by the present application to achieve the predetermined invention purpose, the following, in combination with the accompanying drawings and preferred embodiments, describes in detail the specific implementation method, structure, features and effects of the anonymous identity energy big data sharing security audit method based on data update proposed in the present application. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0052] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0053] An anonymous identity energy big data sharing security audit method based on data update implementation example:

[0054] The specific scheme of the anonymous identity energy big data sharing security audit method based on data update provided by this application is described in detail below with reference to the accompanying drawings.

[0055] See also Figure 1 , which shows a flow chart of an anonymous identity energy big data sharing security audit method based on data update provided by an embodiment of the present application, the method comprising the following steps:

[0056] Step S1, initialize the system and calculate the master private key information and public parameters.

[0057] The cross-domain management center sets a safety factor λ through a pseudo-random generator; defines a residual class ring Z on the module q q The elliptic curve E on which q is a large prime number, P is a point on the elliptic curve E with a prime number q, and is an additive cyclic group G = A generator of the key generation center KGC from Z q Select a random number s as the master private key of the key generation center KGC and calculate P Pub =sP obtains the master public key P of the key generation center KGC Pub ; The cross-domain management center sets three anti-collision hash functions h1, h2, and h3; The cross-domain management center discloses public parameters Para = (P, q, E, P Pub ,h1,h2,h3), the master private key s of KGC needs to be kept secret. In this embodiment, the method for obtaining public parameters and master private key information can adopt the method in patent CN108712259B, and the implementer can also adopt other methods, which are not particularly limited in this embodiment.

[0058] The cross-domain management center is a center for managing cross-domain data. It can help enterprises manage and coordinate energy data from different sources to achieve effective use and sharing of energy data. It usually includes functions such as data integration, data storage, data processing, data security and data sharing, which can help enterprises achieve efficient management and utilization of energy data, improve the quality and reliability of energy data, and reduce the cost and risk of energy data management. However, the cross-domain management center is a semi-trusted entity, and the cross-domain management center or external attackers may completely destroy the integrity of outsourced energy data. Therefore, it is very necessary for TPA to periodically audit the integrity of outsourced data.

[0059] The Key Generation Center (KGC) is an authoritative and fully trusted entity, which is mainly responsible for initializing some system parameters and generating the user's public and private key pairs based on the user's uniquely identifiable identity.

[0060] In step S2, the original data owner and the agent act as users, generate anonymous identities and signature private keys for them, register their identities, and authorize the agents.

[0061] The original data owner and agent are users, and the key generation center KGC is from Z q Select a random number from , and get a binary array based on the random number;

[0062] The original data owner is from Z q A random number is selected from the binary array, and the digital signature of the authorization letter is obtained based on the random number and the binary array. The digital signature is sent to the agent for authorization. Figure 2 shown.

[0063] Among them, the original data owner (Original Data Owner): On the one hand, the data owner will entrust a trusted agent to help him / her complete a series of data processing tasks. On the other hand, the data owner will also hire a TPA to periodically verify the integrity of the energy data stored in the cross-domain management center on his / her behalf. Procy: As a proxy worker, once a proxy authorization agreement is reached with the original data owner, the Procy will comply with the provisions of the agreement to perform a series of data processing transactions such as encryption, compression and upload of energy data.

[0064] Step S3, grouping the data provided by the data owner and encrypting the grouped data blocks.

[0065] The original data owner sends the energy data he owns to the agent through a secure channel according to the provisions of the agent agreement. The agent needs to divide the data file into n energy data blocks and determine a unique identification tag for each energy data file. The traditional division method is to select continuous fixed-length energy data from the energy data file as energy data blocks, which leads to a large data correlation between energy data blocks and the problem of low security of the ciphertext information obtained by encrypting the subsequent energy data blocks separately.

[0066] Step S3.1, obtaining the energy entity text of the original data owner.

[0067] The original data owners are all anonymous energy data entities. Through the secure channel between the agent and the original data owner, the energy data of the original data owner is obtained, and different types of energy data are recorded as energy elements. In this embodiment, energy elements take energy project analysis reports, energy efficiency assessment reports, energy facility operation logs, energy facility maintenance records, energy facility failure reports, and energy transaction records as examples. Energy elements are composed of multiple energy entity texts, such as Figure 3 At the same time, in order to avoid errors in the data acquisition process, the acquired data is pre-processed using data cleaning technology, wherein the data cleaning technology is a well-known technology, and the specific process will not be repeated.

[0068] Step S3.2, obtaining the vocabulary level value coefficient of energy vocabulary according to the distance and similarity between energy vocabulary in the energy entity text and the difference between the hierarchical vocabulary sequences;

[0069] Energy entity text has obvious structural hierarchical characteristics. The energy entity data in different parts of the same energy entity text have different emphases. The appearance of vocabulary also has obvious hierarchical characteristics. Specific vocabulary will only appear in specific parts of the energy entity text. For words that appear evenly in the entire energy entity text, they are usually the theme information of the text or common words for energy elements.

[0070] The energy entity text of each energy element is equally divided according to the length, and is divided into α1 text levels. The data length of each text level is the same. In this embodiment, α1=10. All energy entity texts are used as input and processed using Jieba word segmentation. All segmented words are recorded as energy vocabulary. The number of all types of words in the energy vocabulary is counted, recorded as Nom. The number of each type of energy vocabulary in each text level in the same energy entity text is counted, recorded as the level vocabulary. All level vocabulary of the same energy vocabulary is organized into a level vocabulary sequence according to the order of the text level.

[0071] In the energy entity text of the original data owner, there are different energy terms, but they have similar semantics and can be replaced with each other. Taking energy term a as an example, the normalized Google distance between energy term a and all other types of energy terms is calculated, and they are sorted in ascending order. The first α2 energy terms in the sequence are obtained as semantic replacement terms for energy term a. α2 is 3 in this implementation. The normalized Google distance is a well-known technology, and the specific process is not repeated here.

[0072] For any energy vocabulary, obtain the information entropy of the hierarchical vocabulary sequence of the energy vocabulary. For the hierarchical vocabulary sequence of the energy vocabulary, calculate the information entropy of the hierarchical vocabulary sequence after reducing one level of vocabulary. For each energy vocabulary, the energy vocabulary appears multiple times in each text level. Any energy vocabulary that appears is recorded as the target energy vocabulary. In each text level, the previous energy vocabulary of the same type of the target energy vocabulary is recorded as the first front energy vocabulary, and the next energy vocabulary of the same type of the target energy vocabulary is recorded as the first back energy vocabulary. The number of energy vocabulary between the target energy vocabulary and the first front energy vocabulary is recorded as the first number, and the number of energy vocabulary between the target energy vocabulary and the first back energy vocabulary is recorded as the second number. The sum of the first number and the second number is recorded as the vocabulary distance of the target energy vocabulary. For example: the text level is abcbadca, each letter represents an energy vocabulary, the middle a is the target energy vocabulary, the front a is the first front energy vocabulary, and the back a is the first back energy vocabulary. Then the first number is 3, the second number is 2, and the vocabulary distance of the target energy vocabulary is 5. It is worth noting that if the target energy vocabulary does not have a first preceding energy vocabulary or a first succeeding energy vocabulary, only the existing one will be counted.

[0073] The lexical level feature strength of energy vocabulary at the text level is obtained based on the information entropy difference of the hierarchical vocabulary sequence of each energy vocabulary before and after removing a certain level of vocabulary and the lexical distance of energy vocabulary in the text level corresponding to the level of vocabulary. The formula is as follows:

[0074]

[0075] In the formula, N represents the hierarchical vocabulary of energy word a in the yth energy entity text at the cth text level, y (c) represents the total number of energy words at the cth text level in the yth energy entity text, Ent a The information entropy of the hierarchical vocabulary sequence representing energy vocabulary a, Ent a (c) represents the information entropy of the hierarchical vocabulary sequence of energy vocabulary a after removing the cth text level, ρ a,n′ (c) represents the lexical distance of the n′th energy word a at the cth text level, β1 represents the adjustment factor, which is used to place the denominator at 0. Represents the lexical level feature strength of energy word a at the cth text level in the yth energy entity text.

[0076] When the proportion of energy vocabulary in the hierarchical vocabulary is larger, the energy vocabulary interval of the same type of energy vocabulary in the same text level is shorter, that is, The bigger, The smaller the value is, the higher the frequency of energy vocabulary appears at the text level and the denser the distribution is at the text level. At the same time, when the level of energy vocabulary is more affected by the text level, that is, Ent a -Ent a The larger (c) is, the more distinctive the energy vocabulary is at the text level from other text levels of the same energy entity text. This can eliminate the problem of large errors in the intensity of vocabulary-level features caused by the high frequency of occurrence of the text's subject information or common words for energy elements at each text level. The more obvious and accurate the structural-level features of energy vocabulary are, the more The bigger.

[0077] According to the lexical level feature strength of the semantic replacement vocabulary of the energy vocabulary and the similarity between the energy vocabulary and its semantic replacement vocabulary, the replacement vocabulary homomorphic level of the energy vocabulary is obtained. The formula is as follows:

[0078]

[0079] Where Gog(a,a′) represents the normalized Google distance between energy word a and its a′th semantic replacement word, dtw(ω a ,ω a′ ) represents the DTW distance between the hierarchical vocabulary sequence of energy vocabulary a and its a′th semantic replacement vocabulary, ω a Represents the hierarchical vocabulary sequence of energy vocabulary a, ω a′ represents the hierarchical vocabulary sequence of the a′th semantic replacement vocabulary of energy vocabulary a, represents the lexical level feature strength of the a′th semantic replacement word of energy word a in the yth energy entity text at the cth text level, α2 represents the number of semantic replacement words, β1 represents the adjustment factor, It represents the degree of homomorphism of the replacement vocabulary of energy vocabulary a in the yth energy entity text at the cth text level.

[0080] Among them, when the hierarchical feature strength of the semantic replacement vocabulary of energy vocabulary is greater, that is, The larger the value, the more obvious the structural characteristics of the semantic replacement vocabulary are at the same text level. At the same time, when the semantics of energy vocabulary and semantic replacement vocabulary are more similar, the distribution of energy vocabulary under the same energy entity text is closer, that is, the larger the value of Gog(a,a′), the greater the value of dtw(ω a ,ω a′ ) is smaller, The larger it is, the more energy words and semantic replacement words can be replaced with each other from macroscopic natural semantics to actual energy entity text vocabulary distribution, and the more likely other energy entity texts are to use semantic replacement words. The bigger.

[0081] The lexical level value coefficient of energy vocabulary at the text level is obtained according to the lexical level feature strength of energy vocabulary and the isomorphic level of replacement vocabulary. The formula is as follows:

[0082]

[0083] In the formula, represents the lexical level feature strength of energy word a in the yth energy entity text at the cth text level, represents the degree of homomorphism of the replacement vocabulary of energy vocabulary a in the yth energy entity text at the cth text level, represents the lexical level feature strength of energy word a in the vth energy entity text at the cth text level, represents the degree of homomorphism of the replacement vocabulary of energy vocabulary a in the vth energy entity text at the cth text level, Y x represents the number of energy entity texts under the x-th energy element, β1 represents the adjustment factor, Represents the vocabulary-level value coefficient of energy vocabulary a in the y-th energy entity text at the c-th text level.

[0084] Among them, when the structural hierarchical characteristics of the same energy vocabulary in other energy entity texts are less obvious, that is, The smaller, The larger it is, the more the energy vocabulary can represent the text level, and the higher the level value of the energy vocabulary. The larger the value is, the more the energy vocabulary represents the text-level part of the energy entity text.

[0085] Step S3.3, determine the representative score sequence according to the lexical level value coefficient of the energy vocabulary, obtain the specific vocabulary convergence index according to the characteristics of the representative score sequence, obtain the topic elimination strength by combining the lexical level value coefficients of different specific representative words, and obtain the element text distinctiveness according to the specific vocabulary convergence index and the topic elimination strength.

[0086] The energy entity data of the same energy element of the original data owner belongs to the same type, and there are differences in the similarity of energy entity texts of different energy elements. For example, energy facilities usually follow strict operating standards and processes. The operating activities of energy facilities may be very similar every day. In addition, modern energy facilities are highly dependent on automated control systems. The standardized maintenance and operation procedures of energy facilities reduce the differences between the operation logs of energy facilities. The energy entity texts of such energy elements are low in distinction and small in difference, and the data encryption value is low; however, there are diverse trading participants and many types of transactions in the energy market. Energy transactions are greatly affected by the market and have the characteristics of dynamic fluctuations. The energy entity texts of such energy elements are high in distinction and large in difference, and the data encryption value is high. Therefore, different energy elements have different encryption priorities.

[0087] The energy words with the largest vocabulary level value coefficient in each text level in the energy entity text are recorded as special representative words, and all special representative words are used as inputs of the Word2Vec model, where the dimension of the embedding vector is α3, which is 50 in this embodiment, and the word embedding vectors of each special representative word are output. Then, the word embedding vectors of the special representative words are used as inputs of the PCA principal component analysis algorithm, and the special vocabulary vectors of the special representative words are output. The dimension of the special vocabulary vector is α4, which is 3 in this embodiment. The PCA algorithm and the Word2Vec model are both well-known technologies and will not be described in detail here.

[0088] Furthermore, each component in the specific vocabulary vector of the specific representative word is used as a dimension to construct a feature space with a dimension of α4, and the specific vocabulary vectors of all specific representative words are projected into the α4-dimensional feature space to obtain the feature space projection data points of the specific representative words. All the projected data points in the α4-dimensional feature space are used as the input of the INFLO algorithm, where the number of nearest neighbor data points k is obtained by cross-validation, and the INFLO score of each specific representative word is output. The INFLO scores of the specific representative words of different energy entity texts at the same text level are sorted in ascending order according to the energy entity text number to form a representative score sequence, that is, each text level corresponds to a representative score sequence.

[0089] The Herfindahl index of each representative score sequence is calculated, and the specific vocabulary convergence index of energy elements at the corresponding text level is obtained according to the Herfindahl index of the representative score sequence and the projection data points of different energy entity texts at the text level. The formula is as follows:

[0090]

[0091] In the formula, h x (c) The Herfindahl index representing the score sequence of energy element x at the cth text level, L x,y (c) represents the projection data point of the yth energy entity text in the energy element x at the cth text level, L x,y′ (c) represents the projection data point of the y′th energy entity text in the energy element x at the cth text level, dis[L x,y (c),L x,y′ (c)] represents the projected data point L x,y (c) and L x,y′ (c) The Euclidean distance, Y x represents the number of energy entity texts in energy element x, β1 represents the adjustment factor, and w x (c) represents the specific vocabulary aggregation index of energy element x at the cth text level.

[0092] Among them, at the same text level, the larger the distance between the projected data points of any energy entity text in the feature space, the more discrete the representative score sequence is, that is, The bigger, h x The smaller (c) is, the less similar the specific representative words of the energy entity texts at the same text level are, the more discrete the distribution in the feature space is, and the less similar the energy entity texts of the energy elements are. x (c) Smaller.

[0093] For the projected data point L x,y (c) The number of the nearest neighbor data points of the projected data point in the feature space belonging to the projected data point of the y-th energy entity text is recorded as the specific word distance of the specific representative word The topic elimination intensity of energy elements is obtained according to the vocabulary level value coefficient of the specific representative word and the specific word distance. The formula is as follows:

[0094]

[0095] In the formula, G x,y (c) represents the vocabulary-level value coefficient of the specific representative word of the y-th energy entity text of energy element x at the c-th text level, ΔG represents the mean of the vocabulary-level value coefficients of the specific representative words of all energy entity texts of energy element x at each text level, represents the unique word distance of the yth energy entity text of energy element x at the cth text level, Y x represents the number of energy entity texts in energy element x, β1 represents the adjustment factor, α1 represents the number of text levels, exp[] represents the exponential function with natural constant as the base, τ x Represents the topic elimination strength of energy element x.

[0096] Among them, when the vocabulary level value coefficient is larger, the other specific representative words of the same energy entity text in the feature space are not the nearest neighbor data points, that is, G x,y (c)-The larger the ΔG, The smaller, The larger the value is, the more the specific representative word can represent the text level, the less the specific representative word can represent other text levels at the same time, and the less likely the specific representative word is the subject word of the energy entity text. x The bigger.

[0097] The element text distinction of energy elements is obtained according to the subject elimination intensity, specific vocabulary convergence index and safety factor of energy elements. The formula is as follows:

[0098]

[0099] In the formula, w x (c) represents the specific vocabulary convergence index of energy element x at the cth text level, τ x represents the topic elimination intensity of energy element x, α1 represents the number of text levels, β1 represents the adjustment factor, β2 represents the second adjustment factor, and in this embodiment, it is set to 1 to prevent the situation where the base of the composite function is less than 1 and the monotonicity cannot be determined, round{} represents the rounding function, V x Indicates the element text distinctiveness of energy element x.

[0100] Among them, the greater the security factor of the cross-domain management center, that is, the greater the λ, the more it can avoid the problem of small number of energy data blocks, large data correlation, and low security of encrypted energy data blocks. At this time, more data blocks should be divided for data owners. The energy entity texts of this type of energy elements are highly distinguished and have large differences. V x The bigger.

[0101] At this point, the element text distinctiveness of each energy element is obtained. The larger the value, the more distinctive and different the energy entity text is. The higher the encryption value is when encrypting, and it needs to be divided into different energy data blocks to provide encryption security.

[0102] Step S3.4, clustering is performed according to the distinctiveness of the element text to obtain the energy data block, thereby obtaining a unique ordered identifier.

[0103] In the α4-dimensional feature space, the text levels are sorted in ascending order, and the projection data points corresponding to each text level are connected with the projection data points corresponding to the next text level. The projection data points are used as graph nodes, and the connection results are used as the undirected graph of energy entity texts. The undirected graph corresponding to all energy entity texts of energy element x is used as the input of the AGNES clustering algorithm, and the WL (Weisfeiler-Lehman) Kernel distance is used as the measurement method. The element text distinctiveness V x As the number of clusters, output the cluster of energy element x for energy entity text. Use the same method to obtain the cluster corresponding to each energy element, and record the cluster as a similar text cluster.

[0104] Each round of energy data extraction starts from the energy element numbered 1 and ends with the energy element with the maximum number. An energy entity text is randomly extracted from all similar text clusters of each energy element, and the energy data blocks of each round are formed in the order of extraction. For example, in the first round of energy data extraction, starting from the energy element numbered 1, an energy entity text f is randomly extracted from the first similar text cluster. 1,1 , and then randomly extract an energy entity text f from the second similar text cluster 1,2 , and so on, until the energy entity text f of the last similar text cluster in the energy element with the maximum number 1,σ1 So far, σ1 is the total number of energy entity texts extracted in the first round. According to the order of extraction, the energy data blocks of the first round are formed in sequence. f1 = [f 1,1 ,f 1,2 ,f 1,3 ,…,f 1,σ1 ]. Further, the number of energy data blocks of the original energy owner is recorded as n, and all energy data blocks are converted into binary data to form a binary data file F = [f1, f2, f1, ... f n ] T , T is the transposition symbol, and the binary data converted from the n energy data blocks are used as the input of MD5 (Message Digest Algorithm 5), and the output is a unique ordered identifier tag for each energy data block.

[0105] Step S3.5, complete encryption according to the unique ordered identifier.

[0106] The agent obtains each energy data block and the corresponding unique ordered identifier tag of the original data owner, and then uses a symmetric encryption algorithm to encrypt the energy data blocks one by one according to the identifier, and then encrypts the data file F into a ciphertext F′, wherein the symmetric encryption algorithm in this embodiment adopts the 3DES encryption algorithm, and the agent calls a pseudo-random number generator to generate a corresponding random variable for each energy data block, and uses a pseudo-random generation function to generate a corresponding random number for each energy data block; the agent uses the sub-vector value of the random variable and the random number to compress the data block m times to generate a homomorphic message authentication code σ i .

[0107] The above encryption process is as follows Figure 4 shown.

[0108] Step S4, generating a digital signature of the data block, and uploading the corresponding data block to the cross-domain management center for storage.

[0109] The agent selects a random number, generates a digital signature for the data block, and uploads the corresponding data block to the cross-domain management center for verification. If the verification is successful, it means that the storage information uploaded by the user is valid, and the cross-domain management center immediately stores it. Otherwise, it is rejected and the cross-domain management center sends a rejection signal to the agent.

[0110] Step S5: The third-party auditor TPA generates challenge information based on relevant data and parameters.

[0111] After receiving the audit request from the agent, the third-party auditor TPA will verify it. If the authorization information is verified to be valid, a challenge message will be generated and sent to the cross-domain management center. If the authorization information is verified to be invalid, the corresponding audit request will be rejected and the generation of the challenge message will be refused.

[0112] Step S6: The cross-domain management center audits and verifies the corresponding data based on the challenge information.

[0113] After receiving the challenge information sent by the third-party auditor, the cross-domain management center calculates the combined data block and the aggregate signature, and then sends the proof response information to the third-party auditor.

[0114] Step S7: The third-party auditor TPA verifies the audit certification information.

[0115] The third-party auditor TPA verifies the audit certificate information of the cross-domain management center. If the verification is successful, it means that the energy data stored by the original data owner in the cross-domain management center is complete. Otherwise, TPA will immediately issue an emergency notice to the original data owner, key generation center, agent, and cross-domain management center, indicating that the data file is no longer available.

[0116] Step S8, the audit results are fed back to the user, who may make some modifications and update the data after completion.

[0117] The audit results are fed back to the user, who may make some modification operations and update the data after completion. This step algorithm is mainly executed by the user and verified by the cross-domain management center. The users here include the original data owner and the agent, and mainly include insert data update, delete data update and modify data update.

[0118] The first one is insert data update. The algorithm performs an insert operation on a data block, inputs the insert position, the insert data block and its unique identifier, and outputs the insert result.

[0119] If the user needs to insert a new data block into an existing data block, first generate the tag of the new data block, then construct an insertion request and send it to the cross-domain management center. After receiving the request, the cross-domain management center first verifies and stores the data block tag, stores the data block in the corresponding position, completes the verification, and updates the relevant information.

[0120] The second type is deletion data update. The algorithm performs a deletion operation on the data block, inputs the location of the deleted data block, and outputs the deletion result.

[0121] If a user needs to delete a data block, the user constructs a deletion request and sends it to the cross-domain management center. After receiving the request, the cross-domain management center directly locates the data block to be deleted and deletes it, and updates the relevant information after completing the verification process.

[0122] The third type is to modify data and update. The algorithm performs a modification operation on the data block, inputs the modified data block location, the new data block and its unique identifier, and outputs the modification result.

[0123] If the user needs to modify a data block, first generate a label for the modified new data block, then construct a modification request and send it to the cross-domain management center. After receiving the request, the cross-domain management center first verifies the correctness of the label. After the verification is successful, the modification of the data block and its label is completed, and the relevant information is updated after the verification is completed. If the verification fails, the user is required to recalculate, thereby completing the data update. The process of data encryption, uploading, data security auditing, and data update in steps s3 to s8 is as follows: Figure 5 shown.

[0124] It should be noted that the sequence of the above embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0125] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. An anonymous identity energy big data sharing security audit method based on data update, characterized by: The method comprises the following steps: Obtain all energy elements of energy data and several energy entity texts corresponding to the energy elements; Each energy entity text is divided into a preset number of text levels and energy vocabulary is obtained. The hierarchical vocabulary sequence and the vocabulary distance of each energy vocabulary are obtained according to the text level; the vocabulary level feature strength of the energy vocabulary is obtained according to the difference of the hierarchical vocabulary sequence, the vocabulary distance of the energy vocabulary and the number of energy vocabulary at the text level; the replacement vocabulary isomorphism level of the energy vocabulary is obtained according to the vocabulary level feature strength of the energy vocabulary and the similarity between the energy vocabulary; the vocabulary level value coefficient of the energy vocabulary is obtained according to the vocabulary level feature strength of the energy vocabulary and the replacement vocabulary isomorphism level; Determine specific representative words according to the lexical level value coefficient of energy vocabulary, obtain specific vocabulary vectors according to the specific representative words, determine feature space according to the specific vocabulary vector to obtain representative score sequence; obtain specific vocabulary convergence index of energy elements according to the distance of data points in the feature space and the characteristics of representative score sequence; obtain the topic elimination strength of energy elements according to the lexical level value coefficients of different specific representative words and the number of adjacent data points of feature space data points; preset the safety factor, and obtain the element text distinctiveness of energy elements according to the topic elimination strength of energy elements, the specific vocabulary convergence index and the safety factor; Obtain a unique ordered identifier according to the element text distinction of the energy element, and complete encryption according to the unique ordered identifier; After the encryption results are stored, the audit results are fed back to the user through a third-party auditor to complete the data update.

2. The anonymous identity energy big data sharing security audit method based on data update as claimed in claim 1 is characterized in that: The method for obtaining the vocabulary distance between the hierarchical vocabulary sequence and each energy vocabulary according to the text hierarchy is: The number of each type of energy vocabulary appearing at each text level in the same energy entity text is counted, recorded as the level vocabulary, and all the level vocabulary of the same energy vocabulary are organized into a level vocabulary sequence according to the order of text levels; Any energy word that appears is recorded as the target energy word. In each text level, the energy word of the same type before the target energy word is recorded as the first previous energy word, and the energy word of the same type after the target energy word is recorded as the first subsequent energy word. The number of energy words between the target energy word and the first previous energy word is recorded as the first number, the number of energy words between the target energy word and the first subsequent energy word is recorded as the second number, and the sum of the first number and the second number is recorded as the lexical distance of the target energy word.

3. The anonymous identity energy big data sharing security audit method based on data update as claimed in claim 2 is characterized in that: The method for obtaining the lexical level feature strength of energy vocabulary according to the difference of hierarchical vocabulary sequences, the lexical distance of energy vocabulary and the number of energy vocabulary at the text level is: Any energy vocabulary at each text level in the energy entity text is recorded as a central vocabulary, any text level is recorded as a central text level, and the ratio of the level vocabulary of the central vocabulary at the central text level to the number of all energy vocabulary at the central text level is recorded as the first quantity ratio; The information entropy of the hierarchical vocabulary sequence of the central vocabulary is recorded as the first information entropy, and the information entropy of the hierarchical vocabulary sequence after the central vocabulary removes the central text level is recorded as the second information entropy; the absolute value of the difference between the first information entropy and the second information entropy is recorded as the information difference; Obtain the lexical distance of each energy word of each central word, and record all lexical distances as the first cumulative distance; record the ratio of the information difference to the first cumulative distance as the first ratio, and record the product of the first ratio and the first quantity ratio as the lexical level feature strength of the central word at the central text level.

4. The anonymous identity energy big data sharing security audit method based on data update as claimed in claim 1 is characterized in that: The method for obtaining the replacement vocabulary homomorphic level of energy vocabulary according to the vocabulary level feature strength of energy vocabulary and the similarity between energy vocabulary is as follows: Record any energy vocabulary as the first energy vocabulary, calculate the normalized Google of the first energy vocabulary and another energy vocabulary, sort the normalized Google in ascending order, and select a preset number of energy vocabulary as the semantic replacement vocabulary of the first energy vocabulary; The normalized Google distance between the first energy word and each semantic replacement word is recorded as the first distance; the dtw distance between the hierarchical word sequence of the first energy word and the hierarchical word sequence of each semantic replacement word is recorded as the second distance; the product of the ratio of the first distance and the second distance and the word-level feature strength of the first energy word at the text level is recorded as the first-level product, and the first-level products of the first energy word and all semantic replacement words are accumulated as the replacement word homomorphism level of the first energy word at the text level.

5. The anonymous identity energy big data sharing security audit method based on data update as claimed in claim 1 is characterized in that: The method for obtaining the vocabulary level value coefficient of energy vocabulary according to the vocabulary level feature strength of energy vocabulary and the level of homomorphism of replacement vocabulary is as follows: For any energy entity text, the lexical level feature strength of energy vocabulary in the energy entity text at the text level and the level of homomorphism of replacement vocabulary are recorded as the first level sum; The first-level sum of the same energy vocabulary at the same text level in all energy text entities is accumulated as the first accumulated-level sum, and the ratio of the first-level sum to the first accumulated-level sum is recorded as the lexical-level value coefficient of the energy vocabulary at the text level in a certain energy entity text.

6. The anonymous identity energy big data sharing security audit method based on data update as claimed in claim 1 is characterized in that: The method of determining the specific representative words according to the vocabulary level value coefficient of the energy vocabulary, obtaining the specific vocabulary vector according to the specific representative words, and determining the feature space according to the specific vocabulary vector to obtain the representative score sequence is: The energy words with the largest vocabulary level value coefficient in each text level in the energy entity text are recorded as the specific representative words; Take all the special representative words as the input of the Word2Vec model and output the word embedding vector of each special representative word; Use the PCA algorithm to embed the word vector to obtain the specific vocabulary vector; Each component of the specific vocabulary vector is used as a dimension to construct a feature space, and each specific representative word corresponds to a projected data point in the feature space; all data points are used as inputs of the INFLO algorithm to obtain the INFLO score of each specific representative word; The INFLO scores of the specific representative words of different energy entity texts at the same text level are sorted in ascending order according to the energy entity text number to form a representative score sequence.

7. The anonymous identity energy big data sharing security audit method based on data update as claimed in claim 6 is characterized in that: The method for obtaining the specific vocabulary convergence index of energy elements according to the distance of data points in the feature space and the features representing the score sequence is: Calculate the Euclidean distances between the projected data points of all energy entity texts of the energy element at the same text level, and add up all the Euclidean distances to record them as the total Euclidean distance; The ratio of the Herfindahl index representing the score sequence at the same text level to the total Euclidean distance at the same text level is taken as the specific vocabulary convergence index of energy elements.

8. The anonymous identity energy big data sharing security audit method based on data update as claimed in claim 6 is characterized in that: The method for obtaining the topic elimination intensity of energy elements according to the vocabulary level value coefficients of different specific representative words and the number of adjacent data points of feature space data points is: If the nearest neighbor data point of any projected data point in the feature space and the projected data point are in the same energy entity text, the nearest neighbor data point in the same energy entity text is used as the specific word distance of the intended representative word, and the number of nearest neighbor data points is obtained by cross-validation method; the expression of topic elimination strength is: In the formula, G x,y (c) represents the vocabulary-level value coefficient of the specific representative word of the y-th energy entity text of energy element x at the c-th text level, ΔG represents the mean of the vocabulary-level value coefficients of the specific representative words of all energy entity texts of energy element x at each text level, represents the unique word distance of the yth energy entity text of energy element x at the cth text level, Y x represents the number of energy entity texts in energy element x, β1 represents the adjustment factor, α1 represents the number of text levels, exp[] represents the exponential function with natural constant as the base, τ x Represents the topic elimination strength of energy element x.

9. The anonymous identity energy big data sharing security audit method based on data update as claimed in claim 1 is characterized in that: The method for obtaining the element text distinction of energy elements according to the subject elimination intensity, the specific vocabulary convergence index and the safety factor of energy elements is: In the formula, w x (c) represents the specific vocabulary convergence index of energy element x at the cth text level, τ x represents the topic elimination strength of energy factor x, α1 represents the number of text levels, β1 represents the adjustment factor, β2 represents the second adjustment factor, round{} represents the rounding function, V x represents the element text distinction of energy element x, and λ represents the preset safety factor.

10. The anonymous identity energy big data sharing security audit method based on data update as claimed in claim 1 is characterized in that: The method for obtaining a unique ordered identifier according to the element text distinction of the energy element is: Clustering the energy entity text of each energy element to obtain similar text clusters; wherein the number of similar text clusters is set to be the element text distinctiveness of the energy element; For each energy element number, an energy entity text is extracted from each similar text cluster of each energy element; the extracted energy entity text is used as an energy data block; For each energy element, an energy data block is obtained; all energy data blocks are converted into binary data, and the binary data file composed of all energy data blocks is used as the input of the MD5 algorithm, and a unique ordered identifier is output.

Images