Management optimization method based on automatic detection of low-voltage metering box

Abstract

The invention provides a management optimization method based on automatic detection of a low-voltage metering box, and belongs to the technical field of intelligent power grids, and the method comprises the following steps: S1, storing a user password based on an enhanced Hash matrix encryption algorithm, controlling user access and operation authority based on an authority management optimization algorithm, and establishing a detection data management library; s2, performing decomposition and analysis processing on the keywords after the basic preprocessing, and performing optimization processing on the keywords after the decomposition and analysis processing; s3, based on a weighted keyword matching algorithm, calculating similarities between the keywords and the detection data, sorting according to the similarities, and returning the detection data with the highest similarity to the user; and S4, recording the keyword, the matching process and the result of each query in a system, and constructing a data log. According to the method and the device, the access safety of the user can be effectively improved, the password storage safety is effectively enhanced, and the method and the device have relatively high query efficiency and accuracy and relatively good user experience.

Description

Technical Field

[0001] The present invention belongs to the technical field of smart grids, and in particular relates to a management optimization method based on automatic detection of a low-voltage meter box. Background Art

[0002] The low-voltage meter box is an important device for measuring electric energy in the power system. With the development of smart grid, the demand for detection and management of low-voltage meter boxes is getting higher and higher. However, the current low-voltage meter box detection data management system has many deficiencies in data storage, backup, recovery, query, matching and security, which affects the overall performance of the system and user experience.

[0003] In terms of data matching, existing matching algorithms are mostly based on simple string matching or traditional similarity calculation methods, which are difficult to cope with complex query requirements, and the accuracy and reliability of matching results are low. Especially when faced with massive detection data, the performance bottleneck of traditional matching algorithms is particularly prominent, making it difficult to meet the needs of real-time query and high-precision matching.

[0004] In terms of data security and access control, the existing user authentication mechanisms mostly use simple username and password authentication, which is less secure and vulnerable to brute force cracking and information leakage. Although some multi-factor authentication technologies have improved security, they are complex to implement and provide a poor user experience. In addition, the existing role-based access control model and permission management mechanism are not refined enough, making it difficult to achieve strict control and audit of data access. Summary of the invention

[0005] In order to solve the above problems existing in the prior art, a management optimization method based on automatic detection of low-voltage metering boxes is provided.

[0006] The technical solution adopted by the present invention to solve its technical problem is:

[0007] This technical solution proposes a management optimization method based on automatic detection of low-voltage metering boxes, including the following steps:

[0008] S1: Store user passwords based on the enhanced hash matrix encryption algorithm, control user access and operation permissions based on the permission management optimization algorithm, and establish a detection data management library;

[0009] S2: Perform basic preprocessing on the query keywords, decompose and analyze the keywords after the basic preprocessing, and optimize the keywords after the decomposition and analysis;

[0010] S3: Based on the weighted keyword matching algorithm, the similarity between the keyword and the test data is calculated, sorted according to the similarity, and the test data with the highest similarity is returned to the user;

[0011] S4: Record the keywords, matching process and results of each query in the system to build a data log.

[0012] Preferably, in S1, a database architecture is designed based on a regularization method, data tables and relationships are defined, partitioning and sharding strategies are implemented, and data is distributed on multiple storage nodes.

[0013] Preferably, in S1, a distributed data storage method based on a hash algorithm is selected, a Gossip protocol is used to synchronize the states between nodes, and a distributed consensus algorithm is used to achieve distributed consistency.

[0014] Preferably, in S1, a monitoring tool is used to regularly monitor database performance, an automatic expansion method is used to dynamically adjust resource allocation, sharding and replication strategies are adjusted, and a backup tool and a cloud storage service are used to implement a backup strategy to achieve data backup.

[0015] Preferably, in S1, a multi-factor authentication technology is used to design a user authentication mechanism, and the multi-factor authentication technology includes using SMS verification codes, email verification and hardware tokens.

[0016] Preferably, in S1, the enhanced hash matrix encryption algorithm includes:

[0017] S11: Slice the user password and divide it into n substrings {p1, p2, ..., p n};

[0018] S12: Construct a randomly generated hash matrix H of size (m×m), the elements of the H matrix H ij Generated by the hash function hash(), where H ij = hash(s ij ), s ij is the random seed;

[0019] S13: For each substring p i Perform hash calculation to get the hash value h i ,h i = hash(p i );

[0020] S14: Each h i Convert to matrix M i The size is (m×m), and the matrix M i Perform normalization:

[0021] S15: Calculate the final hash value H final , the formula is as follows:

[0022]

[0023] In the formula, H final is the final hash value, n is the number of password fragments, m is the size of the hash matrix, M i,j,k is the i-th codon string p i The hash matrix M i The j-th row and k-th column element of H j,k is the jth row and kth column element of the randomly generated hash matrix H, log(i+1) is the natural logarithm of (i+1), which is used for normalization processing, M i,l,l is the i-th codon string p i The hash matrix M i diagonal elements.

[0024] Preferably, in S1, the rights management optimization algorithm includes:

[0025] S21: Assign a permission matrix P of size (k×k) to each role. The element P of the matrix P is ij It is the permission value of the role in a certain permission dimension;

[0026] S22: Classify each data access request R and set a priority matrix Q of size (k×k). The element Q of the Q matrix is ij The priority of data access requests on a certain dimension;

[0027] S23: When a user initiates a data access request, the user's role and specific request information are obtained, and the matching degree M between the role and the request is calculated. The formula is as follows:

[0028]

[0029] Where M is the matching degree between the role and the request, which is used to determine the user's access rights to the data, k is the size of the permission matrix P and the priority matrix Q, and P ij is the i-th row and j-th column element of the permission matrix P, representing the permission value of the role, Q ij is the i-th row and j-th column element of the priority matrix Q, which represents the priority value of the data access request. Log(Q ij +1) for (Q ij +1) Take the natural logarithm for normalization, is the square root of the sum of the squares of the elements of the priority matrix Q, which is used to comprehensively measure the priority of the request. ww are the diagonal elements of the permission matrix P, representing the specific permission values ​​of the role;

[0030] S24: Make a decision based on the calculated matching degree M and the preset access control rules. If M≥M threshold, then the user is allowed to access the requested data if M <M threshold , the user is denied access to the requested data, or is granted partial access rights.

[0031] Preferably, in S2, the basic preprocessing includes removing redundant spaces, special characters and punctuation marks, and the decomposition and analysis processing includes using a word segmentation tool to split the input keywords into smaller words or phrases, filtering out meaningless words through a predefined stop word list, and replacing synonyms using a synonym dictionary.

[0032] Preferably, in S2, the decomposed and parsed keywords are optimized, and the optimization includes using a rule-based keyword expansion method to expand the query keywords into more relevant words and applying Boolean logic to optimize keyword combinations.

[0033] Preferably, in S3, the weighted keyword matching algorithm includes:

[0034] Extract feature information from the test data in the test data management library and the optimized keywords, obtain feature vectors in the test data and query keywords, and obtain feature representation:

[0035] The feature vector of the detection data is expressed as: is the number of features of the detection data feature vector;

[0036] The feature vector of the query keyword is expressed as: is the number of keywords in the query keyword feature vector.

[0037] Preferably, the similarity between the keyword and the detection data is calculated based on a similarity calculation formula, and the formula is as follows:

[0038]

[0039] In the formula, Q is the query keyword feature vector, T is the detection data feature vector, and w i is the weight of the i-th query keyword feature vector, f(q i ,T) is the query keyword q i The number of occurrences in the test data feature vector T indicates the frequency of the query keyword in the test data, which is used to measure the relevance of the keyword to the test data;

[0040] q i is the i-th keyword in the query keyword feature vector, indicating the specific keyword queried by the user, t i is the tth feature vector in the detection data i feature values, representing the specific features in the low-pressure metering box detection data, is the adjustment factor of keyword matching, which performs nonlinear adjustment on the matching degree between query keywords and detection data features;

[0041] log(1+f(q i ,T)) is the logarithmic transformation of the number of occurrences, and the number of keyword occurrences is logarithmically transformed. is the normalization factor, which normalizes the query keywords and the detection data feature vectors;

[0042] According to the similarity sorting, the detection data results with the highest similarity are returned to the user.

[0043] Compared with the prior art, the present invention has the following advantages:

[0044] 1. This application uses an enhanced hash matrix encryption algorithm to store user passwords. By sharding the passwords and performing complex matrix hash operations, the security of password storage is enhanced, conventional brute force cracking and rainbow table attacks are prevented, and the security of user access is effectively improved.

[0045] 2. This application can achieve refined data access management by introducing matching degree calculation. The matching degree calculation not only takes into account the permissions of user roles, but also takes into account the priority of access requests, ensuring the security and rationality of data access and effectively improving the user experience.

[0046] 3. This application introduces a weighted keyword matching algorithm to calculate the similarity between keywords and test data, thereby ensuring the accuracy and stability of the matching results, making it have higher query efficiency and accuracy. By recording all access logs, it is convenient for subsequent auditing and tracking, which is conducive to subsequent optimization analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:

[0048] Figure 1 It is the overall flow chart of the present invention. DETAILED DESCRIPTION

[0049] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0050] like Figure 1 As shown, this embodiment proposes a management optimization method based on automatic detection of low-voltage metering boxes, including the following steps:

[0051] S1: Store user passwords based on the enhanced hash matrix encryption algorithm, control user access and operation permissions based on the permission management optimization algorithm, and establish a detection data management library;

[0052] S2: Perform basic preprocessing on the query keywords, decompose and analyze the keywords after the basic preprocessing, and optimize the keywords after the decomposition and analysis;

[0053] S3: Based on the weighted keyword matching algorithm, the similarity between the keyword and the test data is calculated, sorted according to the similarity, and the test data with the highest similarity is returned to the user;

[0054] S4: Record the keywords, matching process and results of each query in the system to build a data log.

[0055] In S1, in order to ensure the centralized storage, backup and recovery functions of the detection data, a detection data management library is established. The process of establishing the detection data management library includes professional technicians selecting a suitable database management system, such as MySQL, PostgreSQL or NoSQL database, according to the data type and query requirements.

[0056] Design the database architecture based on the regularization method, define data tables and relationships, ensure the data structure is reasonable, support efficient query and storage, implement partitioning and sharding strategies such as hash sharding and range sharding, distribute data on multiple storage nodes to improve the processing power and scalability of the database, use technologies such as composite indexes or full-text indexes to define and implement data indexes, optimize common queries, and improve query speed.

[0057] In S1, in order to improve data processing efficiency and scalability, a distributed data storage method based on a hash algorithm is selected to ensure balanced data distribution and availability. The Gossip protocol is used to synchronize the status between nodes, and a distributed consensus algorithm (such as Paxos or Raft algorithm) is used to achieve distributed consistency, ensuring balanced distribution and redundant storage of data among multiple nodes.

[0058] In S1, data replication and sharding strategies are implemented based on Quorum's read-write strategy to ensure high availability and fault tolerance of data. Monitoring tools (such as Prometheus and Grafana) are used to regularly monitor database performance. Automatic expansion methods are used to dynamically adjust resource allocation, adjust sharding and replication strategies, optimize performance, and use backup tools and cloud storage services to implement backup strategies and achieve data backup.

[0059] In S1, in order to ensure that only authorized users can access data, a user authentication mechanism is designed and an authentication mechanism based on username and password is implemented: ensure the uniqueness and reliability of user identity, use an enhanced hash matrix encryption algorithm to store user passwords to prevent password leakage, and use multi-factor authentication technology to add an additional security layer. Multi-factor authentication technology includes the use of SMS verification codes, email verification, and hardware tokens.

[0060] Configure the authentication server to centrally manage user identity information and authentication processes, and implement unified authentication and authorization management by using OAuth2.0 or OpenID Connect protocols. In order to control data access and operation permissions and ensure that users can only access and operate data within the authorized scope, assign access permissions of different roles to users.

[0061] Based on user responsibilities and data sensitivity, a role-based access control model is used to define the mapping relationship between roles and permissions. Access Control List (ACL) or Policy-based Access Control (PBAC) technology is used to manage roles and permissions and implement role-based access control in the database. In order to ensure that each data access is logged for auditing and tracking, data access control and logging are implemented to facilitate subsequent analysis and query.

[0062] In S1, the enhanced hash matrix encryption algorithm enhances the security of password storage by sharding and complex matrix hashing operations on passwords, preventing conventional brute force cracking and rainbow table attacks. The enhanced hash matrix encryption algorithm includes:

[0063] S11: Slice the user password and divide it into n substrings {p1, p2, ..., p n};

[0064] S12: Construct a randomly generated hash matrix H of size (m×m), the elements of the H matrix H ij Generated by the hash function hash(), where H ij = hash(s ij ), s ij is the random seed;

[0065] S13: For each substring p i Perform hash calculation to get the hash value h i ,h i = hash(p i );

[0066] S14: Each h i Convert to matrix M iThe size is (m×m), and the matrix M i Perform normalization:

[0067] S15: Calculate the final hash value H final , the formula is as follows:

[0068]

[0069] In the formula, H final is the final hash value, which is used to store the hash value of the user password. n is the number of password fragments, the number of substrings into which the user password is divided. m is the size of the hash matrix, that is, the dimension of the H matrix. M i,j,k is the i-th codon string p i The hash matrix M i The j-th row and k-th column element of H j,k is the jth row and kth column element of the randomly generated hash matrix H, log(i+1) is the natural logarithm of (i+1), which is used for normalization processing, M i,l,l is the i-th codon string p i The hash matrix M i diagonal elements.

[0070] In S1, in order to achieve refined management of data access control, the permission management optimization algorithm is introduced. The permission management optimization algorithm performs complex calculations on the matching degree between permissions and requests, and finely controls data access rights to ensure the security and rationality of each request. The permission management optimization algorithm includes:

[0071] S21: Assign a permission matrix P of size (k×k) to each role. The element P of the matrix P is ij It is the permission value of the role in a certain permission dimension;

[0072] S22: Classify each data access request R and set a priority matrix Q of size (k×k). The element Q of the Q matrix is ij The priority of data access requests on a certain dimension;

[0073] S23: When a user initiates a data access request, the user's role and specific request information are obtained, and the matching degree M between the role and the request is calculated. The formula is as follows:

[0074]

[0075] Where M is the matching degree between the role and the request, which is used to determine the user's access rights to the data, k is the size of the permission matrix P and the priority matrix Q, that is, the dimension of the matrix, P ij is the i-th row and j-th column element of the permission matrix P, representing the permission value of the role, Qij is the element in the \(i\)-th row and \(j\)-th column of the priority matrix \(Q\), representing the priority value of the data access request. \(\log(Q ij + 1)\) is to take the natural logarithm of \((Q ij + 1)\) for normalization processing. is the square root of the sum of squares of the elements of the priority matrix \(Q\), used to comprehensively measure the priority of the request. \(P ww is the diagonal element of the permission matrix \(P\), representing the specific permission value of the role.

[0076] S24: Make a decision based on the calculated matching degree \(M\) and the preset access control rules. If \(M\geq M threshold \), the user is allowed to access the requested data. If \(M < M threshold \), the user's access request for the data is rejected, or partial access permission is granted. Whether the access request is allowed or not, an access log needs to be recorded for subsequent auditing and tracking.

[0077] By introducing the calculation of the matching degree, refined data access management can be achieved. Calculating the matching degree not only considers the permissions of the user role but also the priority of the access request, ensuring the security and rationality of data access. Recording all access logs facilitates subsequent auditing and tracking, thus constructing a secure, efficient, and flexible permission management.

[0078] In S2, the basic preprocessing includes removing redundant spaces, special characters, and punctuation marks. The decomposition and parsing processing include using a word segmentation tool (such as Jieba for Chinese or NLTK for English) to split the input keywords into smaller words or phrases. For example, splitting "low-voltage metering box detection data" into "low voltage", "metering box", "detection", and "data". Filter out meaningless words (such as "of", "and", "the", "is") through a predefined stop word list, and use a synonym dictionary to replace synonyms to ensure that the query covers more relevant expressions.

[0079] In S2, the decomposed and parsed keywords are optimized. The optimization processing includes using a rule-based keyword expansion method, such as using a dictionary, a thesaurus, etc., to expand the query keywords into more relevant words, and applying Boolean logic (such as AND, OR, NOT) to optimize the keyword combination, thereby improving the accuracy of the query.

[0080] In S3, the weighted keyword matching algorithm includes:

[0081] Extract the feature information from the detection data in the detection data management library and the optimized keywords to obtain the feature vectors in the detection data and the query keywords, and obtain the feature representation form, such as the keyword list:

[0082] The feature vector of the detection data is expressed as: Such as [low pressure, metering box, detection], is the number of features of the detection data feature vector;

[0083] The feature vector of the query keyword is expressed as: The number of keywords in the query keyword feature vector, where the query keyword is included in the features of the detection data in the monitoring data management library.

[0084] Based on the similarity calculation formula, the similarity between the keyword and the detection data is calculated. The formula is as follows:

[0085]

[0086] In the formula, Q is the query keyword feature vector, T is the detection data feature vector, and w i is the weight of the i-th query keyword feature vector, f(q i ,T) is the query keyword q i The number of occurrences in the test data feature vector T indicates the frequency of the query keyword in the test data, which is used to measure the relevance of the keyword to the test data;

[0087] q i is the i-th keyword in the query keyword feature vector, indicating the specific keyword queried by the user, t i is the tth feature vector in the detection data i feature values, representing specific features in the low-voltage meter box detection data, such as voltage, current, or words in the text description. The adjustment factor of the keyword matching degree is used to make nonlinear adjustments to the matching degree of the query keyword and the detection data features to increase the complexity and accuracy of the matching calculation;

[0088] log(1+f(q i ,T)) is the logarithmic transformation of the number of occurrences. The number of keyword occurrences is logarithmically transformed to reduce the impact of high-frequency words and increase the stability of the calculation. and is the normalization factor, which normalizes the query keywords and the detection data feature vectors to eliminate the influence of different number of features on the matching degree calculation;

[0089] According to the similarity sorting, the most similar test data results are returned to the user. The keywords, matching process and results of each query are recorded in the system, and a data log is constructed for future optimization and analysis, thereby effectively realizing the management optimization of the automatic detection data of the low-voltage metering box, ensuring accurate query and matching of data, and improving query efficiency and user satisfaction.

[0090] Although the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.

Claims

1. A management optimization method based on automatic detection of low-voltage metering boxes, characterized in that: The following steps are involved: S1: Store user passwords based on the enhanced hash matrix encryption algorithm, control user access and operation permissions based on the permission management optimization algorithm, and establish a detection data management library; S2: Perform basic preprocessing on the query keywords, decompose and analyze the keywords after the basic preprocessing, and optimize the keywords after the decomposition and analysis; S3: Based on the weighted keyword matching algorithm, the similarity between the keyword and the test data is calculated, sorted according to the similarity, and the test data with the highest similarity is returned to the user; S4: Record the keywords, matching process and results of each query in the system to build a data log.

2. A management optimization method based on automatic detection of low-voltage metering boxes according to claim 1, characterized in that: In S1, the database architecture is designed based on the regularization method, data tables and relationships are defined, partitioning and sharding strategies are implemented, and data is distributed on multiple storage nodes.

3. A management optimization method based on automatic detection of low-voltage metering boxes according to claim 1, characterized in that: In S1, a distributed data storage method based on a hash algorithm is selected, the Gossip protocol is used to synchronize the states between nodes, and a distributed consensus algorithm is used to achieve distributed consistency.

4. A management optimization method based on automatic detection of low-voltage metering boxes according to claim 1, characterized in that: In S1, a monitoring tool is used to regularly monitor database performance, an automatic expansion method is used to dynamically adjust resource allocation, sharding and replication strategies are adjusted, and a backup tool and a cloud storage service are used to implement a backup strategy to achieve data backup.

5. A management optimization method based on automatic detection of low-voltage metering boxes according to claim 1, characterized in that: In S1, a user identity authentication mechanism is designed using multi-factor identity authentication technology, and the multi-factor identity authentication technology includes using SMS verification codes, email verification, and hardware tokens.

6. A management optimization method based on automatic detection of low-voltage metering boxes according to claim 1, characterized in that: In S1, the enhanced hash matrix encryption algorithm includes: S11: Slice the user password and divide it into n substrings {p1, p2, ..., p n }; S12: Construct a randomly generated hash matrix H of size (m×m), the elements of the H matrix H ij Generated by the hash function hash(), where H ij = hash(s ij ), s ij is the random seed; S13: For each substring p i Perform hash calculation to get the hash value h i ,h i = hash(p i ); S14: Each h i Convert to matrix M i The size is (m×m), and the matrix M i Perform normalization: S15: Calculate the final hash value H final , the formula is as follows: In the formula, H final is the final hash value, n is the number of password fragments, m is the size of the hash matrix, M i，j，k is the i-th codon string p i The hash matrix M i The j-th row and k-th column element of H j，k is the jth row and kth column element of the randomly generated hash matrix H, log(i+1) is the natural logarithm of (i+1), which is used for normalization processing, M i，l，l is the i-th codon string p i The hash matrix M i diagonal elements.

7. A management optimization method based on automatic detection of low-voltage metering boxes according to claim 1, characterized in that: In S1, the rights management optimization algorithm includes: S21: Assign a permission matrix P of size (k×k) to each role. The element P of the matrix P is ij It is the permission value of the role in a certain permission dimension; S22: Classify each data access request R and set a priority matrix Q of size (k×k). The element Q of the Q matrix is i j is the priority of the data access request in a certain dimension; S23: When a user initiates a data access request, the user's role and specific request information are obtained, and the matching degree M between the role and the request is calculated. The formula is as follows: Where M is the matching degree between the role and the request, which is used to determine the user's access rights to the data, k is the size of the permission matrix P and the priority matrix Q, and P ij is the i-th row and j-th column element of the permission matrix P, representing the permission value of the role, Q ij is the i-th row and j-th column element of the priority matrix Q, which represents the priority value of the data access request. Log(Q ij +1) for (Q ij +1) Take the natural logarithm for normalization, is the square root of the sum of the squares of the elements of the priority matrix Q, which is used to comprehensively measure the priority of the request. ww are the diagonal elements of the permission matrix P, representing the specific permission values ​​of the role; S24: Make a decision based on the calculated matching degree M and the preset access control rules. If M≥M threshold , then the user is allowed to access the requested data if M <M threshold , the user is denied access to the requested data, or is granted partial access rights.

8. A management optimization method based on automatic detection of low-voltage metering boxes according to claim 1, characterized in that: In S2, the basic preprocessing includes removing redundant spaces, special characters and punctuation marks, and the decomposition and analysis processing includes using a word segmentation tool to split the input keywords into smaller words or phrases, filtering out meaningless words through a predefined stop word list, and replacing synonyms using a synonym dictionary.

9. A management optimization method based on automatic detection of low-voltage metering boxes according to claim 1, characterized in that: In S2, the decomposed and parsed keywords are optimized, and the optimization includes using a rule-based keyword expansion method to expand the query keywords into more relevant words and applying Boolean logic to optimize keyword combinations.

10. A management optimization method based on automatic detection of low-voltage metering boxes according to claim 1, characterized in that: In S3, the weighted keyword matching algorithm includes: Extract feature information from the test data in the test data management library and the optimized keywords, obtain feature vectors in the test data and query keywords, and obtain feature representation: The feature vector of the detection data is expressed as: is the number of features of the detection data feature vector; The feature vector of the query keyword is expressed as: is the number of keywords in the query keyword feature vector.

11. A management optimization method based on automatic detection of low-voltage metering boxes according to claim 10, characterized in that: Based on the similarity calculation formula, the similarity between the keyword and the detection data is calculated. The formula is as follows: In the formula, Q is the query keyword feature vector, T is the detection data feature vector, and w i is the weight of the i-th query keyword feature vector, f(q i ,T) is the query keyword q i The number of occurrences in the test data feature vector T indicates the frequency of the query keyword in the test data, which is used to measure the relevance of the keyword to the test data; q i is the i-th keyword in the query keyword feature vector, indicating the specific keyword queried by the user, t i is the tth feature vector in the detection data i feature values, representing the specific features in the low-pressure metering box detection data, is the adjustment factor of keyword matching, which performs nonlinear adjustment on the matching degree between query keywords and detection data features; log(1+f(q i ,T)) is the logarithmic transformation of the number of occurrences, and the number of keyword occurrences is logarithmically transformed. and is the normalization factor, which normalizes the query keywords and the detection data feature vectors; According to the similarity sorting, the detection data results with the highest similarity are returned to the user.

Images