A verification method, device and equipment of a power service system and a medium

Abstract

The application provides a verification method, device and equipment of a power service system and a medium, relates to the field of computers, and is used for solving the problem of low verification efficiency of the power service system. The method comprises the following steps: constructing an index tree according to the latest policy text of the power industry; determining a target keyword set according to a function module of the power service system; searching in the index tree according to the target keyword set to obtain key sentences related to the power service system; converting the key sentences into SQL statements; executing the SQL statements in a database of the power service system to obtain a verification result; and using the verification result to indicate whether the power service system is compliant. The method realizes end-to-end conversion from a policy text to an SQL statement, automatically generates executable compliance verification logic, and thus improves the verification efficiency of the power service system.

Description

TECHNICAL FIELD

[0001] The application relates to the computer technical field and provides a verification method, device and equipment of a power service system and a medium. BACKGROUND

[0002] The power service system refers to a specialized information system supporting production, operation, management and service of the power industry. With the increasing requirements of the power grid industry on safe, stable and economic operation, new policies and new standards are constantly introduced, which puts forward higher requirements on the operation of the power service system. Compliance verification is not only a key step to ensure that the system meets new policies and standards, but also an important means to improve the safety, efficiency of the power grid and cope with future challenges.

[0003] At present, the artificial verification method is mainly used to verify the power service system according to the latest policy text, but the efficiency is low. SUMMARY

[0004] The application provides a verification method, device and equipment of a power service system and a medium to solve the problem of low verification efficiency of the power service system.

[0005] In a first aspect, a verification method of a power service system is provided, comprising:

[0006] According to the latest policy text of the power industry, an index tree is constructed;

[0007] According to the function modules of the power service system, a target keyword set is determined;

[0008] According to the target keyword set, retrieval is performed in the index tree to obtain key sentences related to the power service system;

[0009] The key sentences are converted into SQL statements;

[0010] The SQL statements are executed in the database of the power service system to obtain a verification result; the verification result is used to indicate whether the power service system is compliant.

[0011] Optionally, the target keyword set is determined according to the function modules of the power service system, comprising:

[0012] The basic keywords are extracted from the function modules of the power service system;

[0013] The basic keywords are expanded according to the power industry standard synonym library to obtain extended keywords;

[0014] The target keyword set is obtained according to the basic keywords and the extended keywords.

[0015] Optionally, obtaining the target keyword set based on the basic keywords and the extended keywords includes:

[0016] Determine the similarity threshold based on the type of the policy text;

[0017] The basic keywords and the extended keywords are combined to obtain an initial keyword set;

[0018] Calculate the semantic similarity between each keyword in the initial keyword set and the policy text, and retain keywords with semantic similarity higher than the similarity threshold to obtain the target keyword set.

[0019] Optionally, determining the similarity threshold based on the type of the policy text includes:

[0020] If the policy text is of the type of legal provisions, then the first threshold is determined as the similarity threshold;

[0021] If the policy text is of the technical specification type, then the second threshold is determined as the similarity threshold;

[0022] If the policy text is of the notification or announcement type, then the third threshold is determined as the similarity threshold; wherein the first threshold is greater than the second threshold, and the second threshold is greater than the third threshold.

[0023] Optionally, the step of constructing an index tree based on the latest policy texts in the power industry includes:

[0024] The latest policy text for the power industry is divided into multiple paragraphs, and vector representations of these paragraphs are generated using text embedding technology. These vector representations are then used as leaf nodes.

[0025] Cluster the vector representations of the multiple paragraphs to obtain multiple clusters;

[0026] A summary for each cluster is generated using a large language model;

[0027] The vector representation of each cluster is obtained by weighting the vector representations of the segments in each cluster;

[0028] The summaries and vector representations of the multiple clusters are embedded into the parent node of the leaf node through recursive embedding until the root node is constructed to obtain the index tree; wherein, the index tree contains multiple nodes, and each node contains a node index, child nodes, a summary, and a vector representation.

[0029] Optionally, the clustering of the vector representations of the multiple paragraphs to obtain multiple clusters includes:

[0030] The vector representations of the multiple paragraphs are treated as independent clusters;

[0031] Use the exchange test to calculate the fidelity between clusters;

[0032] The two clusters with the highest fidelity are merged until the number of clusters reaches the preset number, resulting in multiple clusters after clustering.

[0033] Optionally, converting the key statement into an SQL statement includes:

[0034] Identify the fields and constraint types in the key statements;

[0035] Based on a predefined field mapping table, the fields in the key statements are mapped to fields in the database table;

[0036] If the constraint type is a numeric range constraint, then an SQL statement with a BETWEEN condition or comparison operator is generated based on the fields of the database table; if the constraint type is an enumeration value constraint, then an SQL statement with an IN condition is generated based on the fields of the database table; if the constraint type is a regular expression constraint, then an SQL statement matching REGEXP is generated based on the fields of the database table.

[0037] Secondly, a verification device for a power business system is provided, comprising:

[0038] The building module is used to construct an index tree based on the latest policy texts in the power industry;

[0039] The determination module is used to determine the target keyword set based on the functional modules of the power business system;

[0040] The retrieval module is used to perform a retrieval in the index tree based on the target keyword set to obtain key statements related to the power business system.

[0041] The conversion module is used to convert the key statements into SQL statements;

[0042] The verification module is used to execute the SQL statement in the database of the power business system and obtain the verification result; the verification result is used to indicate whether the power business system is compliant.

[0043] Thirdly, this application provides a computer device including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the verification method for the power business system described in the first aspect.

[0044] Fourthly, this application provides a computer-readable storage medium storing a computer program, on which a processor executes the computer program to implement the verification method for the power business system described in the first aspect.

[0045] Compared with the prior art, the beneficial effects of the embodiments of this application are as follows:

[0046] This application provides a verification method for a power business system. The method includes: constructing an index tree based on the latest policy texts of the power industry; determining a target keyword set based on the functional modules of the power business system; searching the index tree based on the target keyword set to obtain key statements related to the power business system; converting the key statements into SQL statements; executing the SQL statements in the database of the power business system to obtain verification results; and using the verification results to indicate whether the power business system is compliant.

[0047] As can be seen, this application can automatically match the relevant content of the latest policy texts based on the target keyword set of the power business system, and retrieve relevant data from the database through SQL statements to automatically verify whether the power business system complies with the latest policy requirements. This reduces the tedious process of manual inspection, improves verification efficiency, and enables decision-makers to quickly obtain feedback on compliance, thereby accelerating the decision-making process and avoiding risks and delays caused by compliance issues. Attached Figure Description

[0048] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0049] Figure 1 A schematic diagram of the computing device structure of the hardware operating environment involved in the embodiments of this application;

[0050] Figure 2 A flowchart illustrating the verification method for a power business system provided in this application embodiment;

[0051] Figure 3 This is a schematic diagram of the index tree structure provided in an embodiment of this application;

[0052] Figure 4 A schematic diagram illustrating the formation of intermediate layer nodes provided in an embodiment of this application;

[0053] Figure 5 This is a schematic diagram of the structure of the verification device for the power business system provided in the embodiments of this application.

[0054] The diagram is labeled as follows: 101-Processor, 102-Communication bus, 103-Network interface, 104-User interface, 105-Memory. Detailed Implementation

[0055] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application. Unless otherwise specified, the embodiments and features in the embodiments of this application can be arbitrarily combined with each other. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than that shown here.

[0056] To address the low verification efficiency of power business systems, this application provides a verification method for power business systems, which can be executed by computer equipment. Please refer to... Figure 1 This is a schematic diagram of the computer device structure of the hardware operating environment involved in the embodiments of this application.

[0057] like Figure 1 As shown, the computer device may include: a processor 101, such as a central processing unit (CPU), a communication bus 102, a user interface 104, a network interface 103, and a memory 105. The communication bus 102 is used to enable communication between these components. The user interface 104 may include a display screen and an input unit such as a keyboard; the user interface 104 may include standard wired and wireless interfaces. The network interface 103 may include standard wired and wireless interfaces (such as a Wi-Fi interface). The memory 105 may be high-speed random access memory (RAM) or stable non-volatile memory (NVM), such as a disk drive. The memory 105 may also be a storage device independent of the aforementioned processor 101.

[0058] Those skilled in the art will understand that Figure 1 The structure shown does not constitute a limitation on the computer device and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0059] like Figure 1 As shown, the memory 105, which serves as a storage medium, may include an operating system, a network communication module, a user interface module, and a verification device for the power business system.

[0060] exist Figure 1 In the computer device shown, the network interface 103 is mainly used for data communication with the network server; the user interface 104 is mainly used for data interaction with the user; the processor 101 and the memory 105 in the computer device of the present invention can be set in the computer device, and the computer device calls the power business system verification device stored in the memory 105 through the processor 101 and executes the power business system verification method provided in the embodiment of this application.

[0061] based on Figure 1 The computer equipment shown below, in conjunction with Figure 2 The present application provides a verification method for a power business system, the steps of which are as follows:

[0062] S201. Construct an index tree based on the latest policy texts in the power industry.

[0063] In practice, policy texts are official documents issued by governments, organizations, or other management agencies to convey policy directions, regulations, decisions, or action plans, and to guide or regulate behavior and operations in related fields, such as the "National Unified Electricity Market Development Plan Blue Book." The latest policy texts for the power industry can be obtained through official website downloads, issuance by higher-level or relevant departments, etc. Policy texts can be in formats such as PDF, TXT, Word, and web pages.

[0064] Policy texts can be classified as legal provisions, technical specifications, or notices and announcements. These will be described in detail below:

[0065] 1. Policy texts containing legal provisions

[0066] These types of documents have legal force and are mainly drafted by governments or legislative bodies. Their purpose is to regulate the basic operations of the power industry and ensure the safety and stability of power supply. Examples include legal documents concerning the power market, electricity management, and environmental regulations.

[0067] 2. Policy texts related to technical specifications

[0068] These types of texts mainly concern the technical requirements and standards of the power industry, and are usually issued by the state or relevant industry authorities to ensure the safety, stability and sustainability of the power system.

[0069] 3. Policy texts such as notices and announcements

[0070] These types of documents are typically used to release information such as administrative measures, policy changes, and management regulations, and are highly time-sensitive. They are usually issued by government departments or industry regulatory bodies to inform power companies or the public about the latest developments or policy changes in the power industry. Examples include annual plans for the power industry, price adjustment announcements, and policy implementation notices.

[0071] In one possible embodiment, the specific steps of S201 include S1.1-S1.6.

[0072] S1.1 Divide the latest policy texts for the power industry into multiple paragraphs.

[0073] Specifically, after obtaining the latest policy texts for the power industry, corresponding preset rules can be determined based on the type of policy text. These preset rules identify delimiters within the policy text, which are then used to divide the text into multiple paragraphs. If the policy text is a legal provision, "Article X", "Section X", and document number identifiers are used as delimiters. If the policy text is a technical specification, tables containing power standard numbers (GB / T / DL, etc.) and technical parameters are used as delimiters, and splitting across tables is prohibited. If the policy text is a notice or announcement, hierarchical numbers such as "(I)" and "1." are used as delimiters.

[0074] Alternatively, a power policy corpus can be constructed, which contains multiple policy texts with manually annotated paragraph boundaries. A semantic segmentation model can be trained based on the power policy corpus, and then the trained semantic segmentation model can be used to calculate the boundary probabilities between adjacent sentences in the policy text to generate multiple paragraphs.

[0075] S1.2 generates vector representations of multiple paragraphs using text embedding technology, and uses these vector representations as leaf nodes.

[0076] Specifically, text embedding techniques (such as Word2Vec, BERT, RoBERTa, etc.) can be used to convert each paragraph into a fixed-length vector representation, and each paragraph's vector representation can be used as a leaf node in the index tree. This approach is not only based on text content but can also identify semantically similar paragraphs, avoiding the limitations of traditional keyword-based retrieval.

[0077] S1.3 Cluster the vector representations of multiple paragraphs to obtain multiple clusters.

[0078] Specifically, clustering algorithms can be used to cluster the vector representations of multiple paragraphs, grouping similar paragraphs into the same cluster. This helps the system better identify and organize semantically similar content, thereby supporting efficient multi-dimensional search. For example, when a user queries a policy-related topic, the system can find the most relevant parts by calculating the distance to the cluster center.

[0079] S1.4. Generate a summary for each cluster using a large language model.

[0080] Specifically, a large language model (such as GPT or T5) can be used to generate a summary for each cluster. The summary for each cluster summarizes the core content of all paragraphs in the cluster, thus reducing redundancy and extracting the key points of the cluster.

[0081] S1.5. Perform a weighted average of the vector representations of the segments in each cluster to obtain the vector representation of each cluster.

[0082] Specifically, a weighted average can be taken from the vector representations of all paragraphs in each cluster to generate the vector representation for each cluster. The weighted average can be assigned weights based on paragraph length, importance, or other metrics to ensure that the most representative paragraphs within a cluster dominate the vector representation of that cluster.

[0083] S1.6. Embed the summaries and vector representations of multiple clusters into the parent node of the leaf node, and repeat the embedding process until the root node is constructed to obtain the index tree.

[0084] Specifically, the summary and vector representation of each cluster are embedded into the parent node of the leaf node. Through recursive embedding, an index tree is finally constructed. The root node of the tree represents the overall content summary and vector representation of the policy text.

[0085] The index tree contains multiple nodes, each node containing a node index, child nodes, a summary, and a vector representation. Please refer to [link / reference]. Figure 3 This is a schematic diagram of the index tree structure provided in an embodiment of this application. A, B, C, D, and E are leaf nodes. Clustering the leaf nodes forms intermediate layer nodes F, G, and H, and further clustering the intermediate layer nodes forms root nodes I and J.

[0086] Please refer to Figure 4 This is a schematic diagram illustrating the formation of intermediate layer nodes provided in an embodiment of this application. Clustering is performed on A, B, C, D, and E, resulting in: C and E forming one cluster, A and D forming one cluster, and B forming another cluster. C and E are merged into F, A and D into G, and B is grouped into H.

[0087] In this embodiment, by embedding summaries and vector representations of multiple text paragraphs into a tree structure, a large amount of policy text can be effectively organized and indexed. When specific information needs to be found, the system can more efficiently locate relevant content based on the node's summary, reducing unnecessary full scans. By constructing the index tree level by level through recursive embedding, it can be ensured that high-level nodes represent the essence of the entire document, thus improving the retrieval efficiency of policies related to a specific topic or field.

[0088] In one possible embodiment, the vector representations of multiple paragraphs are clustered to obtain multiple clusters, including: treating the vector representation of each paragraph as an independent cluster; using a cross-validation test to calculate the fidelity between clusters; merging the two clusters with the highest fidelity until the number of clusters is a preset number, thus obtaining multiple clusters after clustering.

[0089] In the specific implementation process, a quantum agglomerative hierarchical clustering algorithm is selected. Each sample (i.e., the vector representation of each segment) is initially considered an independent cluster. The vector representation of each cluster is converted into a quantum state, and the fidelity between every two clusters is calculated using a swap test. The two clusters with the highest fidelity are merged. Each merge updates the cluster center and recalculates the fidelity between every two clusters. The process continues to select cluster pairs with the highest fidelity for merging. Merging stops when the number of clusters reaches a preset number.

[0090] For example, clusters C i and cluster C j The fidelity between clusters is defined as the cluster fidelity. C i All samples and clusters C j The maximum fidelity between the means of all samples is calculated using the following formula:

[0091]

[0092] in, L ij Cluster C i and cluster C j The fidelity between them, max represents the maximum value function, w i_t Cluster C i All samples, where L is the cluster C j The number of samples in w j_k_t Cluster C j The k-th sample in the dataset, where k is a positive integer from 0 to L-1. Cluster Ci All samples and clusters C j The fidelity between the means of all samples.

[0093] In this embodiment of the application, considering that language words may have multiple meanings that need to be determined by context, and that quantum mechanics has multiple states that need to be determined by observation, the similarity between natural language and quantum mechanics is utilized to construct index trees at different summary levels through recursive embedding, quantum clustering, and text fragment summarization. This not only captures high-level and low-level details of the text, but also improves the generation speed and aggregation accuracy of the index tree.

[0094] S201. Determine the target keyword set based on the functional modules of the power business system.

[0095] In practical implementation, the functional modules of the power business system can include power production, power transmission, power distribution, power marketing and billing, power dispatching and monitoring, and power safety and protection modules. Furthermore, core terms can be extracted from these functional modules to obtain basic keywords, which can then be merged into a target keyword set. These basic keywords can be artificially defined based on the key business functions of each module. For example, the basic keywords for the power production module could be power plant, power generation dispatching, power generation, and power plant equipment; the same applies to the basic keywords for other functional modules, which will not be elaborated further here.

[0096] While basic keywords can provide keywords for specific functional modules, they may not take into account the differences and diversity in other professional fields or regions within the industry. By expanding basic keywords, more synonyms, variants, and related terms can be generated, improving the diversity and accuracy of keywords. This way, during retrieval, not only can precise terms be matched, but also related terms can be found, avoiding information omissions. Therefore, in one possible embodiment, the specific steps of S201 include: extracting basic keywords from the functional modules of the power business system; expanding the basic keywords according to the power industry standard thesaurus to obtain expanded keywords; and obtaining a target keyword set based on the basic keywords and expanded keywords.

[0097] The aforementioned power industry standard thesaurus refers to a collection of terms and their synonyms used within the power industry. For example, synonyms for power plant include power plant and electric power plant. By merging the basic keyword (power plant) and the extended keyword (electric power plant), a target keyword set is obtained to improve subsequent search efficiency.

[0098] In one possible embodiment, the step of obtaining a target keyword set based on basic keywords and extended keywords includes: determining a similarity threshold based on the type of policy text; merging the basic keywords and extended keywords to obtain an initial keyword set; calculating the semantic similarity between each keyword in the initial keyword set and the policy text, retaining keywords with semantic similarity higher than the similarity threshold, and obtaining the target keyword set.

[0099] In practice, the BERT model can be used to convert each keyword in the initial keyword set into a context-dependent vector representation of the policy text, and then the cosine similarity of these vectors can be calculated. By comparing this vector with a similarity threshold, keywords with a cosine similarity lower than the threshold are removed, while those with a cosine similarity higher than the threshold are retained, thus ultimately selecting the target keyword set.

[0100] In this embodiment of the application, by setting a similarity threshold, only keywords that are highly related to the policy text will be retained. This can effectively eliminate keywords that do not conform to the semantics of the policy text, reduce the interference of irrelevant information, and improve the accuracy of subsequent analysis or processing.

[0101] The similarity threshold is determined based on the type of policy text, such as:

[0102] If the policy text is of the legal provisions type, then the first threshold will be determined as the similarity threshold;

[0103] If the policy text is of the technical specification type, then the second threshold will be determined as the similarity threshold;

[0104] If the policy text is of the notification or announcement type, then the third threshold is determined as the similarity threshold; wherein, the first threshold is greater than the second threshold, and the second threshold is greater than the third threshold.

[0105] In practice, legal texts are typically written in rigorous and formal language, with standardized expressions. To ensure the accuracy of keyword extraction, a high first threshold (e.g., 0.9) is set to ensure that only keywords highly relevant to the text content are selected, avoiding the inclusion of low-relevance or misleading keywords. Technical specifications usually contain a large amount of industry terminology and operational procedures, exhibiting a certain degree of complexity and professionalism. Setting a moderate second threshold (e.g., 0.8) helps capture keywords related to technical details while avoiding over-screening for excessively irrelevant keywords. Notices and announcements generally use concise and clear language, with more general and announcement-oriented content. Keyword selection can be relatively broader, and a lower third threshold (e.g., 0.7) helps capture more potential and relevant keywords, preventing the omission of potentially crucial information.

[0106] In this embodiment, by setting different similarity thresholds for different types of policy texts, more flexible and accurate keyword filtering can be achieved. The thresholds are stricter for legal texts, while lower thresholds are used for technical specifications and notices to broaden the scope of keyword extraction. This method effectively improves the accuracy and relevance of keyword filtering.

[0107] S203. Search the index tree based on the target keyword set to obtain key statements related to the power business system.

[0108] In the specific implementation process, all keywords in the target keyword set are iteratively searched in the index tree using a depth-first search strategy, with the number of iterations equal to the number of keywords in the target keyword set. For example, if the target keyword set is {ancillary services, price}, the policy document can be found to contain two key statements regarding ancillary services and prices: "In principle, the upper limit of peak-shaving service prices shall not exceed the on-grid electricity price of local grid parity new energy projects" and "The upper limit of standby service prices shall not exceed the upper limit of local electricity market prices."

[0109] S204. Convert key statements into SQL statements.

[0110] In one possible implementation, the fields and constraint types in the key statements are identified; based on a predefined field mapping table, the fields in the key statements are mapped to fields in the database table; if the constraint type is a numeric range constraint, an SQL statement with a BETWEEN condition or a comparison operator (<=) is generated based on the fields in the database table; if the constraint type is an enumeration value constraint, an SQL statement with an IN condition is generated based on the fields in the database table; if the constraint type is a regular expression constraint, an SQL statement matching REGEXP is generated based on the fields in the database table.

[0111] For example, the key statement "In principle, the upper limit of the peak-shaving service price shall not be higher than the on-grid electricity price of local grid-parity new energy projects" has a numerical range constraint type. The SQL statement that generates the comparison operator (<=) is as follows:

[0112] SELECT * FROM energy_prices WHERE peak_price<= base_price

[0113] Here, SELECT * means selecting data from all columns, FROM energy_prices means the data is stored in the energy_prices table, and WHERE peak_price<= base_price is a constraint that ensures the peak-shaving service price ceiling (peak_price) does not exceed the local grid-connected electricity price (base_price) of grid-parity renewable energy projects. Only records that meet this condition will be returned.

[0114] For example, the key statement: "The status of power equipment can only be normal, faulty, or under maintenance," has an enumeration value constraint type. The SQL statement that generates the IN condition is as follows:

[0115] SELECT * FROM equipment WHERE status IN (“Normal”, “Fault”, “Under Maintenance”)

[0116] The SQL statement means: Retrieve all devices from the equipment table whose device numbers are correctly formatted and whose status is either normal, faulty, or under maintenance.

[0117] For example, the key statement: "Electrical equipment number must conform to the format 'ED-' followed by 5 digits", its constraint type is regular expression constraint, and the generated SQL statement for REGEXP matching is as follows:

[0118] ELECT * FROM equipment WHERE equipment_id REGEXP '^ED-\d{5}$'

[0119] This SQL statement means: Retrieve all devices from the `equipment` table if the device ID (`equipment_id`) conforms to the format 'ED-' followed by 5 digits. The regular expression `^ED-\d{5}$` ensures that the device IDs follow this pattern.

[0120] In this embodiment, by identifying fields and constraint types in natural language and combining them with a predefined field mapping table, the natural language is converted into corresponding SQL statements. For different constraint types, SQL statements with different conditions (BETWEEN, <=, IN, REGEXP, etc.) can be generated, thereby ensuring that the generated SQL statements can correctly query data in the database.

[0121] S205. Execute SQL statements in the database of the power business system to obtain the verification results.

[0122] In practice, the verification results are used to indicate whether the power business system is compliant. If the verification results indicate compliance, it means that the power business system meets the latest policy requirements. If the verification results indicate non-compliance, decision-makers can formulate corresponding decisions based on the verification results, thereby optimizing the power business system to meet the latest policy requirements.

[0123] For example, you can select a data table reflecting ancillary services in the Electricity Marketing 2.0 system database, and then select the peak-shaving service price and the renewable energy feed-in tariff columns from this table. Using the generated SQL statement, you can compare the values ​​of the same row in these two columns to obtain the verification result for the peak-shaving service price. If the peak-shaving service price is lower than the renewable energy feed-in tariff, the verification result is compliant; if the peak-shaving service price is higher than the renewable energy feed-in tariff, the verification result is non-compliant.

[0124] In summary, this application provides a verification method for power business systems. By constructing an extensible index tree structure, it achieves multi-level semantic representation of power industry policy texts. Employing a recursive embedding clustering method, it abstracts the semantic information of policy texts hierarchically, improving the retrieval efficiency of large-scale policy texts. A three-level keyword processing flow is established, from basic keyword extraction to synonym expansion and semantic filtering, ensuring that the target keyword set is both comprehensive and accurate. An adaptive similarity threshold mechanism based on policy type is proposed, setting differentiated matching standards for different policy types such as legal provisions and technical specifications. End-to-end conversion from policy text to SQL queries is achieved. Through field mapping and constraint type recognition, executable compliance verification logic is automatically generated, significantly improving verification efficiency. This application, through the deep integration of natural language processing and database technology, constructs a policy-business mapping system unique to the power industry, providing an effective compliance assurance tool for the construction of new power systems. It can be used for rapid system compliance assessment after the introduction of new policies and verification of settlement rules in power trading systems.

[0125] Based on the same inventive concept, such as Figure 5 As shown in the illustration, this application also provides a verification device for a power business system, comprising:

[0126] The building module is used to construct an index tree based on the latest policy texts in the power industry;

[0127] The determination module is used to determine the target keyword set based on the functional modules of the power business system;

[0128] The retrieval module is used to search the index tree based on the target keyword set to obtain key statements related to the power business system.

[0129] The conversion module is used to convert key statements into SQL statements;

[0130] The verification module is used to execute SQL statements in the database of the power business system and obtain verification results; the verification results are used to indicate whether the power business system is compliant.

[0131] It should be noted that each module in the verification device of the power business system in this embodiment corresponds one-to-one with each step in the verification method of the power business system in the aforementioned embodiment. Therefore, the specific implementation of this embodiment can refer to the implementation of the verification method of the aforementioned power business system, and will not be repeated here.

[0132] Furthermore, in one embodiment, this application also provides a computer device, the computer device including a processor, a memory and a computer program stored in the memory, the computer program being executed by the processor to implement the aforementioned verification method for the power business system.

[0133] In addition, in one embodiment, this application also provides a computer storage medium storing a computer program, which is executed by a processor to implement the aforementioned verification method for a power business system.

[0134] In some embodiments, the computer-readable storage medium may be a memory such as FRAM, ROM, PROM, EPROM, EEPROM, flash memory, magnetic surface memory, optical disk, or CD-ROM; or it may be a device including one or any combination of the above-mentioned memories. The computer may be a variety of computing devices, including smart terminals and servers.

[0135] In some embodiments, executable instructions may take the form of a program, software, software module, script, or code, written in any form of programming language (including compiled or interpreted languages, or declarative or procedural languages), and may be deployed in any form, including as a standalone program or as a module, component, subroutine, or other unit suitable for use in a computing environment.

[0136] As an example, executable instructions may, but do not necessarily, correspond to files in the file system. They may be stored as part of a file that holds other programs or data, for example, in one or more scripts in a Hyper Text Markup Language (HTML) document, in a single file dedicated to the program in question, or in multiple collaborative files (e.g., a file that stores one or more modules, subroutines, or code sections).

[0137] As an example, executable instructions can be deployed to execute on a single computing device, or on multiple computing devices located in one location, or on multiple computing devices distributed across multiple locations and interconnected via a communication network.

[0138] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0139] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0140] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as read-only memory / random access memory, magnetic disk, optical disk) and includes several instructions to cause a multimedia terminal device (which may be a mobile phone, computer, television receiver, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0141] The above are merely preferred embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. A verification method for a power business system, characterized in that, include: Based on the trained semantic segmentation model, the latest policy texts of the power industry are divided into multiple paragraphs. Vector representations of these paragraphs are generated using text embedding technology, and these vector representations are used as leaf nodes. The trained semantic segmentation model is obtained by training a power policy corpus containing multiple policy texts with manually annotated paragraph boundaries. Treat the vector representation of each paragraph as an independent cluster; Use the exchange test to calculate the fidelity between clusters; Merge the two clusters with the highest fidelity until the number of clusters reaches the preset number, resulting in multiple clusters after clustering. A summary for each cluster is generated using a large language model; The vector representation of each cluster is obtained by weighted averaging of the vector representations of the segments in each cluster; The summaries and vector representations of the multiple clusters are embedded into the parent node of the leaf node. This process is repeated recursively until the root node is constructed, thus obtaining an index tree. The index tree contains multiple nodes, each of which contains a node index, child nodes, a summary, and a vector representation. Based on the functional modules of the power business system, determine the target keyword set; Based on the target keyword set, a search is performed in the index tree to obtain key statements related to the power business system; Identify the fields and constraint types in the key statements; Based on a predefined field mapping table, the fields in the key statements are mapped to fields in the database table; If the constraint type is a numeric range constraint, then an SQL statement with a BETWEEN condition or comparison operator is generated based on the fields of the database table; if the constraint type is an enumeration value constraint, then an SQL statement with an IN condition is generated based on the fields of the database table; if the constraint type is a regular expression constraint, then an SQL statement with a REGEXP match is generated based on the fields of the database table. The SQL statement is executed in the database of the power business system to obtain the verification result; the verification result is used to indicate whether the power business system is compliant.

2. The verification method for a power business system as described in claim 1, characterized in that, The determination of the target keyword set based on the functional modules of the power business system includes: Extract basic keywords from the functional modules of the power business system; The basic keywords are expanded based on the power industry standard thesaurus to obtain extended keywords; Based on the basic keywords and the extended keywords, a target keyword set is obtained.

3. The verification method for a power business system as described in claim 2, characterized in that, The step of obtaining the target keyword set based on the basic keywords and the extended keywords includes: Determine the similarity threshold based on the type of the policy text; The basic keywords and the extended keywords are combined to obtain an initial keyword set; Calculate the semantic similarity between each keyword in the initial keyword set and the policy text, and retain keywords with semantic similarity higher than the similarity threshold to obtain the target keyword set.

4. The verification method for a power business system as described in claim 3, characterized in that, Determining the similarity threshold based on the type of the policy text includes: If the policy text is of the type of legal provisions, then the first threshold is determined as the similarity threshold; If the policy text is of the technical specification type, then the second threshold is determined as the similarity threshold; If the policy text is of the notification or announcement type, then the third threshold is determined as the similarity threshold; wherein the first threshold is greater than the second threshold, and the second threshold is greater than the third threshold.

5. A verification device for a power business system, characterized in that, include: The module is used to divide the latest policy texts of the power industry into multiple paragraphs based on the trained semantic segmentation model, generate vector representations of the multiple paragraphs through text embedding technology, and use the vector representations of the multiple paragraphs as leaf nodes; wherein, the trained semantic segmentation model is obtained by training on a power policy corpus containing multiple policy texts with manually annotated paragraph boundaries; each paragraph's vector representation is treated as an independent cluster; the fidelity between clusters is calculated using a commutation test; the two clusters with the highest fidelity are merged until the number of clusters reaches a preset number, resulting in multiple clusters after clustering; a summary of each cluster is generated using a large language model; the vector representations of the paragraphs in each cluster are weighted and averaged to obtain the vector representation of each cluster; the summaries and vector representations of the multiple clusters are embedded into the parent node of the leaf node through recursive embedding until the root node is constructed, resulting in an index tree; wherein, the index tree contains multiple nodes, each node containing a node index, child nodes, a summary, and a vector representation; The determination module is used to determine the target keyword set based on the functional modules of the power business system; The retrieval module is used to perform a retrieval in the index tree based on the target keyword set to obtain key statements related to the power business system. The conversion module is used to identify the fields and constraint types in the key statements; based on a predefined field mapping table, it maps the fields in the key statements to fields in the database table; if the constraint type is a numeric range constraint, it generates an SQL statement with a BETWEEN condition or comparison operator based on the fields in the database table; if the constraint type is an enumeration value constraint, it generates an SQL statement with an IN condition based on the fields in the database table; if the constraint type is a regular expression constraint, it generates an SQL statement with a REGEXP match based on the fields in the database table. The verification module is used to execute the SQL statement in the database of the power business system and obtain the verification result; the verification result is used to indicate whether the power business system is compliant.

6. A computer device, characterized in that, The computer device includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the verification method for the power business system as described in any one of claims 1-4.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, and the processor executes the computer program to implement the verification method for the power business system as described in any one of claims 1-4.

Images