Method and system for constructing virtual knowledge graph of power distribution network

Abstract

The invention discloses a construction method and system for a virtual knowledge graph of a power distribution network, and relates to the technical field of graph construction, after a virtual user inputs a target domain demand, data in a virtual sub-chain is screened according to the target domain demand, and the data in the virtual sub-chain is obtained through confidence evaluation and a cost optimization algorithm. Data elements meeting the requirements of the virtual user are recognized and preferentially called, calling results form a virtual result set, the virtual result set is fed back to the virtual user, the virtual user generates virtual feedback based on the virtual result set, the confidence degree is dynamically updated according to the virtual feedback, then the virtual feedback is written back to three virtual sub-chains of the technical element, the price element and the efficiency element, and the technical element, the price element and the efficiency element are obtained. And completing the construction of the knowledge graph. The construction system adopts a data annotation and classification technology, fine and specific data calling can be conveniently realized under different scenes and requirements, a virtual feedback mechanism is introduced to dynamically update the confidence degree, it is ensured that the knowledge graph can be continuously optimized, and high timeliness and intelligence are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of graph construction, and in particular to a method and system for constructing a virtual knowledge graph of a distribution network. Background Art

[0002] As the process of digital transformation of enterprises continues to accelerate, knowledge graphs have developed rapidly in various industries. The construction of cross-organizational virtual knowledge graphs has become a valuable research direction. The construction of cross-organizational virtual knowledge graphs is the product of the deep integration of knowledge engineering, artificial intelligence, and business fields, providing a more open, efficient, and sustainable knowledge sharing and production model. The technical and economic analysis of distribution networks involves multiple market players and involves the entire distribution network cycle, including distribution network planning, design, construction, and operation. It is a typical application field of cross-organizational knowledge sharing. At present, the research areas of the academic community are mainly concentrated on theoretical calculations, reference price compilation, and typical engineering designs. With the integration and evolution of energy technology, artificial intelligence, and blockchain technology, distribution network knowledge graph practices have emerged in the industry. Vigorously promoting cross-organizational knowledge sharing and evolution will further promote the commercialization and practical application of knowledge graphs.

[0003] The prior art has the following defects:

[0004] 1. In existing distribution network data applications, the retrieved data is usually not evaluated for value, resulting in some data lacking practicality and pertinence. In addition, data processing is often based on general classification and fails to be refined to specific factors such as technology, price, and efficiency, making it difficult to efficiently support analysis of different needs.

[0005] 2. When constructing a knowledge graph, it is often static and fails to collect and utilize user feedback in real time to dynamically update the system, causing the graph content to lose timeliness over time. Summary of the invention

[0006] The purpose of the present invention is to provide a method and system for constructing a virtual knowledge graph for a distribution network, which adopts data labeling and classification technology to facilitate refined and specific data retrieval under different scenarios and requirements, and introduces a virtual feedback mechanism to dynamically update the confidence level, ensuring that the knowledge graph can be continuously optimized with high timeliness and intelligence.

[0007] In order to achieve the above object, the present invention provides the following technical solution: a method for constructing a virtual knowledge graph of a distribution network, the construction method comprising the following steps: the construction method comprising the following steps:

[0008] The construction system obtains data from the databases of various organizations and standardizes the data. Through data annotation and classification technology, the data is classified according to technical factors, price factors and efficiency factors, and corresponding virtual sub-chains are created;

[0009] After the virtual user inputs the target domain requirements, the data in the virtual sub-chain is screened according to the target domain requirements, and the data elements that meet the virtual user's requirements are identified and retrieved first through confidence evaluation and cost optimization algorithms;

[0010] The retrieved results are formed into a virtual result set and fed back to the virtual user. The virtual user generates virtual feedback based on the virtual result set. After dynamically updating the confidence level according to the virtual feedback, the virtual feedback is written back to the three virtual sub-chains of technical factors, price factors, and efficiency factors to complete the construction of the knowledge graph.

[0011] In a preferred embodiment, the data elements that meet the requirements of the virtual user are identified and preferentially retrieved through confidence evaluation and cost optimization algorithms, including the following steps:

[0012] Based on the demand conditions input by the virtual user, calculate the demand matching confidence of each data element;

[0013] Technical confidence: Evaluate the technical match between each data element and user requirements, and determine whether the technical parameters meet user requirements;

[0014] Price confidence: Check whether the price information of the data element is consistent with the user's cost budget;

[0015] Efficiency confidence: Evaluate the efficiency matching of data elements based on user's demand for efficiency;

[0016] The confidence levels of technology, price, and efficiency are combined to obtain an overall confidence score for each data element;

[0017] Cost optimization analysis of all data elements:

[0018] Prioritize the confidence-assessed and cost-optimized data elements.

[0019] In a preferred embodiment, the demand matching confidence of each data element is calculated, and the technology, price and efficiency confidence are comprehensively calculated to obtain the overall confidence score of each data element, including the following steps:

[0020] Evaluate whether the data elements meet the needs of the virtual user in the technical dimension, and set the technical confidence between 0 and 1 to indicate the matching degree of technical parameters;

[0021] Check whether the price of the data element meets the budget requirements of the virtual user. The price confidence is between 0 and 1, indicating the degree of price matching. The calculation expression of price confidence is:

[0022] In the formula, C 价格is the price confidence, B is the virtual user budget, P is the data element price, and max(*) means selecting the maximum value;

[0023] The efficiency matching degree of data elements is evaluated according to the user's efficiency requirements. The efficiency confidence is set between 0 and 1, indicating the matching degree of efficiency. The efficiency confidence calculation formula is:

[0024] In the formula, C 效率 is the efficiency confidence, E 需求 is the efficiency requirement of the virtual user, E is the data element efficiency, and min(*) indicates selecting the minimum value;

[0025] The confidence scores of technology, price and efficiency are combined to obtain the overall confidence score of each data element. The calculation expression is:

[0026] C 整体 =w 技术 ·C 技术 +w 价格 ·C 价格 +w 效率 ·C 效率 , where C 整体 is the overall confidence score, C 技术 is the technical confidence, C 价格 is the price confidence, C 效率 is the efficiency confidence, w 技术 、w 价格 、w 效率 are the weights of technology confidence, price confidence, and efficiency confidence, respectively, and C 技术 +C 价格 +C 效率 =1.

[0027] In a preferred embodiment, cost optimization analysis is performed on all data elements, and the data elements that have undergone confidence evaluation and cost optimization are sorted by priority, including the following steps:

[0028] Calculate the cost-effectiveness ratio for each data element using the formula: Where CBV is the cost-effectiveness ratio, C 整体 is the overall confidence score, P is the data element price;

[0029] The cost-effectiveness ratio of the data elements and the overall confidence score are normalized so that the value range of the cost-effectiveness ratio and the overall confidence score is mapped to [0,1]. The normalized cost-effectiveness ratio and the overall confidence score are summed to obtain the priority value. All data elements are sorted from large to small according to the priority value to generate a list of data elements.

[0030] In a preferred embodiment, after dynamically updating the confidence according to the virtual feedback, the virtual feedback is written back to the three virtual sub-chains of technical factors, price factors, and efficiency factors to complete the construction of the knowledge graph, including the following steps:

[0031] Extract data from the generated virtual feedback and classify the feedback information according to technical factors, price factors and efficiency factors;

[0032] Based on the collected feedback data, the confidence calculation model is used to dynamically update each factor, including weight adjustment and confidence score update;

[0033] Update node information in the virtual subchain based on feedback, including adding new data items, updating the properties of existing data items, or modifying data item weights;

[0034] Adjust the relationship between elements in the knowledge graph, integrate the updated virtual sub-chain data, and build or update the knowledge graph.

[0035] In a preferred embodiment, after the virtual user inputs the target domain requirements, the data in the virtual sub-chain is screened according to the target domain requirements, including the following steps:

[0036] Receive input requirements from virtual users, parse the requirements and extract relevant parameters, identify key factors, including technology categories, price ranges, and efficiency requirements, and convert user requirements into screening conditions through natural language processing technology;

[0037] Based on the extracted demand parameters, define the screening conditions in each virtual sub-chain:

[0038] Technical screening conditions: determine the technical elements and related parameters required by users;

[0039] Price screening conditions: limit the cost range;

[0040] Efficiency screening conditions: Determine the operating efficiency and energy consumption level requirements based on the efficiency target;

[0041] Screening is performed simultaneously in the technology, price and efficiency virtual sub-chains, and multi-factor cross-matching is used to screen out data that meets the conditions of each factor:

[0042] Use the intersection operation to retrieve data items that meet multiple factor conditions at the same time, and use labels and index structures for cross-matching.

[0043] In a preferred embodiment, data is classified into technical factors, price factors and efficiency factors through data annotation and classification technology, and corresponding virtual sub-chains are created, including the following steps:

[0044] Define data annotation rules based on business needs and the structure of the knowledge graph, including the following elements:

[0045] Technical elements: define the technical category, including equipment type, application scenario, and technical parameters;

[0046] Price elements: determine the cost-related annotation information, including purchase cost, maintenance cost, and update cost;

[0047] Efficiency factors: define efficiency indicators, including operational efficiency, resource utilization, and energy consumption levels;

[0048] Assign labels to data one by one based on labeling rules, label data in multiple dimensions according to technology, price, and efficiency factors, and assign multiple labels to data items involving multiple factors;

[0049] Classify and group the labeled data by factors, dividing them into technical factors, price factors, and efficiency factors;

[0050] Technical element classification: classify the data marked with technical labels into technical element data sets;

[0051] Price element classification: classify the data with price tags into price element datasets;

[0052] Efficiency factor classification: classify the data marked with efficiency labels into efficiency factor datasets;

[0053] According to the classified data, a corresponding virtual sub-chain is created for each element;

[0054] Technology virtual sub-chain: storing technology factor data;

[0055] Price virtual sub-chain: stores price factor data;

[0056] Efficiency virtual sub-chain: stores efficiency factor data;

[0057] An index structure is established for each virtual sub-chain, and the association between sub-chains is realized through mapping relationships.

[0058] In a preferred embodiment, it includes a virtual sub-chain creation module, an element retrieval module, and a graph construction module;

[0059] Virtual sub-chain creation module: obtain data from the databases of various organizations and standardize the data. Through data labeling and classification technology, classify the data according to technical factors, price factors and efficiency factors, and create corresponding virtual sub-chains;

[0060] Element retrieval module: After the virtual user inputs the target domain requirements, the data in the virtual sub-chain is screened according to the target domain requirements, and the data elements that meet the virtual user's requirements are identified and preferentially retrieved through confidence evaluation and cost optimization algorithms;

[0061] Graph construction module: The retrieved results are formed into a virtual result set and fed back to the virtual user. The virtual user generates virtual feedback based on the virtual result set. After dynamically updating the confidence level according to the virtual feedback, the virtual feedback is written back to the three virtual sub-chains of technical factors, price factors, and efficiency factors to complete the construction of the knowledge graph.

[0062] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0063] After the virtual user inputs the target domain requirements, the present invention screens the data in the virtual subchain according to the target domain requirements, identifies and preferentially retrieves the data elements that meet the virtual user's requirements through confidence evaluation and cost optimization algorithms, forms the retrieved results into a virtual result set, and feeds back to the virtual user. The virtual user generates virtual feedback based on the virtual result set, and after dynamically updating the confidence according to the virtual feedback, the virtual feedback is written back to the three virtual subchains of technical elements, price elements, and efficiency elements to complete the construction of the knowledge graph. The construction system adopts data annotation and classification technology to facilitate the realization of refined and specific data retrieval under different scenarios and requirements, and introduces a virtual feedback mechanism to dynamically update the confidence, ensuring that the knowledge graph can be continuously optimized with high timeliness and intelligence. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0065] Figure 1 The figure is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0066] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0067] Example 1: Please refer to Figure 1As shown, the method for constructing a virtual knowledge graph of a distribution network described in this embodiment includes the following steps:

[0068] The construction system obtains data from the databases of various organizations and standardizes the data. Through data labeling and classification technology, the data is classified according to technical factors, price factors and efficiency factors, and corresponding virtual sub-chains are created. After the virtual user enters the target field requirements, the data in the virtual sub-chain is screened according to the target field requirements. Through confidence evaluation and cost optimization algorithms, the data elements that meet the requirements of the virtual user are identified and retrieved first. Through correlation analysis, it is ensured that the retrieved data has high application value and can meet the specific needs of the virtual user in terms of technology, price and efficiency dimensions. The retrieved results are formed into a virtual result set and fed back to the virtual user. The virtual user generates virtual feedback based on the virtual result set. After dynamically updating the confidence according to the virtual feedback, the virtual feedback is written back to the three virtual sub-chains of technical factors, price factors and efficiency factors to complete the construction of the knowledge graph.

[0069] After the virtual user inputs the target domain requirements, this application screens the data in the virtual subchain according to the target domain requirements, identifies and prioritizes the data elements that meet the virtual user's requirements through confidence evaluation and cost optimization algorithms, forms the retrieved results into a virtual result set, and feeds back to the virtual user. The virtual user generates virtual feedback based on the virtual result set, and after dynamically updating the confidence according to the virtual feedback, the virtual feedback is written back to the three virtual subchains of technical elements, price elements, and efficiency elements to complete the construction of the knowledge graph. The construction system adopts data annotation and classification technology to facilitate the realization of refined and specific data retrieval under different scenarios and requirements, and introduces a virtual feedback mechanism to dynamically update the confidence, ensuring that the knowledge graph can be continuously optimized with high timeliness and intelligence.

[0070] Example 2: Building a system to obtain data from the databases of various organizations and standardize the data, including the following steps:

[0071] Extract raw data from the databases of various organizations, identify and connect data sources of different databases. The extracted data may contain multiple formats and structures, such as relational databases, document databases, etc. Establish data collection interfaces to achieve secure and stable data transmission.

[0072] Clean the extracted raw data to remove invalid, duplicate, missing or inconsistent data. The cleaning process may include: deleting redundant fields and duplicate records, filling missing values ​​or filling them according to preset rules, and correcting errors in the data, such as irregular formats and character encoding problems;

[0073] Convert cleaned data into a unified format for subsequent processing and standardization. Data conversion may include:

[0074] Data type conversion (such as string to date or value), parsing and converting unstructured data (such as document content, JSON format data) into structured data

[0075] The converted data is standardized to ensure the consistency of data format and content. The main standardization steps include:

[0076] Define and apply standard field names, standardize and unify similar fields of different organizations, set data range and precision according to business needs and industry standards, and standardize the valid range and decimal places of data values; standardize the format of fields such as date, time and geographic location information to ensure consistency among organizations;

[0077] Data annotation is performed on standardized data sets, and labels are added based on distribution network-related technology, price, efficiency and other factors to support subsequent classification and classified storage. The annotation process usually includes the following:

[0078] Label according to specific elements (such as technology category, application scenario);

[0079] Add metadata (such as data source, update time, accuracy) as auxiliary information;

[0080] Based on the annotation information, the data is classified into virtual sub-chains of different dimensions such as technical factors, price factors, and efficiency factors, which facilitates subsequent rapid call and accurate screening. The classification method is defined according to the specific characteristics and application requirements of the data, and an index mechanism is established to improve the speed of data query and retrieval.

[0081] Store standardized data by virtual sub-chain and establish an efficient index structure to support fast query and call. Store different categories of data separately during storage to enhance data scalability. Indexing mechanisms can include primary key indexes, full-text indexes, or other multi-dimensional indexes to speed up data retrieval. Regularly verify standardized data to ensure data quality. Verification content includes field integrity, format consistency, and data timeliness. Quality monitoring should also track data changes and issue warnings in a timely manner when anomalies are detected. Establish a regular update and synchronization mechanism to ensure that new data in the databases of each organization can be automatically collected and added to the system database after standardized processing to maintain the real-time and consistency of data.

[0082] Through data labeling and classification technology, the data is classified according to technical factors, price factors and efficiency factors, and the corresponding virtual sub-chains are created, including the following steps:

[0083] Define data annotation rules based on business needs and the structure of the knowledge graph, including the following elements:

[0084] Technical elements: define the technical categories (such as equipment type, application scenario, technical parameters, etc.) and clarify the technical details involved.

[0085] Price elements: Determine the cost-related annotation information (such as purchase cost, maintenance cost, update cost, etc.).

[0086] Efficiency factors: define efficiency indicators (such as operating efficiency, resource utilization, energy consumption level, etc.) and mark relevant data with efficiency.

[0087] If the data volume is large or the data attributes are complex, machine learning or natural language processing (NLP) models can be introduced for automatic labeling:

[0088] Train the model based on historical data or manually annotated data to generate an annotated model.

[0089] Apply the annotation model to automatically annotate new data, generate multi-dimensional labels, and improve annotation efficiency.

[0090] Based on the labeling rules and labeling models, labels are assigned to the data one by one, and the data is labeled in multiple dimensions according to factors such as technology, price, and efficiency:

[0091] Label the data according to the three major factors of technology, price and efficiency.

[0092] For data items involving multiple elements, multiple labels are assigned to ensure the comprehensiveness and refinement of the annotations.

[0093] Verify the quality of annotations through manual sampling to ensure the accuracy of model or rule annotations. Verification includes:

[0094] Check whether the labeling complies with the rules. Correct incorrect labeling, especially the cross-labeling of multi-factor data.

[0095] The labeled data are classified and grouped by factors, and divided into three categories: technical factors, price factors, and efficiency factors: Technical factor classification: the data labeled with technical tags are classified as technical factor data sets, which are subdivided into different technical categories. Price factor classification: the data labeled with price tags are classified as price factor data sets, which are subdivided into different cost dimensions. Efficiency factor classification: the data labeled with efficiency tags are classified as efficiency factor data sets, which are subdivided into different efficiency indicators.

[0096] According to the classified data, a corresponding virtual sub-chain is created for each element: Technology virtual sub-chain: stores technology element data, which is convenient for retrieval and analysis based on technical attributes. Price virtual sub-chain: stores price element data, which is convenient for retrieval based on cost optimization needs. Efficiency virtual sub-chain: stores efficiency element data, which supports analysis needs based on performance and efficiency.

[0097] Establish an index structure for each virtual sub-chain to improve data retrieval efficiency, and realize the association between sub-chains through mapping relationships: create multi-dimensional indexes to quickly retrieve data by technology, price, and efficiency factors. Establish mapping relationships between data to facilitate cross-factor queries or retrieve multi-factor comprehensive data.

[0098] After the virtual user inputs the target domain requirements, the data in the virtual sub-chain is screened according to the target domain requirements, including the following steps:

[0099] Receive input requirements from virtual users, parse the requirements and extract relevant parameters, and identify key elements such as technology category, price range, efficiency requirements, etc. Through natural language processing (NLP) technology or keyword matching algorithm, convert user requirements into specific screening conditions.

[0100] Based on the extracted demand parameters, define the screening conditions in each virtual sub-chain:

[0101] Technical screening conditions: Determine the technical elements and related parameters (such as type, applicable scenarios) required by users.

[0102] Price screening conditions: limit the cost range (such as purchase cost, maintenance cost, etc.) to filter data that meets the price requirements.

[0103] Efficiency screening conditions: According to the efficiency target, determine the requirements such as operating efficiency and energy consumption level, and screen the data that meet the efficiency standards.

[0104] Screening is performed simultaneously in the technology, price and efficiency virtual sub-chains, and multi-factor cross-matching is used to screen out data that meets the conditions of each factor:

[0105] Use intersection operations to retrieve data items that meet multiple conditions at the same time, ensuring that the data meets both technical requirements and meets the conditions within the price and efficiency range. Use labels and index structures to speed up the cross-matching process and improve screening efficiency.

[0106] Receive the input requirements of virtual users, parse the requirements and extract relevant parameters, identify key factors, including technology categories, price ranges, and efficiency requirements, and convert user requirements into screening conditions through natural language processing technology; define screening conditions in each virtual sub-chain based on the extracted demand parameters: technical screening conditions: determine the technical factors and related parameters required by the user; price screening conditions: limit the cost range; efficiency screening conditions: determine the operating efficiency and energy consumption level requirements based on the efficiency target; screen in the technology, price and efficiency virtual sub-chains at the same time, and use multi-factor cross-matching to screen out data that meets the conditions of each factor: use the intersection operation to retrieve the data items that meet the conditions of multiple factors at the same time, and use the label and index structure for cross-matching, including the following steps:

[0107] Assume that after receiving the following virtual user requirements, the system extracts the following screening conditions:

[0108] Technology category: New energy technology;

[0109] Price range: less than 5000;

[0110] Efficiency requirement: higher than 90%;

[0111] Assume that the data items in the three virtual subchains are as shown in Table 1 (for example only):

[0112]

[0113]

[0114] Table 1

[0115] The results after filtering are as follows:

[0116] Technical screening conditions (new energy): filter out data items A, C, D, and E.

[0117] Price filter condition (less than 5000): filter out data items A, B, D, and E.

[0118] Efficiency screening condition (efficiency>90%): filter out data items A, C, and E.

[0119] Through the intersection operation, the data items that meet all conditions are retained, namely, data items A and E.

[0120] The screening formula is as follows:

[0121] S 技术 ={x|Technology Category=New Energy};

[0122] S 价格 ={x|price≤5000};

[0123] S效率 ={x|efficiency ≥ 90%};

[0124] Intersection matching formula: S 交叉 =S 技术 ∩S 价格 ∩S 效率 ;

[0125] Find the intersection of the screening result sets of each condition, and finally get the intersection set that meets all conditions: S 交叉 ={A,E}; In this way, the system can filter out data that meets all the element conditions from the virtual sub-chain and provide the most suitable result set for the virtual user based on the cross-matching results.

[0126] Through confidence assessment and cost optimization algorithms, identify and prioritize data elements that meet the requirements of virtual users. Through correlation analysis, ensure that the retrieved data has high application value and can meet the specific needs of virtual users in terms of technology, price and efficiency. The following steps are included:

[0127] Based on the requirements input by the virtual user, the requirements matching confidence of each data element is calculated. The confidence calculation mainly includes the following aspects:

[0128] Technical confidence: Evaluate the technical match between each data element and user requirements, and determine whether the technical parameters (such as specifications, type, and applicability) meet user requirements.

[0129] Price confidence: Check whether the price information of the data element is consistent with the user's cost budget or expected price.

[0130] Efficiency confidence: Evaluate the efficiency matching of data elements based on user requirements for efficiency (such as operating efficiency and resource utilization).

[0131] The confidence of technology, price and efficiency is calculated comprehensively to obtain the overall confidence score of each data element, which is used as the basis for identifying priority data. The comprehensive confidence calculation method can be based on weighted summation, and the weight value is determined by the user's priority for each element.

[0132] Conduct cost optimization analysis on data elements, including:

[0133] Cost model selection: Select an appropriate cost optimization model based on the application scenario, such as the lowest cost model or the cost-effectiveness model.

[0134] Cost optimization algorithm: Apply cost optimization algorithms (such as linear programming and dynamic programming) to perform cost calculations, giving priority to retrieving data elements with lower costs and higher matching degrees.

[0135] Prioritize the data elements that have undergone confidence assessment and cost optimization to ensure that the data is called in the order of meeting user needs and cost-effectiveness.

[0136] Correlation analysis is used to ensure that the retrieved data has high application value. The specific analysis steps include:

[0137] Multi-dimensional correlation analysis: Check the correlation of data elements in the dimensions of technology, price, and efficiency to ensure that data items can complement and enhance each other.

[0138] Historical relevance verification: Verify the applicability of data items in real scenarios based on historical data and domain knowledge base, and filter out potential low-relevance or inapplicable data items.

[0139] Consistency check: Check whether the selected data is consistent in multiple dimensions to ensure stability and consistency in actual applications.

[0140] Calculate the demand matching confidence of each data element, and calculate the technology, price and efficiency confidence comprehensively to get the overall confidence score of each data element, including the following steps:

[0141] Evaluate whether the data element meets the needs of the virtual user in the technical dimension, set the technical confidence between 0 and 1, and indicate the matching degree of the technical parameters. Assume that the technical confidence is based on the matching degree of the technical attributes, such as the exact match score is 1, the partial match score is 0.8, and the complete non-match score is 0; if the virtual user's demand is "new energy technology", and the technical attribute of data element A is "new energy", then the technical confidence is equal to 1; if the technical attributes are partially matched, such as "renewable energy", then the technical confidence is equal to 0.8;

[0142] Check whether the price of the data element meets the budget requirements of the virtual user. The price confidence is between 0 and 1, indicating the degree of price matching. The calculation expression of price confidence is:

[0143] In the formula, C 价格 is the price confidence, B is the virtual user budget, P is the data element price, max(*) means selecting the maximum value, when the data element price is equal to the user budget, the price confidence is 1; as the price deviates from the budget, the price confidence gradually decreases;

[0144] The efficiency matching degree of data elements is evaluated according to the user's efficiency requirements. The efficiency confidence is set between 0 and 1, indicating the matching degree of efficiency. The efficiency confidence calculation formula is:

[0145] In the formula, C 效率 is the efficiency confidence, E 需求is the virtual user efficiency requirement, E is the data element efficiency, min(*) indicates selecting the minimum value, when the data element efficiency is greater than or equal to the user requirement, the efficiency confidence is 1; when it is lower than the requirement, the efficiency confidence is calculated based on the relative efficiency.

[0146] The confidence scores of technology, price and efficiency are combined to obtain the overall confidence score of each data element. The calculation expression is:

[0147] C 整体 =w 技术 ·C 技术 +w 价格 ·C 价格 +w 效率 ·C 效率 , where C 整体 is the overall confidence score, C 技术 is the technical confidence, C 价格 is the price confidence, C 效率 is the efficiency confidence, w 技术 、w 价格 、w 效率 are the weights of technology confidence, price confidence, and efficiency confidence, respectively, and C 技术 +C 价格 +C 效率 =1.

[0148] Perform cost optimization analysis on all data elements and prioritize the data elements that have undergone confidence assessment and cost optimization, including the following steps:

[0149] On the premise of meeting user needs, data elements with high cost performance are given priority, and the cost-effectiveness ratio of each data element is calculated to maximize the ratio of overall confidence to price. The cost-effectiveness ratio calculation formula is: Where CBV is the cost-effectiveness ratio, C 整体 is the overall confidence score, P is the data element price;

[0150] The cost-effectiveness ratio of the data elements and the overall confidence score are normalized so that the value range of the cost-effectiveness ratio and the overall confidence score is mapped to [0,1]. The normalized cost-effectiveness ratio and the overall confidence score are summed to obtain the priority value. All data elements are sorted from large to small according to the priority value to generate a list of data elements.

[0151] The retrieved results are formed into a virtual result set and fed back to the virtual user. The virtual user generates virtual feedback based on the virtual result set, including the following steps:

[0152] The retrieved data elements that meet the needs of the virtual user are integrated to form a virtual result set, ensuring that the data items contained in the result set fully meet the needs in terms of technology, price, and efficiency. The steps to generate a virtual result set include:

[0153] Information integration: The key information of the screening results (such as technical parameters, cost details, efficiency indicators) are concentrated in the result set so that users can view the data details at a glance.

[0154] Formatted presentation: Sort the virtual result sets by category and priority, and present them in a structured format (such as tables, charts, etc.) to improve the readability and logical clarity of the information.

[0155] Result set preview and description

[0156] Before providing feedback to the virtual user, a concise result set preview and description is generated to help the virtual user quickly understand the overall situation of the screening results. The description includes:

[0157] Overall Match: Displays the overall confidence or match score of the result set in meeting the requirement.

[0158] Key Indicator Overview: Displays the main indicators of the optimal data items (such as technical attributes, price range, efficiency level) and their specific values.

[0159] The virtual result set is fully fed back to the virtual user, with options for further filtering and customization. The feedback steps include:

[0160] User interaction interface: Presents the result set through a user-friendly interface, provides interactive functions such as sorting and filtering, and allows users to customize the viewing method.

[0161] Detailed information expansion: Provides an expanded details option for each data item, allowing users to drill down to the technical, price, efficiency and other detailed information of individual data.

[0162] The virtual user generates feedback based on the content of the result set and the degree to which the requirements are met. The virtual feedback includes the following:

[0163] Satisfaction feedback: The user's overall satisfaction rating for the result set, reflecting their satisfaction with the data in terms of technology, price, and efficiency.

[0164] Improvement suggestions: Users can enter specific improvement suggestions, such as adjusting filter conditions, priority settings, or adding specific data dimensions to help the system optimize the filtering logic.

[0165] Individual data evaluation: Users can evaluate individual data items to indicate whether they meet expectations or whether there is room for improvement to meet higher requirements.

[0166] After dynamically updating the confidence level based on the virtual feedback, the virtual feedback is written back to the three virtual sub-chains of technical factors, price factors, and efficiency factors to complete the construction of the knowledge graph, including the following steps:

[0167] Data Extraction: Extract key data from user-generated virtual feedback, including satisfaction ratings, specific recommendations, and data reviews.

[0168] Feedback classification: Classify feedback information according to technical factors, price factors and efficiency factors to facilitate subsequent processing and feedback.

[0169] Confidence calculation model: Based on the collected feedback data, the confidence calculation model is used to dynamically update each factor. The specific steps include:

[0170] Weight adjustment: According to the user's evaluation of each factor, the weight in the confidence calculation is adjusted. For example, if the user emphasizes the importance of price in the feedback, the weight of the price factor is increased.

[0171] Confidence score update: The confidence score is recalculated for each data item to reflect its performance and importance in user feedback.

[0172] Definition of virtual sub-chain: Ensure that the virtual sub-chain structure of technical factors, price factors and efficiency factors is clear and can effectively accommodate feedback information.

[0173] Information integration: Integrate the updated confidence score and user feedback information into the corresponding virtual subchain. The specific steps include:

[0174] Node update: Update the node information in the virtual subchain based on feedback, such as adding new data items, updating the properties of existing data items, or modifying their weights.

[0175] Edge relationship adjustment: Adjust the relationship between elements in the knowledge graph to ensure that the relevance and priority in the new feedback are reflected.

[0176] Build a knowledge graph: Integrate the updated virtual sub-chain data to build or update the knowledge graph to ensure that it reflects the latest user needs and market changes.

[0177] Performance evaluation: Conduct performance evaluation on the constructed knowledge graph to verify its effectiveness and accuracy in terms of technology, price, and efficiency.

[0178] Extract data from the generated virtual feedback, and classify the feedback information according to technical factors, price factors, and efficiency factors; based on the collected feedback data, use the confidence calculation model to dynamically update each factor, including weight adjustment and confidence score update; update the node information in the virtual sub-chain according to the feedback, including adding new data items, updating the attributes of existing data items, or modifying the weight of data items; adjust the relationship between the elements in the knowledge graph, integrate the updated virtual sub-chain data, and build or update the knowledge graph, including the following steps:

[0179] In the virtual user feedback, extract and classify the feedback information into technical, price and efficiency factors:

[0180] Technical elements: Feedback information involves product performance, technical functions, compatibility, etc. For example, feedback "new energy technology has high applicability" belongs to the technical elements.

[0181] Price factor: The feedback information involves the rationality of the price, budget matching, etc. For example, the feedback "the price is higher than expected" belongs to the price factor.

[0182] Efficiency factor: Feedback information involves efficiency, response speed, product capacity, etc. For example, feedback "operation efficiency is faster than expected" belongs to the efficiency factor.

[0183] Re-adjust the weights of technology, price, and efficiency based on the frequency and importance of feedback. If user feedback is more focused on technology, adjust the weights to increase the weight of technical factors. For example, adjust the weight of technology from 1 / 3 to 0.5, and reduce the weights of price and efficiency to 0.25.

[0184] In the virtual subchain, node data is dynamically updated based on feedback, including the following operations:

[0185] (1) Add new data items

[0186] If a new data item (such as a new technical indicator) is mentioned in the feedback, the data item is added to the virtual subchain and its confidence score and weight are initialized.

[0187] (2) Update the attributes of existing data items

[0188] Adjust the properties of existing data items based on feedback, such as increasing the technical confidence of the "New Energy Applicability" node.

[0189] (3) Modify data item weights

[0190] Adjust the weight of data items in the subchain according to their importance in the feedback. For example, if efficiency becomes a priority, increase the weight of the relevant nodes.

[0191] Assume that the nodes in the original virtual sub-chain are "new energy technology applicability", "price" and "efficiency", and are updated based on feedback as shown in Table 2:

[0192] Data Item Original confidence New confidence Original weight New weights Applicability of new energy technologies 0.8 0.9 0.3 0.4 price 0.7 0.65 0.35 0.3 efficiency 0.85 0.9 0.35 0.3

[0193] Table 2

[0194] In the knowledge graph, the relationship between elements is adjusted through feedback information, and new connections are constructed:

[0195] (1) Identify the relationship between elements

[0196] The feedback information is converted into a relational link to represent the interaction between technology, price, and efficiency. For example, if the feedback indicates that price affects efficiency confidence, a "price-efficiency" relation edge is established in the knowledge graph.

[0197] (2) Build or update the knowledge graph structure

[0198] In the knowledge graph, graph nodes and edges are established based on the updated virtual sub-chain data and relationships.

[0199] Example: In the knowledge graph, establish relationship links between "applicability of new energy technologies" and "price" and "efficiency", and mark edge weights to indicate their association strength.

[0200] Assume that the virtual user feedback is as follows:

[0201] “New energy technologies are highly applicable and can meet the latest needs”;

[0202] “The price is beyond budget”;

[0203] “Efficiency is higher than expected, and response speed is faster”;

[0204] Based on the above feedback, we can make the following updates:

[0205] In the virtual sub-chain, the confidence level of “applicability of new energy technologies” is increased to 0.9, and the weight is increased to 0.4; the confidence level of “efficiency” is increased to 0.9.

[0206] In the knowledge graph, a strong connection between “new energy technology applicability” and “efficiency” is constructed, and a new “price-efficiency” edge is added to represent the potential impact of price on efficiency.

[0207] Finally, an updated knowledge graph structure is generated for subsequent data decision-making and recommendation.

[0208] Through the above steps, we have realized the complete process of extracting data from feedback, classifying, updating confidence, adjusting weights, updating virtual sub-chains and building knowledge graphs to support more intelligent data management and user demand response.

[0209] Embodiment 3: A system for constructing a virtual knowledge graph of a distribution network described in this embodiment includes a virtual subchain creation module, an element retrieval module, and a graph construction module;

[0210] Virtual sub-chain creation module: obtains data from the databases of various organizations and standardizes the data. Through data annotation and classification technology, the data is classified according to technical factors, price factors and efficiency factors, and the corresponding virtual sub-chain is created. The virtual sub-chain is sent to the element retrieval module and the graph construction module;

[0211] Element retrieval module: After the virtual user inputs the target domain requirements, the data in the virtual sub-chain is screened according to the target domain requirements. Through confidence evaluation and cost optimization algorithms, the data elements that meet the virtual user's requirements are identified and retrieved first. Through correlation analysis, it is ensured that the retrieved data has high application value and can meet the specific needs of the virtual user in terms of technology, price and efficiency. The retrieval results are sent to the graph construction module;

[0212] Graph construction module: The retrieved results are formed into a virtual result set and fed back to the virtual user. The virtual user generates virtual feedback based on the virtual result set. After dynamically updating the confidence level according to the virtual feedback, the virtual feedback is written back to the three virtual sub-chains of technical factors, price factors, and efficiency factors to complete the construction of the knowledge graph.

[0213] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0214] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0215] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to only specific implementation methods. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for constructing a virtual knowledge graph of a distribution network, characterized in that: The construction method comprises the following steps: The construction system obtains data from the databases of various organizations and standardizes the data. Through data annotation and classification technology, the data is classified according to technical factors, price factors and efficiency factors, and corresponding virtual sub-chains are created; After the virtual user inputs the target domain requirements, the data in the virtual sub-chain is screened according to the target domain requirements, and the data elements that meet the virtual user's requirements are identified and retrieved first through confidence evaluation and cost optimization algorithms; The retrieved results are formed into a virtual result set and fed back to the virtual user. The virtual user generates virtual feedback based on the virtual result set. After dynamically updating the confidence level according to the virtual feedback, the virtual feedback is written back to the three virtual sub-chains of technical factors, price factors, and efficiency factors to complete the construction of the knowledge graph.

2. The method for constructing a virtual knowledge graph of a distribution network according to claim 1, characterized in that: Through confidence evaluation and cost optimization algorithms, data elements that meet the requirements of virtual users are identified and prioritized, including the following steps: Based on the demand conditions input by the virtual user, calculate the demand matching confidence of each data element; Technical confidence: Evaluate the technical match between each data element and user requirements, and determine whether the technical parameters meet user requirements; Price confidence: Check whether the price information of the data element is consistent with the user's cost budget; Efficiency confidence: Evaluate the efficiency matching of data elements based on user's demand for efficiency; The confidence levels of technology, price, and efficiency are combined to obtain an overall confidence score for each data element; Cost optimization analysis of all data elements: Prioritize the confidence-assessed and cost-optimized data elements.

3. The method for constructing a virtual knowledge graph of a distribution network according to claim 2, characterized in that: Calculate the demand matching confidence of each data element, and calculate the technology, price and efficiency confidence comprehensively to get the overall confidence score of each data element, including the following steps: Evaluate whether the data elements meet the needs of the virtual user in the technical dimension, and set the technical confidence between 0 and 1 to indicate the matching degree of technical parameters; Check whether the price of the data element meets the budget requirements of the virtual user. The price confidence is between 0 and 1, indicating the degree of price matching. The calculation expression of price confidence is: In the formula, C 价格 is the price confidence, B is the virtual user budget, P is the data element price, and max(*) means selecting the maximum value; The efficiency matching degree of data elements is evaluated according to the user's efficiency requirements. The efficiency confidence is set between 0 and 1, indicating the matching degree of efficiency. The efficiency confidence calculation formula is: In the formula, C 效率 is the efficiency confidence, E 需求 is the efficiency requirement of the virtual user, E is the data element efficiency, and min(*) indicates selecting the minimum value; The confidence scores of technology, price and efficiency are combined to obtain the overall confidence score of each data element. The calculation expression is: C 整体 =w 技术 ·C 技术 +w 价格 ·C 价格 +w 效率 ·C 效率 , where C 整体 is the overall confidence score, C 技术 is the technical confidence, C 价格 is the price confidence, C 效率 is the efficiency confidence, w 技术 、w 价格 、w 效率 are the weights of technology confidence, price confidence, and efficiency confidence, respectively, and C 技术 +C 价格 +C 效率 =1.

4. The method for constructing a virtual knowledge graph of a distribution network according to claim 3, characterized in that: Perform cost optimization analysis on all data elements and prioritize the data elements that have undergone confidence assessment and cost optimization, including the following steps: Calculate the cost-effectiveness ratio for each data element using the formula: Where CBV is the cost-effectiveness ratio, C 整体 is the overall confidence score, P is the data element price; The cost-effectiveness ratio of the data elements and the overall confidence score are normalized so that the value range of the cost-effectiveness ratio and the overall confidence score is mapped to [0,1]. The normalized cost-effectiveness ratio and the overall confidence score are summed to obtain the priority value. All data elements are sorted from large to small according to the priority value to generate a list of data elements.

5. The method for constructing a virtual knowledge graph of a distribution network according to claim 4, characterized in that: After dynamically updating the confidence level based on the virtual feedback, the virtual feedback is written back to the three virtual sub-chains of technical factors, price factors, and efficiency factors to complete the construction of the knowledge graph, including the following steps: Extract data from the generated virtual feedback and classify the feedback information according to technical factors, price factors and efficiency factors; Based on the collected feedback data, the confidence calculation model is used to dynamically update each factor, including weight adjustment and confidence score update; Update node information in the virtual subchain based on feedback, including adding new data items, updating the properties of existing data items, or modifying data item weights; Adjust the relationship between elements in the knowledge graph, integrate the updated virtual sub-chain data, and build or update the knowledge graph.

6. The method for constructing a virtual knowledge graph of a distribution network according to claim 5, characterized in that: After the virtual user inputs the target domain requirements, the data in the virtual sub-chain is screened according to the target domain requirements, including the following steps: Receive input requirements from virtual users, parse the requirements and extract relevant parameters, identify key factors, including technology categories, price ranges, and efficiency requirements, and convert user requirements into screening conditions through natural language processing technology; Based on the extracted demand parameters, define the screening conditions in each virtual sub-chain: Technical screening conditions: determine the technical elements and related parameters required by users; Price screening conditions: limit the cost range; Efficiency screening conditions: Determine the operating efficiency and energy consumption level requirements based on the efficiency target; Screening is performed simultaneously in the technology, price and efficiency virtual sub-chains, and multi-factor cross-matching is used to screen out data that meets the conditions of each factor: Use the intersection operation to retrieve data items that meet multiple factor conditions at the same time, and use labels and index structures for cross-matching.

7. The method for constructing a virtual knowledge graph of a distribution network according to claim 6, characterized in that: Through data labeling and classification technology, the data is classified according to technical factors, price factors and efficiency factors, and the corresponding virtual sub-chains are created, including the following steps: Define data annotation rules based on business needs and the structure of the knowledge graph, including the following elements: Technical elements: define the technical category, including equipment type, application scenario, and technical parameters; Price elements: determine the cost-related annotation information, including purchase cost, maintenance cost, and update cost; Efficiency factors: define efficiency indicators, including operational efficiency, resource utilization, and energy consumption levels; Assign labels to data one by one based on labeling rules, label data in multiple dimensions according to technology, price, and efficiency factors, and assign multiple labels to data items involving multiple factors; Classify and group the labeled data by factors, dividing them into technical factors, price factors, and efficiency factors; Technical element classification: classify the data marked with technical labels into technical element data sets; Price element classification: classify the data with price tags into price element datasets; Efficiency factor classification: classify the data marked with efficiency labels into efficiency factor datasets; According to the classified data, a corresponding virtual sub-chain is created for each element; Technology virtual sub-chain: storing technology factor data; Price virtual sub-chain: stores price factor data; Efficiency virtual sub-chain: stores efficiency factor data; An index structure is established for each virtual sub-chain, and the association between sub-chains is realized through mapping relationships.

8. A system for constructing a virtual knowledge graph of a distribution network according to claim 7, characterized in that: Including virtual sub-chain creation module, element retrieval module, and graph construction module; Virtual sub-chain creation module: obtain data from the databases of various organizations and standardize the data. Through data labeling and classification technology, classify the data according to technical factors, price factors and efficiency factors, and create corresponding virtual sub-chains; Element retrieval module: After the virtual user inputs the target domain requirements, the data in the virtual sub-chain is screened according to the target domain requirements, and the data elements that meet the virtual user's requirements are identified and preferentially retrieved through confidence evaluation and cost optimization algorithms; Graph construction module: The retrieved results are formed into a virtual result set and fed back to the virtual user. The virtual user generates virtual feedback based on the virtual result set. After dynamically updating the confidence level according to the virtual feedback, the virtual feedback is written back to the three virtual sub-chains of technical factors, price factors, and efficiency factors to complete the construction of the knowledge graph.

Images