A system and method for constructing a standard system model

Abstract

The application provides a standard system model construction system and method, and relates to the technical field of data processing, and comprises the following steps: a parsing module is used for content parsing of standard files in a standard file set, and a dependency network between standards is generated; a positioning module is used for determining that two system observation points are located at an intersection core area of a business stage logical reference axis and a technical level logical reference axis in a two-dimensional logical space constituted by a logical topology skeleton, and determining that three system observation points are respectively located at left upper, right upper and lower edge logical areas of the logical space away from the intersection core area. Through the cooperation of multiple modules, the application realizes the following steps: the target industry full life cycle work range is combed, the standard dependency relationship is parsed, the observation point is positioned, the evaluation reference plane is constructed, the standard distribution is compared and analyzed, the network is corrected and integrated, and finally the standard system model covering the full life cycle of the industry and fitting the actual technical logic is generated, thereby providing comprehensive and objective support for the industrial standardization management.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a system and method for constructing a standard system model. Background Technology

[0002] In the process of constructing an industry standard system model, existing methods usually rely on the experience of domain experts for manual sorting. For example, when constructing a standard system for a sub-sector of manufacturing (such as the three-electric system of new energy vehicles), the expert team may need to manually collect, read and analyze hundreds of standard documents covering R&D, testing, production and recycling. Then, based on their understanding of the industry process, they manually classify the standard content into pre-set, lifecycle-based framework nodes. This type of method may have some limitations when faced with a large number of standard documents, highly specialized technical content and complex industry processes.

[0003] On the one hand, the efficiency of manual processing is relatively limited, and the cycle of system construction and subsequent updates may be long. On the other hand, the integrity of the system architecture and the accuracy of standard classification largely depend on the personal knowledge and experience of the participating experts. When implemented by different teams or at different times, inconsistencies may occur in the system framework or mapping results. When industry technology iterates or standards are updated, the maintenance and adjustment of the existing system usually also require a lot of manual labor for re-comparison and classification, making it difficult to reflect the latest status of the standard system in a timely and dynamic manner. Summary of the Invention

[0004] The technical problem to be solved by this invention is to provide a system and method for constructing a standard system model, which avoids the subjective bias of manually setting correlations, reduces the human cost of standardization management, and ensures the objectivity and consistency of model results.

[0005] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows:

[0006] Firstly, a system for constructing a standard system model, comprising:

[0007] The modeling module is used to generate a work scope catalog covering the entire life cycle of the target industry system based on knowledge of the target industry system, and extract two dimensions, business stage and technology level, as logical baseline axes to construct a logical topology skeleton;

[0008] The parsing module is used to parse the content of standard files in the standard file set and generate a dependency network between standards.

[0009] The positioning module is used to determine, within the two-dimensional logical space formed by the logical topology skeleton, the intersection core area of ​​two system observation points located at the business stage logical reference axis and the technical level logical reference axis, and to determine the upper left, upper right and lower edge logical areas of three system observation points located in the logical space far from the intersection core area.

[0010] The module is used to determine an evaluation reference plane in a three-dimensional logical attribute space formed by expanding a two-dimensional logical space, based on the coordinates of three system observation points and in combination with standard density attributes.

[0011] The extraction module maps the standard dependency network to the evaluation reference plane to form a standard coverage area and defines a logical polygon based on the boundary of the standard coverage area; it generates a logical ellipse based on the distribution of the five system observation points, detects and extracts the intersection and difference areas of the logical polygon and the logical ellipse on the evaluation reference plane, and generates standard distribution comparison data.

[0012] The model generation module is used to correct and integrate the standard dependency network mapped to the logical topology skeleton based on the standard distribution comparison data, and generate a standard system model of the entire life cycle of the target industry system.

[0013] Secondly, a method for constructing a standard system model includes:

[0014] Step 1: Based on the knowledge of the target industry system, generate a work scope catalog covering the entire life cycle of the target industry system, and extract two dimensions, business stage and technology level, as logical baselines to construct a logical topology skeleton;

[0015] Step 2: Parse the content of the standard files in the standard file set to generate a dependency network between the standards;

[0016] Step 3: Within the two-dimensional logical space formed by the logical topology skeleton, determine that two system observation points are located in the core area where the business stage logical reference axis and the technical level logical reference axis intersect, and determine that three system observation points are located in the upper left, upper right and lower edge logical areas of the logical space, respectively, far from the core area where the intersection occurs.

[0017] Step 4: Based on the coordinates of the three system observation points and combined with the standard density attribute, solve for and determine an evaluation reference plane in the three-dimensional logical attribute space formed by expanding the two-dimensional logical space.

[0018] Step 5: Map the standard dependency network onto the evaluation reference plane to form a standard coverage area, and define a logical polygon based on the boundary of the standard coverage area; generate a logical ellipse based on the distribution of the five system observation points, detect and extract the intersection and difference areas of the logical polygon and the logical ellipse on the evaluation reference plane, and generate standard distribution comparison data.

[0019] Step 6: Based on the standard distribution comparison data, correct and integrate the standard dependency network mapped to the logical topology skeleton to generate a standard system model for the entire life cycle of the target industry system.

[0020] The above-described solution of the present invention has at least the following beneficial effects:

[0021] It eliminates the need for domain experts to manually collect, organize, and classify massive amounts of standard documents. Through automated collaboration across modules, it efficiently advances core processes such as generating work scope directories and analyzing standard dependencies, shortening the construction and subsequent update cycle of the standard system. A topological framework is constructed using business stages and technical levels as fixed logical benchmarks. Combined with content parsing and spatial mapping technologies, it quantifies the correlation characteristics and distribution patterns of standards, avoiding subjective biases caused by differences in individual knowledge and experience during manual classification. This ensures consistency in the system framework and standard mapping results built by different teams at different times. Based on comparative standard distribution data, it can quickly identify system deviations caused by industry technology iterations or standard updates, and efficiently complete adjustments through a network correction mechanism. This allows the standard system to reflect the latest state of industry standards in real time, solving the problems of high system maintenance costs and delayed response. Through the difference analysis of logical polygons and logical ellipses, it locates standard coverage blind spots and redundant nodes. The generated standard system model can fully cover the entire lifecycle of the target industry, while clearly defining the dependencies between standards, providing clear and reliable guidance for standardization practices at all stages of the industry. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of a standard system model construction system provided by an embodiment of the present invention.

[0023] Figure 2 This is a flowchart illustrating a method for constructing a standard system model according to an embodiment of the present invention. Detailed Implementation

[0024] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0025] like Figure 1 As shown, an embodiment of the present invention proposes a system for constructing a standard system model, comprising:

[0026] The modeling module is used to generate a work scope catalog covering the entire life cycle of the target industry system based on knowledge of the target industry system, and extract two dimensions, business stage and technology level, as logical baseline axes to construct a logical topology skeleton;

[0027] The parsing module is used to parse the content of standard files in the standard file set and generate a dependency network between standards.

[0028] The positioning module is used to determine, within the two-dimensional logical space formed by the logical topology skeleton, the intersection core area of ​​two system observation points located at the business stage logical reference axis and the technical level logical reference axis, and to determine the upper left, upper right and lower edge logical areas of three system observation points located in the logical space far from the intersection core area.

[0029] The module is used to determine an evaluation reference plane in a three-dimensional logical attribute space formed by expanding a two-dimensional logical space, based on the coordinates of three system observation points and in combination with standard density attributes.

[0030] The extraction module maps the standard dependency network to the evaluation reference plane to form a standard coverage area and defines a logical polygon based on the boundary of the standard coverage area; it generates a logical ellipse based on the distribution of the five system observation points, detects and extracts the intersection and difference areas of the logical polygon and the logical ellipse on the evaluation reference plane, and generates standard distribution comparison data.

[0031] The model generation module is used to correct and integrate the standard dependency network mapped to the logical topology skeleton based on the standard distribution comparison data, and generate a standard system model of the entire life cycle of the target industry system.

[0032] In this embodiment of the invention, there is no need to rely on domain experts to manually collect, sort, and classify massive amounts of standard documents. Through automated collaboration of various modules, core processes such as generating a work scope directory and analyzing standard dependencies are efficiently advanced, shortening the construction and subsequent update cycle of the standard system. A topological skeleton is constructed using business stages and technical levels as fixed logical benchmarks. Combined with content parsing, spatial mapping, and other technologies, the correlation characteristics and distribution patterns of standards are quantified, avoiding subjective biases caused by differences in personal knowledge and experience in manual classification. This ensures that the system framework and standard mapping results built by different teams at different times remain consistent. Based on standard distribution comparison data, system deviations caused by industry technology iteration or standard updates can be quickly identified. Adjustments are efficiently completed through a network correction mechanism, enabling the standard system to reflect the latest status of industry standards in real time, solving the problems of high system maintenance costs and delayed response. Through the difference analysis of logical polygons and logical ellipses, standard coverage blind spots and redundant nodes are located. The generated standard system model can fully cover the entire life cycle of the target industry, while clarifying the dependencies between standards, providing clear and reliable guidance for standardization practices in all aspects of the industry.

[0033] In another preferred embodiment of the present invention, the process of obtaining the standard file set is as follows:

[0034] Access the standard bibliographic database, standard full-text database, and standard bulletin database to obtain published standard documents within the target industry system. Specifically, this involves: First, clarifying the core scope of the target industry system, including the business areas covered by the industry, such as the three-electric systems of new energy vehicles, production line links in intelligent manufacturing, key technology directions, and related product types, ensuring the relevance of the database search; Next, initiating the compliant access process to the standard bibliographic database, standard full-text database, and standard bulletin database. For the standard bibliographic database, access the system through the official database interface or compliant access channel. Set search conditions based on the determined industry scope, including core industry keywords (such as new energy vehicle power batteries, intelligent manufacturing assembly processes), the standard issuing organization, and the standard status (limited to published). After executing the search, obtain information including the standard number, standard name, issuing organization, etc. The database includes a list of bibliographic data for basic information, recording the retrieval time and source path for each bibliographic entry. For the full-text standard database, the database uses a list of standard numbers obtained from the bibliographic database as the retrieval basis. Each standard number is entered into the search box of the full-text database to trigger a full-text search, obtaining the full-text data of the corresponding standard document (common formats are PDF or editable text). A unique temporary storage identifier is assigned to each full-text document, recording its storage location and file size. For the standard announcement database, industry-related announcements within a set time range (which can be adjusted according to the industry development cycle, such as the past 15 years) are filtered. The standard number, publication date, and implementation date of the published standards are extracted from the announcement content and preliminarily matched with the information obtained from the bibliographic database and the full-text database. This confirms that the extracted standard numbers all belong to the target industry system and are in the published state, completing the initial collection of standard documents.

[0035] Extract the standard number, standard name, publication date, and main text content of each standard document. Specifically, this includes: for each collected standard document, initiating the information extraction process; parsing the cover title bar, the opening paragraph of the first page, or the introduction of the main text; identifying the standard number with a fixed format; ensuring complete extraction of all characters in the standard number, including the standard code, sequence number, and year code; and filling the extracted standard number into the standard number field of the preset standard information record table; extracting names that fully reflect the standard content from the title bar, table of contents, or the title description at the beginning of the main text, such as "Safety Requirements for Power Batteries for New Energy Vehicles" or "Maintenance Specifications for Intelligent Manufacturing Production Line Equipment," avoiding omission of key qualifiers; and filling the extracted standard name into the standard name field of the standard information record table; and extracting the publication instructions from the bottom of the cover of the standard document, etc. In the release information section at the end of the main text or in the corresponding standard announcement, identify the official release date of the standard and convert the date format to YYYY-MM-DD, such as converting August 10, 2023 to 2023-08-10, and fill it into the release date field of the standard information record table. For standard files in PDF format, use optical character recognition (OCR) technology or professional PDF text extraction tools to convert the full text into editable text. After conversion, perform preliminary text cleaning to delete garbled characters, blank page text, and duplicate headers and footers caused by format conversion, ensuring that the main text content is continuous and without redundancy. Store the cleaned main text content in the dedicated text unit of the corresponding standard file and record the index address of the text unit in the standard information record table. Repeat the above operations until the key information of all standard files is extracted.

[0036] The standard documents are deduplicated based on their standard numbers and integrated into a standard document set. Specifically, this involves: First, creating two empty containers: one as the final standard document set to store the deduplicated standard documents and their corresponding information; and the other as a list of standard numbers already included in the set to record these numbers and avoid duplication. Next, each standard document is processed sequentially according to the order of entries in the standard information record table. For the currently processed standard document, its standard number is first completely matched against all standard numbers already included in the standard number list. If the standard number does not exist in the list, the full text of the standard document and the corresponding entries (standard number, standard name, publication date, and text index address) in the standard information record table are directly added to the standard document set, and the standard number is also added to the list of standard numbers already included. If the standard number already exists in the list of standard numbers (determined to be a duplicate file), the publication date of the previously stored file with the same standard number is retrieved from the standard document set and compared with the currently processed document. The publication dates of the documents are compared numerically (the comparison logic is to convert the year, month, and day of the two dates to integers, first compare the "year," and the date with the larger year number is updated; if the number of years is the same, compare the "month," and the date with the larger month number is updated; if the number of months is the same, compare the "day," and the date with the larger day number is updated); the document with the updated publication date is retained, the older document is deleted, and the information of the corresponding entry in the standard information record table is updated, such as replacing it with the publication date and text index address of the updated document; if two duplicate documents have the same publication date, the text content of the two documents is further compared to check for version identifiers such as revision and update, and the document marked with the update version is retained; if there are no version identifiers, the document that was included in the set first is retained; the above comparison and filtering operations are repeated until all entries in the standard information record table are processed, and finally a set of standard documents containing no duplicate standard documents and complete information is generated, and a set list is output (the list contains the standard number, standard name, publication date, and storage path of each document), completing the integration.

[0037] This embodiment collects standard documents by connecting to a professional database and setting targeted search conditions, reducing the workload of manually screening irrelevant documents. It also covers three core data sources: bibliographic records, full texts, and announcements, avoiding incomplete standard collection due to human oversight and efficiently acquiring published standard documents within the target industry system. By uniformly extracting standard numbers, standard names, publication dates, and text content, and standardizing the information format and text content, it ensures the completeness and consistency of the extracted information, reducing subsequent processing errors caused by inconsistent information formats. Deduplication is performed using the standard number as a unique identifier, and the latest documents are selected by comparing the publication date values, effectively eliminating duplicate older standards.

[0038] In a preferred embodiment of the present invention, based on the knowledge of the target industrial system, a work scope directory covering the entire life cycle of the target industrial system is generated, and two dimensions, namely business stage and technology level, are extracted as the logical reference axes to construct a logical topology skeleton, including:

[0039] Extract the subject keywords describing the activities and technologies of the target industrial system from the standard names and text contents of the standard document set. Specifically, first, for each standard document in the formed standard document set, retrieve its standard name and the cleaned text content respectively to construct a standard text pool; then start the text preprocessing process, use a Chinese word segmentation algorithm (such as Jieba segmentation) to segment all the texts in the standard text pool, splitting the continuous text into independent word units; subsequently, load the preset general stop word list (including words without actual business / technical meanings such as "of", "and", "and", "this standard", "release", etc.), remove the stop words in the word segmentation results, and manually supplement the exclusive stop words of the target industrial system (such as industrial general names, meaningless modifiers) to further purify the word set; after the preprocessing is completed, calculate the word frequency of each remaining word, that is, the total number of times the word appears in all the texts of the standard text pool, and record the context of each word, such as which other words the word often co-occurs with and whether the sentence where the word is located describes industrial activities or technical content. Screen the subject keywords based on the dual dimensions of word frequency and context. First, set a word frequency threshold, which is dynamically adjusted according to the total text volume of the standard document set. For example, when the total text volume is 1 million words, the threshold is set to appear 10 times or more to initially screen out high-frequency words; then, combined with the context judgment, retain the words describing the core activities of the industry, such as power battery research and development, production line assembly, equipment operation and maintenance, or key technologies, such as energy density testing, control algorithm design, sensor calibration, and exclude the words that only describe non-core contents such as file formats and management requirements, and finally form a list of subject keywords.

[0040] Based on the co-occurrence relationships and context of keywords in the standard text, themes describing chronological order and process evolution are clustered into a phase sequence of the target industry system, forming a phase division of the work scope directory. Specifically, this includes: First, calculating the co-occurrence frequency of any two keywords in the keyword list, i.e., the total number of times the two keywords appear together in the same paragraph of the standard text, and constructing a keyword co-occurrence matrix (matrix rows and columns are keyword terms, and matrix elements are the co-occurrence frequencies of corresponding two keywords). Simultaneously, analyzing the context of each keyword, marking keywords with chronological attributes, such as early, middle, late, and later, or process evolution attributes, such as first, then, subsequently, and finally. For example, requirements analysis often co-occurs with solution design. (And the context shows that the requirements analysis precedes the solution design); Based on the keyword co-occurrence matrix and context labels, a hierarchical clustering algorithm is used to cluster the topic keywords. Keywords with a co-occurrence frequency higher than a set threshold (e.g., co-occurrence frequency ≥ 5 times) and whose context shows a temporal / process relationship are grouped into the same cluster. The keywords of each cluster are semantically summarized to extract the core meaning of the cluster. For example, the core meaning of a cluster containing requirements analysis, solution design, and prototype development is R&D-related activities. Then, according to the temporal / process relationship of each cluster, the clusters are arranged in chronological order to form a phase sequence, such as the R&D phase, testing phase, production phase, operation and maintenance phase, and recycling phase. This phase sequence is the preliminary phase division of the work scope directory.

[0041] Based on phase division, the main text of standard documents belonging to the same phase is aggregated to obtain aggregated text. Phrases describing specific tasks, outputs, and technical components are identified and extracted from the aggregated text as sub-nodes for the corresponding phase, forming a complete work scope directory. Specifically, this involves constructing the work scope directory in three steps based on a defined phase sequence. First, standard documents are assigned to phases to ensure each document accurately belongs to its corresponding phase. This involves identifying all keywords contained in each phase cluster (e.g., R&D phase clusters include keywords like requirements analysis, solution design, and prototype development; testing phase clusters include keywords like performance testing, reliability verification, and troubleshooting). Then, standard documents in the standard document set are retrieved one by one, and the main text of each document is analyzed. The document content is scanned, and the total number of times the document contains keywords for each stage cluster is counted. This count represents the document's relevance to that stage. The relevance values ​​of the document to all stages are then compared, and the document is assigned to the stage with the highest relevance value. If a document has similar relevance values ​​to multiple stages (e.g., 8 for the R&D stage and 7 for the testing stage, with a difference less than the set threshold of 2), further judgment is made based on the standard name and the core functional description in the document. For example, if the standard name is "Power Battery Testing Specification" and the document focuses on the testing process and indicators, it is still prioritized for assignment to the testing stage, even if it has some relevance to the R&D stage. If the standard name and the document do not explicitly point to a particular stage, the stage to which other standards cited in the document belong is referenced to help determine the final stage.

[0042] After all standard documents are assigned to specific phases, the aggregated texts for each phase are constructed. For each phase, all standard documents belonging to that phase are first selected, and their main text is pieced together segment by segment according to their release date. During the piecing process, duplicate headers, footers, legal citations, and other non-core text content are removed. Simultaneously, text breaks and garbled characters caused by format conversion are corrected to ensure semantic coherence in the pieced text. This results in a dedicated aggregated text for each phase, such as aggregated text for the R&D phase and aggregated text for the production phase. The aggregated text fully covers the core technologies and business content of all standard documents for the corresponding phase. Next, phrase units are extracted from the aggregated texts, forming a detailed node list for each phase. For each aggregated text, a dependency parsing algorithm is used. This algorithm analyzes the grammatical dependency relationships of words in the text, such as subject-predicate, verb-object, and modifier-head relationships, to identify phrase units with complete semantics. The extraction process focuses on two types of phrases: verb-object phrases consisting of a verb and a noun. The extraction process involves two main types of phrases. First, there are phrases that describe specific business tasks, such as power battery energy density testing, production line layout planning, and maintenance data recording. Second, there are phrases composed of nouns followed by nouns, which mainly describe business outputs or technical components, such as design documents, battery management system chips, and maintenance report templates. After extraction, all phrase units are deduplicated. Each extracted phrase unit is traversed, and each phrase is completely matched against the retained phrases. If a phrase is identical in wording to a retained phrase (e.g., "production line layout planning" versus "production line layout planning"), it is considered a duplicate phrase. The duplicates are deleted, and only the first occurrence of the phrase is retained, ensuring that the initially filtered phrase units do not contain completely duplicate content. Next, semantically similar phrases are merged by calculating the semantic similarity of the phrase units. This calculation uses word overlap and contextual relevance as core dimensions, employing a weighted average to arrive at the final semantic similarity. Specific steps are illustrated with examples of battery performance testing and power battery performance detection.

[0043] The first step is to calculate the word overlap. First, the two phrases to be compared are processed using Chinese word segmentation to adapt industry terminology. Priority should be given to identifying core industry terms to avoid splitting complete semantic units. For example, battery performance testing is split into two core semantic terms: battery performance and testing. Power battery performance detection is split into two core semantic terms: power battery performance and detection. Then, the core semantic intersection words and core semantic union words are determined. Performance is a core semantic term shared by both phrases. Battery performance and power battery performance have an inclusion relationship (power battery performance is a sub-segment of battery performance). Testing and detection are synonymous terms within the industry (both referring to performance verification operations). Therefore, there are 3 core semantic intersection words (performance, battery performance / power battery performance, testing / detection) and 4 core semantic union words (battery performance, power battery performance, testing, detection). Finally, the word overlap is calculated using the formula: (Number of core semantic intersection words ÷ Number of core semantic union words) × 100%.

[0044] The second step is to calculate the contextual relevance. First, determine the scope of context extraction. Centering on the two phrases to be compared (e.g., battery performance test and power battery performance testing), extract 50 characters before and after each phrase from the aggregated text as the context information for that phrase. For example, the context for battery performance test is: ... Conduct battery performance testing on power battery samples in the new energy vehicle field to verify whether the charge / discharge cycle count and range stability meet industry standards... The context for power battery performance testing is: ... For mass-produced power battery samples, through power battery performance testing, confirm the energy density, ... Cycle life, safety protection, and other indicators meet the standards...; Next, keywords related to the business scenario are extracted. Words or phrases strongly related to the target industry's business scenario are selected from the two contexts. These keywords must cover three categories of information: test object, core indicators, and verification objectives. The final extracted keywords include power battery samples, charge / discharge cycle count, range stability, industry standards, energy density, cycle life, safety protection, and compliance; subsequently, the intersection and union of scenario keywords are counted. The extracted keywords are first categorized by type. In the test object category, power battery samples are a keyword shared by both contexts; in the core indicator category, charge / discharge cycle count and range stability... Qualitative analysis, energy density, cycle life, and safety protection all point to power battery performance indicators, falling into the same semantic category and considered as an intersection. In the verification target category, industry standard compliance all point to whether performance indicators meet requirements, falling into the same semantic category and considered as an intersection. In summary, there are 3 types of intersections for scenario keywords (test object category, core indicator category, and verification target category), meaning the number of intersections is 3. The union of scenario keywords consists of all non-repeating keywords, totaling 8 (power battery sample, charge / discharge cycle count, range stability, industry standard, energy density, cycle life, safety protection, compliance), meaning the number of unions is 6. Then, the basic context association is calculated. Substituting the data into the formula of Basic Context Relevance = (Number of Intersections of Scene Keywords ÷ Number of Unions of Scene Keywords) × 100%, we can obtain Basic Context Relevance = (3 ÷ 6) × 100% = 50%. Finally, by adding the scene matching bonus, we further analyze the consistency of the business scenarios of the two contexts. Both revolve around the performance verification of power battery samples. The test objects, core purposes, and application scenarios are completely matched. Therefore, by adding a 40% scene matching bonus to the Basic Context Relevance, the final Context Relevance = 50% + 40% = 90%, which is completely consistent with the feature that both contexts point to battery performance verification.

[0045] The third step is to calculate the final semantic similarity using a weighted average formula. The formula sets two dimensions, word overlap and contextual relevance, each with a weight of 50% (this weighting ensures that both the semantic overlap of the phrases themselves and the relevance of the phrases in the business scenario are considered, avoiding bias in judgment based on a single dimension). Combining the previously calculated word overlap and final contextual relevance, the formula is substituted into the formula: Semantic Similarity = (Word Overlap × 50% + Contextual Relevance × 50%). The calculation result is compared with the preset semantic similarity threshold (75%). If the result is higher than 75%, the battery performance test and the power battery performance test are determined to be semantically similar phrases.

[0046] For phrases deemed semantically similar, a further comparison is made between the completeness of the description and the fit with industry terminology. In the power battery performance testing, the power battery clearly defines the sub-type of the test object, which is more specific than the description of battery performance testing. However, testing is the standardized terminology for performance verification operations in the target industry (such as the new energy vehicle industry), while detection is a general expression. Therefore, the specific description of power battery and the standardized terminology of testing are retained and merged into power battery performance testing. At the same time, the original battery performance testing and power battery performance detection are deleted. The above semantic similarity calculation and merging operation is repeated until all semantically similar phrases are merged. Finally, a list of subdivided nodes for each stage is formed. Each node in the list corresponds to a specific task, output, or technical component of the corresponding stage.

[0047] Finally, all the sub-nodes of all stages are integrated to form a complete work scope directory. The sub-nodes of each stage are arranged in the order of the determined stage sequence. At the same time, the directory hierarchy is supplemented, with the stage name in the stage sequence as the first-level directory, such as the R&D stage and the testing stage. The sub-nodes under each stage are arranged in the order of task logic, such as the preliminary preparation nodes, the execution operation nodes, and the result output nodes, as the second-level directory. Some complex nodes are further broken down into third-level sub-nodes (such as the requirements analysis node under the R&D stage, which is broken down into three third-level sub-nodes: user requirements research, requirements specification review, and requirements review). The final result is a work scope directory that covers the entire life cycle of the target industry system from the initial initiation to the final completion, with clear hierarchy and complete content.

[0048] Attribute analysis is performed on all sub-nodes in the work scope directory. Phrase units with temporal attributes in the node description are categorized to the business stage dimension, and phrase units with hierarchical or technical abstract attributes in the node description are categorized to the technical level dimension. Specifically, this includes: performing attribute analysis on each sub-node in the formed work scope directory, first breaking down the phrase units of each sub-node, identifying the attribute identifiers within them. If a phrase unit contains temporal-related identifiers, such as stage, step, link, early stage, late stage, or if the context shows that the node is directly related to the process progress of a certain stage, such as solution review in the R&D stage or first article inspection in the production stage, then the phrase unit is determined to have a temporal attribute. If a phrase unit contains hierarchical / technical abstraction-related identifiers, such as level, layer, module, system, component, or if the context shows that the node reflects a hierarchical relationship at the technical level, such as sensor selection at the component level or integration testing at the system level, then the phrase unit is determined to have a hierarchical or technical abstraction attribute.

[0049] Based on the attribute determination results, dimensional classification is performed. Phrase units with temporal attributes are uniformly classified into the business stage dimension to ensure that each classified phrase unit corresponds to a certain stage in the stage sequence. Phrase units with hierarchical or technical abstraction attributes are uniformly classified into the technical level dimension and initially sorted according to the degree of technical abstraction, such as component level < equipment level < system level < platform level. If a phrase unit has both attributes during the classification process, such as system-level fault troubleshooting in the operation and maintenance stage, its temporal attribute part (fault troubleshooting in the operation and maintenance stage) is extracted and classified into the business stage dimension, and its technical abstraction attribute part (system-level fault troubleshooting) is extracted and classified into the technical level dimension.

[0050] Using the business stage dimension as the horizontal axis and the technical level dimension as the vertical axis, an orthogonal two-dimensional coordinate system is established, which constitutes a logical topological skeleton. Specifically, this includes: First, determining the definition and arrangement order of the axes of the two-dimensional coordinate system. The business stage dimension is used as the horizontal axis (X-axis) of the coordinate system, and the stage nodes on the axis are arranged in chronological order of the stage sequence, such as from left to right: R&D stage, testing stage, production stage, operation and maintenance stage, and recycling stage. Each stage node occupies an equal logical interval on the horizontal axis. The technical level dimension is used as the vertical axis (Y-axis) of the coordinate system, and the hierarchical nodes on the axis are arranged from low to high level of technical abstraction, such as from bottom to top: component level, equipment level, system level, and platform level. Each hierarchical node occupies an equal logical interval on the vertical axis.

[0051] Next, an orthogonal relationship is established to ensure that the horizontal axis (business stage dimension) and the vertical axis (technology level dimension) are perpendicular to each other and have no intersection. That is, the advancement of business stages does not affect the level of abstraction of technology levels, and the division of technology levels does not depend on the chronological order of business stages. The two dimensions independently constitute a coordinate reference system. Finally, each sub-node is mapped to a unique coordinate position in the coordinate system according to its attribute affiliation in the business stage dimension and the technology level dimension (e.g., the R&D stage - system-level solution design corresponds to the intersection of the horizontal axis R&D stage and the vertical axis system level). This two-dimensional coordinate system with clear axis definitions and node positioning is the logical topological skeleton of the target industry system.

[0052] Compared to methods that rely on experts to manually define stages and nodes, this embodiment automates keyword extraction, clustering, node extraction, and dimension classification through algorithms. This reduces manual intervention, avoids the tedious work of experts analyzing standard texts one by one, and shortens the construction cycle of the work scope directory. Stage division is based on keyword co-occurrence relationships and contextual temporal features, subdivided node extraction relies on the actual content of the standard text, and dimension classification is based on clear attribute identifiers and context. The entire process is data-driven, avoiding problems such as stage division confusion and inconsistent node definitions caused by differences in the knowledge background of different experts, ensuring the objectivity and consistency of the work scope directory and logical topology framework. By covering the entire lifecycle of the target industry system from early-stage R&D to late-stage recycling through stage sequences, and forming an orthogonal logical topology framework by combining technical level dimensions, the positioning of each subdivided node is more accurate, avoiding standard mapping deviations caused by logical ambiguity in the framework. Keyword extraction, clustering algorithms, and dimension classification rules can be flexibly adjusted according to the standard text characteristics of different industry systems, adapting to different target industries without reconstructing the overall process, improving the versatility and scalability of the solution.

[0053] In a preferred embodiment of the present invention, content parsing is performed on the standard files in the standard file set to generate a dependency network between standards, including:

[0054] Based on the business phase and technical level dimensions of the logical topology skeleton, the standard documents in the standard document set are annotated to determine their logical coordinates in a two-dimensional logical space. Specifically, this involves: first, defining the axes of the two-dimensional logical space: the horizontal axis represents the business phase dimension (arranged in the order of R&D, testing, production, operation and maintenance, and recycling, with each phase corresponding to a unique horizontal axis coordinate value, e.g., R&D phase 1, testing phase 2, production phase 3, operation and maintenance phase 4, recycling phase 5); the vertical axis represents the technical level dimension (arranged from low to high abstraction levels of component level, equipment level, system level, and platform level, with each level corresponding to a unique vertical axis coordinate value, e.g., component level 1, equipment level 2, system level 3, platform level 4); then, processing each standard document in the standard document set individually, performing content annotation, and extracting the core content of the standard documents, including the standard name, etc. The technical terms and scope of application of the document are used to determine the business stage corresponding to the core content. For example, the scope of application of the technical specification for power battery R&D is clearly the R&D stage, so it matches the R&D stage on the horizontal axis, corresponding to the horizontal axis coordinate value 1; the technical terms of the production line equipment operation and maintenance guide all revolve around equipment maintenance operations, matching the operation and maintenance stage on the horizontal axis, corresponding to the horizontal axis coordinate value 4; the technical abstraction level of the core content of the standard document is analyzed to determine the corresponding technical level. For example, the technical requirements for battery management system chips focus on a single component, matching the component level on the vertical axis, corresponding to the vertical axis coordinate value 1; the integration standard for the whole vehicle control system of new energy vehicles covers the collaboration of multiple systems, matching the system level on the vertical axis, corresponding to the vertical axis coordinate value 3; the horizontal and vertical axis coordinate values ​​are integrated to form a unique logical coordinate of the standard document in a two-dimensional logical space. For example, the coordinate of the technical specification for power battery R&D is (1, 3), and the coordinate of the production line equipment operation and maintenance guide is (4, 2), and the coordinate information is bound and stored with the standard document.

[0055] This process involves analyzing the normative reference section of each standard document in the standard document set, extracting the standard numbers of the referenced standards listed in that section, and establishing direct reference relationships from the current standard document to the referenced standard documents. Specifically, this includes: retrieving each standard document in the set, locating the normative reference section in the document's main text (located at the beginning of the text, with the title explicitly indicating a normative reference; some standards are marked as referenced and their normative reference attributes are clearly stated in the section description), extracting the standard numbers of all referenced standards listed in that section, and removing standard numbers marked as informative references (only retaining those explicitly marked as normative references or those without). (Note: This is actually a mandatory reference standard number). The extracted referenced standard number is completely matched with the standard numbers in the standard file set to confirm whether the referenced standard is in the set. If the match is successful, a directed association is established from the current standard file to the referenced standard file. This association is the direct reference relationship. If the match fails (the referenced standard is not in the set), the standard number and the reference relationship are recorded, but not included in the subsequent network construction (to avoid invalid associations). The above operation is repeated until the normative reference chapters of all standard files are parsed, forming a preliminary list of direct reference relationships. The list includes the current standard number, the referenced standard number, and the relationship type: direct reference.

[0056] This analysis examines the technical content of the main clauses in the standard document set, calculates the semantic similarity of different standard documents in the set across terminology definitions, technical parameters, and functional requirements, and identifies standard documents with semantic similarity exceeding a preset threshold as having an indirect support relationship. Specifically, for any two standard documents (Document A and Document B) without a direct reference relationship, the analysis covers the technical content of the main clauses from three dimensions: terminology definitions, technical parameters, and functional requirements. Semantic similarity is calculated. Specifically, for the terminology definition dimension, all terms in the terminology and definition sections of Document A and Document B are extracted, such as energy density, cycle life, and charge / discharge efficiency, forming two terminology sets (Set A1 and Set B1). The number of intersections (the number of terms appearing simultaneously in both A1 and B1) and the number of unions (the total number of all unique terms in A1 and B1) of the two sets are counted. The terminology definition similarity is calculated using the formula: Terminology Definition Similarity = (Number of Intersections ÷ Number of Unions) The similarity is calculated as follows: (Number of union sets) × 100%; When calculating the similarity of technical parameters, extract the parameter ranges describing the same technical indicators in the two documents (e.g., power battery energy density ≥ 150Wh / kg, power battery energy density ≥ 140Wh / kg), determine the parameter benchmark value (take the common industry benchmark, such as the energy density benchmark value of 150Wh / kg); calculate the ratio of the absolute value of the parameter difference to the benchmark value, and calculate it according to the formula: Technical Parameter Similarity = (1 - |Parameter A - Parameter B| ÷ Benchmark Value) × 100%; When calculating the similarity of functional requirements, use a semantic analysis algorithm (e.g., cosine similarity based on word vectors) to convert the functional requirement clauses in the two documents, such as ensuring that the battery works normally in an environment of -20℃ to 60℃, and ensuring that the battery operates stably in an environment of -25℃ to 55℃, into semantic vectors; calculate the cosine value of the two semantic vectors, which is the functional requirement similarity. For example, if the cosine value is 0.85, then the functional requirement similarity = 85%.

[0057] Total semantic similarity = (Terminology definition similarity × 30% + Technical parameter similarity × 40% + Functional requirement similarity × 30%), where 30% and 40% represent the weight percentage of the corresponding dimension in the total semantic similarity calculation. The higher the weight percentage, the greater the impact of the calculation result of that dimension on the total semantic similarity. A preset threshold is set (determined according to the technical characteristics of the target industry, set to 70%). If the total semantic similarity of two documents exceeds the threshold, it is identified as an indirect support relationship. A directed association is established from the document with more basic technical content to the document with more applied technical content, such as basic parameter standards pointing to application specifications, forming a list of indirect support relationships.

[0058] Each direct reference relationship is assigned a first weight value based on the explicitness of the reference, and each indirect support relationship is assigned a second weight value based on semantic similarity. Specifically, the first weight value (for direct reference relationships) is assigned based on the explicitness of the reference. If the current standard document explicitly marks the referenced standard as mandatory (e.g., the chapter states that it should comply with standard XX), the first weight value is set to 1.0. If it is marked as a recommended reference (e.g., it states that standard XX should be consulted), the first weight value is set to 0.7. If it is not explicitly marked but is a mandatory reference by default in the industry, the first weight value is set to 0.9. For example, the power battery safety standard mandates the reference to the battery material flame retardant standard, so the first weight value for the direct reference relationship between the two is 1.0; the power battery packaging specification recommends the reference to the packaging material environmental protection standard, so the first weight value is 0.7.

[0059] The second weight value (indirect support relationship) is directly related to the total semantic similarity value, calculated using the formula: Second Weight Value = Total Semantic Similarity ÷ 100 (ensuring a positive correlation between the weight value and the degree of semantic similarity). If the total semantic similarity between the two standard documents is 79.9%, the second weight value is calculated as 79.9% ÷ 100 = 0.799, meaning that the strength of the indirect support relationship between the two documents is highly matched with their semantic similarity. If the total semantic similarity between the two standard documents is 85%, the second weight value is 85% ÷ 100 = 0.85. The essence of this calculation is to convert the percentage value to a decimal (85 as the percentage value divided by 100), ensuring that the weight value is within the range of 0 to 1. The higher the semantic similarity, the larger the second weight value, and the stronger the indirect support relationship.

[0060] Using each standard document in the standard document set as a node, and direct reference relationships with a first weight value and indirect support relationships with a second weight value as directed edges, a dependency network between standards is constructed. Each node in the dependency network contains the logical coordinate attributes of the standard document corresponding to that node. Specifically, this includes: using each standard document in the standard document set as a node, each node stores the basic information (standard number, standard name) and determined logical coordinate attributes of the corresponding standard document; using direct reference relationships and indirect support relationships as directed edges, each directed edge is labeled with a corresponding weight value (direct reference edges are labeled with the first weight value, and indirect support edges are labeled with the second weight value); and using a graph structure (such as an extended directed graph of an undirected graph) to build the network framework. First, import all standard file nodes into the framework; then, according to the list of direct reference relationships and the list of indirect support relationships, add directed edges to each node one by one and bind the corresponding weight values. For example, the node Battery Material Flame Retardant Standard (coordinate (1,1)) points to the node Power Battery Safety Standard (coordinate (2,3)) through a directed edge with a weight of 1.0, and the node Basic Battery Parameter Standard (coordinate (1,2)) points to the node Power Battery Application Specification (coordinate (3,3)) through a directed edge with a weight of 0.799. After the construction is completed, check the integrity of the network to ensure that all standard files have been included as nodes and all identified direct reference relationships and indirect support relationships have been added as directed edges, and finally form a dependency network between standards containing node attributes, directed edges and weights.

[0061] This embodiment covers both direct references (explicit relationships) and indirect support (implicit relationships), and quantifies indirect associations through multi-dimensional semantic similarity calculation, avoiding incomplete relationship characterization due to human omission of implicit technical associations. The allocation of weight values ​​further distinguishes the strength of relationships, making the dependency logic between standards more accurate. From the determination of logical coordinates and semantic similarity calculation to weight allocation, all processes are automated based on clear rules and algorithms, without relying on expert subjective judgment, avoiding inconsistencies in manual classification, and ensuring the stability and reliability of the dependency network construction results. Automated content annotation, relationship identification, and network construction reduce the workload of manually analyzing standard documents one by one, and shorten the dependency network construction cycle. When standards are updated, it is only necessary to recalculate the semantic similarity between the updated file and other files, adjust the association relationships and weights, without reconstructing the entire network, adapting to the needs of dynamic iteration of industry standards.

[0062] In a preferred embodiment of the present invention, within the two-dimensional logical space constituted by the logical topology skeleton, two system observation points are determined to be located in the core area where the business stage logical reference axis and the technical level logical reference axis intersect, and three system observation points are determined to be located in the upper left, upper right, and lower edge logical regions of the logical space, respectively, away from the core area where the intersection occurs, including:

[0063] The origin of the two-dimensional logical space is defined as the core area where the business stage logical reference axis and the technical level logical reference axis intersect. The coordinates of the observation points of the two systems are set as the coordinates of the origin. Specifically, this includes: first, connecting the constructed two-dimensional logical space (horizontal axis for business stage, vertical axis for technical level), clarifying the coordinate rules of this space. The coordinate values ​​of the business stage dimension (horizontal axis) increase sequentially according to the business process progression, specifically in the order of R&D, testing, production, operation and maintenance, and recycling. Each stage corresponds to a unique coordinate value, for example, R&D stage is 1, testing stage is 2, production stage is 3, operation and maintenance stage is 4, and recycling stage is 5. The coordinate values ​​of the technical level dimension (vertical axis) increase sequentially from low to high technical abstraction, specifically at the component level. The system is divided into four levels: component level, equipment level, system level, and platform level. Each level corresponds to a unique coordinate value, for example, component level is 1, equipment level is 2, system level is 3, and platform level is 4. Based on this, the intersection of the starting reference point of the business stage dimension and the starting reference point of the technology level dimension in the two-dimensional logical space is defined as the origin of the coordinate system, with the coordinate value set to (0, 0). This origin represents the core area where the most basic link of the business process and the most basic level of technical abstraction in the target industry system meet, and is the basic anchor point of the entire system logic. Subsequently, two system observation points are set. Since the core area (origin) is the basic support link of the system, it is necessary to focus on its standard coverage. Therefore, the coordinates of the two system observation points are both set to the coordinates of the origin (0, 0), and the two observation points together serve as the position reference of the core area.

[0064] To obtain the logical coordinates of all nodes in the inter-standard dependency network, determine the minimum and maximum logical coordinate values ​​for the business stage dimension and the technical level dimension. Specifically, this involves: first, extracting the logical coordinates bound to all nodes (i.e., all standard files) from the constructed inter-standard dependency network; that is, traversing each node in the network and reading its stored business stage dimension coordinate values ​​(horizontal axis values) and technical level dimension coordinate values ​​(vertical axis values) one by one, forming two sets of values: the business stage coordinate set (denoted as the X set) and the technical level coordinate set (denoted as the Y set); then calculating the extreme values ​​of the two sets respectively. For the X set (e.g., containing values ​​1, 2, 2, 3, 4, 5, etc.), by calculating the extreme values ​​one by one... Compare all values ​​and select the smallest value as the minimum logical coordinate value of the business stage dimension (denoted as Xmin), and select the largest value as the maximum logical coordinate value of the business stage dimension (denoted as Xmax). For example, if the X set is {1, 2, 2, 3, 4, 5}, then Xmin=1 and Xmax=5. For the Y set (such as containing values ​​1, 1, 2, 3, 4, 4, etc.), use the same comparison method to select the minimum logical coordinate value of the technical level dimension (denoted as Ymin) and the maximum logical coordinate value of the technical level dimension (denoted as Ymax). For example, if the Y set is {1, 1, 2, 3, 4, 4}, then Ymin=1 and Ymax=4. Record and archive the calculated Xmin, Xmax, Ymin, and Ymax.

[0065] The region in the two-dimensional logical space where the business stage dimension has the minimum logical coordinate value and the technical level dimension has the maximum logical coordinate value is defined as the upper left edge logical region; the region where the business stage dimension has the maximum logical coordinate value and the technical level dimension has the maximum logical coordinate value is defined as the upper right edge logical region; and the region where the business stage dimension has the median logical coordinate value and the technical level dimension has the minimum logical coordinate value is defined as the lower edge logical region. Specifically, based on determined coordinate extreme values ​​and the meaning of the axes in the two-dimensional logical space, three edge logical regions are defined, each corresponding to a key link at a different location in the system. The upper left edge logical region represents the intersection of the initial stage of the business process and the highest level of technical abstraction. Its definition rule is that the business stage dimension takes Xmin (minimum logical coordinate value), and the technical level dimension takes Ymax (maximum logical coordinate value). The scope is defined as follows: business stage dimension X∈[Xmin, Xmin+1], technical level dimension Y∈[Ymax-1, Ymax]; for example, when Xmin=1 and Ymax=4, the coordinate range of this area is X∈[1,2] and Y∈[3,4], corresponding to the platform-level technical link in the R&D stage; the upper right edge logical area represents the intersection of the business process closing stage and the highest level of technical abstraction. The definition rule is that the business stage dimension takes Xmax (maximum logical coordinate value), and the technical level dimension takes Ymax (maximum logical coordinate value). The coordinate range is business stage dimension X∈[Xmax-1, Xmax], and technical level dimension Y∈[Ymax-1, Ymax]; for example, when Xmax=5 and Ymax=4, the coordinate range is X∈[4,5] and Y∈[3,4], corresponding to the platform-level technical link in the recycling stage.

[0066] The lower edge logical area represents the intersection of the intermediate stage of the business process and the most basic level of technical abstraction. The definition rule is that the business stage dimension takes the median of the logical coordinates, and the technical level dimension takes Ymin (the minimum logical coordinate value). The median of the business stage dimension is calculated as follows: if the number of values ​​in the X set is odd, the median is the value of the middle position after sorting; if it is even, the median is the average of the two middle values ​​after sorting (e.g., the median of the X set {1, 2, 3, 4, 5} is 3, and the median of the X set {1, 2, 3, 4} is (2+3) / 2=2.5). The coordinate range of this area is: business stage dimension X ∈ [median - 0.5, median + 0.5], technical level dimension Y ∈ [Ymin, Ymin+1]. For example, when median = 3 and Ymin = 1, the coordinate range is X ∈ [2.5, 3.5] and Y ∈ [1, 2], corresponding to the component-level technical link in the production stage.

[0067] Within the coordinate ranges corresponding to the upper left, upper right, and lower edge logical regions, the geometric center point coordinates of the upper left edge logical region are selected as the first system observation point coordinates; the geometric center point coordinates of the upper right edge logical region are selected as the second system observation point coordinates; and the geometric center point coordinates of the lower edge logical region are selected as the third system observation point coordinates. Specifically, to ensure that the observation points accurately represent the positional characteristics of each edge region, the geometric center point of each edge region is selected as the system observation point. The calculation method for the geometric center point coordinates is based on the region coordinate range. The specific steps for calculating the average values ​​of the horizontal and vertical axes of the observation point are as follows: For the observation point in the upper left edge region (the first observation point), based on the coordinate range of this region X∈[Xmin, Xmin+1], Y∈[Ymax-1, Ymax], calculate the horizontal axis coordinate of the center point as (Xmin+(Xmin+1))÷2, and the vertical axis coordinate as ((Ymax-1)+Ymax)÷2. For example, when Xmin=1 and Ymax=4, the horizontal axis coordinate is (1+2)÷2=1.5, and the vertical axis coordinate is (3+4)÷2=3.5. Therefore, the coordinates of the first observation point are (1.5, 3.5). The observation point in the upper right edge region (the second observation point) is calculated based on the coordinate range of this region: X∈[Xmax-1,Xmax], Y∈[Ymax-1,Ymax]. The horizontal axis coordinate is calculated as ((Xmax-1)+Xmax)÷2, and the vertical axis coordinate is calculated as ((Ymax-1)+Ymax)÷2. For example, when Xmax=5 and Ymax=4, the horizontal axis coordinate is (4+5)÷2=4.5, and the vertical axis coordinate is 3.5. Therefore, the coordinates of the second observation point are (4.5, 3.5). The observation point in the lower edge region (the third observation point) is calculated based on the coordinate range of this region: X∈[Xmax-1,Xmax], Y∈[Ymax-1,Ymax]. The coordinate range of the region is X∈[median - 0.5, median + 0.5], Y∈[Ymin, Ymin + 1]. Calculate the x-axis coordinate of the center point = ((median - 0.5) + (median + 0.5)) ÷ 2 = median, and the y-axis coordinate = (Ymin + (Ymin + 1)) ÷ 2. For example, when median = 3 and Ymin = 1, the x-axis coordinate = 3, and the y-axis coordinate = (1 + 2) ÷ 2 = 1.5. Therefore, the coordinates of the third observation point are (3, 1.5). Summarize the coordinates of the three edge region observation points with the coordinates of the two core region observation points to form a complete list of system observation point coordinates.

[0068] This embodiment transforms observation point positioning into quantifiable logical operations through explicit coordinate rules, extreme value calculations, and geometric center point algorithms. This avoids positioning deviations caused by subjective expert judgments, ensuring consistent positioning logic for observation points built across different industry systems or at different times, thus improving the reliability of the evaluation benchmark. The observation points simultaneously cover the core area where business and technology intersect, as well as the edge areas at both ends of business and the high and low ends of technology, comprehensively covering the key logical positions throughout the entire lifecycle of the target industry system. This more comprehensively reflects the distribution characteristics of the standard within the system, avoiding evaluation omissions due to one-sided reference points. The coordinates of the five observation points clearly define the key positions in the two-dimensional logical space. When subsequently expanding the three-dimensional logical attribute space, the constraint points of the evaluation benchmark plane can be quickly determined by directly superimposing standard density attributes on these coordinates, without the need for repositioning the reference, ensuring the consistency of coordinate association between the three-dimensional space and the two-dimensional logical space. The coordinate calculation process of the observation points is based on the node coordinate data of the dependency network. The position of each observation point can be traced back to the logical coordinates of the original standard document, allowing for rapid verification through the original data and improving the traceability and credibility of the entire construction process.

[0069] In a preferred embodiment of the present invention, based on the coordinates of three system observation points and combined with standard density attributes, an evaluation reference plane is determined in a three-dimensional logical attribute space formed by expanding a two-dimensional logical space, including:

[0070] For system observation points located in the upper left, upper right, and lower right logical regions, the number of inter-standard dependency network nodes within a circular logical region centered on the observation point and with a preset logical distance as its radius is counted. This number is used as the standard density attribute value for the observation point. Specifically, the rules for determining the preset logical distance are clarified. Combining the coordinate intervals in the two-dimensional logical space (coordinate values ​​for both business stage and technical level dimensions increase by 1 as a base interval, e.g., 1 interval from R&D stage 1 to testing stage 2, 1 interval from component level 1 to equipment level 2), the preset logical distance is set to 0.5. This value covers the most relevant standard nodes around the observation point while avoiding an excessively large range. Irrelevant nodes are mixed in; then, for the three edge system observation points (top left edge, top right edge, and bottom edge), standard node statistics are performed respectively, that is, the two-dimensional logical coordinates of the observation point are retrieved, such as (1.5, 3.5), and a circular logical region is defined in the two-dimensional logical space with the coordinates as the center and 0.5 as the radius; all nodes in the constructed standard inter-dependency network are traversed, and the two-dimensional logical coordinates of each node are read one by one. For example, if the coordinates of a node are (2, 3.5) and the coordinates of another node are (1.5, 4), the distance between the node and the observation point is calculated using the distance formula between the two points. If the distance is less than or equal to the preset logical distance of 0.5, the node is determined to be within the circular region; the total number of all nodes within the circular region is counted. This total number is the standard density attribute value of the top left edge observation point. For example, if the statistical result is 8, the standard density attribute value is 8.

[0071] Using the same method as the top-left edge observation point, a circular region is defined with the two-dimensional logical coordinates of the top-right edge observation point, such as (4.5, 3.5) as the center and 0.5 as the radius. The distance between each node in the network and this observation point is calculated, and the total number of nodes in the region is counted. This count is used as the standard density attribute value of the top-right edge observation point. For example, if the count result is 6, then the standard density attribute value is 6. Similarly, a circular region is defined with the two-dimensional logical coordinates of the bottom edge observation point, such as (3, 1.5) as the center and 0.5 as the radius. The total number of nodes in the region is counted and used as the standard density attribute value of the bottom edge observation point. For example, if the count result is 10, then the standard density attribute value is 10.

[0072] The first three-dimensional coordinate point is formed by combining the two-dimensional logical coordinates of the system observation point located in the upper left logical region with the standard density attribute value of the system observation point; the second three-dimensional coordinate point is formed by combining the two-dimensional logical coordinates of the system observation point located in the upper right logical region with the standard density attribute value of the system observation point; and the third three-dimensional coordinate point is formed by combining the two-dimensional logical coordinates of the system observation point located in the lower logical region with the standard density attribute value of the system observation point. Specifically, this includes: firstly, clarifying the axis definition of the three-dimensional logical attribute space; and secondly, adding a standard density dimension as the vertical axis Z to the existing two-dimensional logical space (horizontal axis X represents the business stage dimension, and vertical axis Y represents the technical level dimension), forming X (business stage) - Y (technical level) - Z (standard density). The three-dimensional logical attribute space of the three edge observation points is then formed by combining the two-dimensional coordinates of the three edge observation points with the corresponding standard density attribute values. The first three-dimensional coordinate point corresponds to the upper left edge observation point. The two-dimensional logical coordinates (e.g., 1.5, 3.5) and the standard density attribute value (e.g., 8) are combined to obtain the first three-dimensional coordinate point (1.5, 3.5, 8). The second three-dimensional coordinate point corresponds to the upper right edge observation point. The two-dimensional logical coordinates (e.g., 4.5, 3.5) and the standard density attribute value (e.g., 6) are combined to obtain the second three-dimensional coordinate point (4.5, 3.5, 6). The third three-dimensional coordinate point corresponds to the lower edge observation point. The two-dimensional logical coordinates (e.g., 3, 1.5) and the standard density attribute value (e.g., 10) are combined to obtain the third three-dimensional coordinate point (3, 1.5, 10).

[0073] Using the first, second, and third 3D coordinate points as three non-collinear constraint points in space, the process includes: determining whether the three 3D coordinate points are collinear. The criterion for collinearity is whether the vectors formed by any two points are proportional. Taking the first 3D coordinate point (1.5, 3.5, 8), the second 3D coordinate point (4.5, 3.5, 6), and the third 3D coordinate point (3, 1.5, 10) as an example, the vector from the first point to the second point is calculated as (4.5 - 1.5, 3.5 - 3.5, 6 - 8) = (3, 0, -2); the vector from the first point to the third point is calculated as (3 - 1.5, 1.5 - 3.5, 10 - 8) = (1.5, -2, 2); the vector ratio is determined as 3 ÷ 1.5 = 2, 0 ÷ (-2) = 0, and -2 ÷ 2 = -1. Since the three ratio values ​​are not equal, it indicates that the two vectors are not proportional. Therefore, the three 3D coordinate points are not collinear and can be used as non-collinear constraint points in the spatial plane.

[0074] Using the geometric principle of determining a spatial plane through three points, the equation of the spatial plane passing through the first, second, and third three-dimensional coordinate points is calculated. Specifically, the general equation of the spatial plane is Ax + By + Cz + D = 0 (where A, B, and C are not simultaneously 0). During calculation, the coordinates of the three non-collinear constraint points need to be substituted into this equation to form a system of equations. The proportional relationship between A, B, C, and D is then solved. The specific process is as follows: First, substitute the three three-dimensional coordinate points into the general equation of the spatial plane to obtain three sets of equations. Specifically, substituting the first three-dimensional coordinate point (1.5, 3.5, 8) into the equation yields 1.5A + 3.5B + ​​8C + D = 0, which is denoted as Equation 1; substituting the second three-dimensional coordinate point (4.5, 3.5, 6) into the equation yields 4.5A + 3.5B + ​​6C + D = 0, which is denoted as Equation 2; substituting the third three-dimensional coordinate point (3, 1... ... Substituting 5 and 10 into the equation, we get 3A + 1.5B + ​​10C + D = 0. This equation is denoted as Equation 3. By eliminating redundant variables through addition and subtraction operations between equations, we gradually derive the relationship between each coefficient and C. That is, subtracting Equation 1 from Equation 2 eliminates B and D, and after simplification, we get 3A - 2C = 0. Further transformation yields A = (2 / 3)C. This equation is denoted as Derivation 1. Subtracting Equation 1 from Equation 3 eliminates D, and after simplification, we get 1.5A - 2C = 0. B+2C=0, this equation is denoted as Equation 4; Substituting Derivation 1 (A=(2 / 3)C) into Equation 4, we get 1C-2B+2C=0, and further rearranging, we get 3C-2B=0, and further transformation, we get B=(3 / 2)C, this equation is denoted as Derivation 2; Substituting Derivation 1 (A=(2 / 3)C) and Derivation 2 (B=(3 / 2)C) into Equation 1 together, we get D=-14.25C, this equation is denoted as Derivation 3.

[0075] To avoid decimal coefficients and simplify subsequent calculations, we choose a suitable value for C and substitute it into derivations 1, 2, and 3 to determine the integer coefficients. We take C = 12 (this value ensures that A, B, and D are all integers and that the coefficients have no common divisor). We calculate A according to derivation 1: A = (2 / 3) × 12 = 8; we calculate B according to derivation 2: B = (3 / 2) × 12 = 18; we calculate D according to derivation 3: D = -14.25 × 12 = -171. Substituting A = 8, B = 18, C = 12, and D = -171 into the general equation of the spatial plane, we get 8x + 18y + 12z - 171 = 0. If further simplification is needed, we can divide both sides of the equation by the common divisor 2 to get 4x + 9y + 6z - 85.5 = 0. The spatial plane positions corresponding to the two forms of the equation are completely consistent.

[0076] The plane defined by the spatial plane equation is determined as the evaluation reference plane for mapping and evaluating the standard distribution. Specifically, the spatial plane corresponding to the solved spatial plane equation, such as 4x+9y+6z-85.5=0 or 8x+18y+12z-171=0, is formally defined as the evaluation reference plane. The core function of this plane is that when mapping the dependency network between standards to this plane, the distribution of standard nodes on the plane can be quantitatively analyzed based on the fixed spatial position of the plane to determine whether the standard distribution is balanced and whether there are gaps in the standard coverage of local areas. Therefore, the final determined plane equation and the position information of the plane in the three-dimensional logical attribute space need to be stored and archived.

[0077] This embodiment transforms the density of standard nodes in a two-dimensional logical space into quantifiable values ​​by calculating standard density attribute values. This avoids the ambiguity of subjective descriptions such as density and sparsity in manual assessments, making the standard distribution characteristics more accurate and comparable. An evaluation benchmark plane is constructed using three-dimensional coordinate points and spatial plane equations. The entire process is based on the actual distribution data and geometric principles of standard nodes, eliminating the need for expert judgment and completely avoiding deviations caused by manually setting benchmarks. This ensures the consistency and objectivity of evaluation benchmarks constructed in different industry systems or at different times. As a reference surface for the standard distribution, the evaluation benchmark plane can identify redundant and gap areas in the standard distribution when mapping the standard dependency network. This addresses the pain point of difficulty in locating subtle coverage issues in manual assessments. From standard density calculation to solving the plane equations, each step has clear algorithmic rules and data basis.

[0078] In a preferred embodiment of the present invention, a standard dependency network is mapped onto an evaluation reference plane to form a standard coverage area, and a logical polygon is defined based on the boundary of the standard coverage area; a logical ellipse is generated based on the distribution of the five system observation points; the intersection and difference regions of the logical polygon and the logical ellipse on the evaluation reference plane are detected and extracted to generate standard distribution comparison data, including:

[0079] The logical coordinates of each node in the inter-standard dependency network are vertically projected onto the evaluation reference plane to obtain the projection point of each node on the evaluation reference plane. All projection points constitute the standard coverage area. Specifically, this involves: First, clarifying the core logic of projection. Vertical projection refers to drawing a straight line (i.e., a perpendicular line) parallel to the normal vector of the evaluation reference plane for each node in the inter-standard dependency network. The intersection of this line and the evaluation reference plane is the projection point of the node. The specific process is as follows: Retrieve the information of a single node from the inter-standard dependency network. This node contains two-dimensional logical coordinates (business stage dimension X1, technical level dimension Y1) and standard density attribute value Z1, which are combined to form the three-dimensional coordinates (X1, Y1, Z1) of the node; At the same time, retrieve the determined evaluation reference plane equation, such as 4x+9y+6z-85.5=0, and clarify the coefficients of the plane equation A=4, B=9, C=6 (coefficients A, B, and C constitute the normal vector of the plane, i.e., the perpendicular direction). Since the perpendicular line is parallel to the normal vector, its direction vector is (A, B, C). Therefore, the parametric equations of the perpendicular line passing through the node (X1, Y1, Z1) are: x = X1 + A × t (t is a parameter representing the distance moved along the perpendicular direction), y = Y1 + B × t, z = Z1 + C × t. Substituting the parametric equations of the perpendicular line into the equation of the evaluation reference plane makes the equations true (because the projection point is on the plane), we get A × (X1 + A × t) + B × (Y1 + B × t) + C × ( Z1+C×t)+D=0 (D is the constant term of the plane equation, here D=-85.5), solve for t based on this; substitute the obtained t into the vertical parametric equation to obtain the projection point (x2, y2, z2), which is the projection point of the node on the evaluation reference plane; repeat the above steps to perform vertical projection on all nodes in the inter-standard dependency network to obtain the projection points of all nodes; treat all projection points as a whole and define it as the standard coverage area.

[0080] Calculate the convex hull boundary of all projected points in the standard coverage area. The convex hull boundary is defined as the boundary contour of a logical polygon. Specifically, the convex hull is the smallest convex polygon containing all points in the plane, and its boundary reflects the outer contour of the standard coverage area. The calculation process uses the Graham scan method, and the specific steps are as follows: From all the obtained projected points, remove duplicate projected points (points with completely identical coordinates); Among the remaining projected points, find the lowest and leftmost point as the base point, i.e., first compare the y-coordinates of the projected points, and the point with the smallest y-coordinate is the candidate; If multiple points have the same y-coordinate, select the point with the smallest x-coordinate, such as projected points (1.2, 2.1), (1.5, 2.1), and (2.3, 3.5), with the base point being (1.2, 2.1); Using the base point as the origin, calculate the polar angle of the line connecting each other projected point to the base point (the polar angle is the angle between the line and the positive x-axis, expressed as arctan²(y / x). i -y0,x iCalculate (x0, y0), where (x0, y0) are the coordinates of the base point, (x0, y0) i y i (The coordinates of other projection points are given); the projection points are sorted in ascending order of polar angle. If two points have the same polar angle, the point that is farther away from the base point is retained.

[0081] Initialize a stack and push the first three sorted projection points onto the stack (elements in the stack represent vertices of the current convex hull). Starting from the fourth projection point, traverse the remaining projection points in sequence. For the current point P, take the top two points of the stack (denoted as S1 and S2, where S1 is the top of the stack and S2 is the next point after the top of the stack), and calculate the vector S2S1(x1-x2, y1-y2) and the vector S1P(x1-x2, y1-y2). -x1, The cross product of (x1 - x2) is (cross product = (x1 - x2) × (y1) -y1) - (y1 - y2) × ( -x1)); If the cross product is less than 0, it means that point S1 is a concave point (breaking the convexity of the convex hull), so pop S1 from the stack; repeat this process until the cross product of the two points at the top of the stack and the current point P is greater than or equal to 0, then push P onto the stack; after traversal, the remaining points in the stack are the vertices of the convex hull; connect the vertices in the stack in order (starting from the base point, clockwise or counterclockwise along the boundary of the convex hull) to form the polygon boundary, which is the boundary of the convex hull, and define the boundary of the convex hull as the boundary contour of the logical polygon.

[0082] Obtain the projected coordinates of two system observation points located in the intersection core area and three system observation points located in the upper left, upper right, and lower edge logical regions, respectively, on the evaluation reference plane. Based on the projected coordinates of the five system observation points, calculate the minimum area ellipse that can enclose the five projected points, and define the ellipse as a logical ellipse. Specifically, this includes: obtaining the three-dimensional coordinates of the five system observation points, including two core area observation points (both with coordinates (0, 0, ..., ...). ), For the standard density attribute values ​​of the observation points in the core area, such as =5), three edge zone observation points (three-dimensional coordinates of the upper left edge observation point (1.5, 3.5, 8), the upper right edge observation point (4.5, 3.5, 6), and the lower edge observation point (3, 1.5, 10)); using the vertical projection method, the projection coordinates of the five observation points are calculated respectively. For example, the projection point of the core area observation point (0, 0, 5) is (0.12, 0.27, 5.08), and the projection point of the upper left edge observation point (1.5, 3.5, 8) is (1.68, 3.82, 8.15), thus obtaining a list of projection coordinates of the five observation points.

[0083] The core function of the minimum area ellipse is that, given the projected coordinates of the five observation points corresponding to the core area and critical peripheral links of the target industrial system, the ellipse with the smallest area that completely encloses these coordinates can objectively reflect the reasonable standard coverage of the system. It covers all critical links without any redundant areas, providing a reference for subsequent judgment on whether the actual standard coverage is balanced. Its calculation is based on ellipse constraint optimization logic, and the specific process is broken down as follows: First, the general equation of the ellipse is set as follows... ,in , , Together they determine the shape of the ellipse (such as its flattening) and its tilt angle (whether it is aligned with the coordinate axes). , Determine the position (left-right, up-down offset) of the ellipse on the evaluation reference plane. This is a constant term that helps adjust the overall position and size of the ellipse, while also satisfying the following conditions: The validity condition is what distinguishes an ellipse from other quadratic curves, such as hyperbolas. ,parabola The key is that only when this condition is met can the graph represented by the equation be a closed ellipse, thus fulfilling the function of enclosing the observation point; the area of ​​the ellipse needs to be calculated quantitatively using the coefficients of the above equation, with the formula: s = 2π To ensure that the ellipse completely encloses the projected coordinates of the five observation points, constraints need to be set for the ellipse equation, namely, the projected coordinates (x, y) of each observation point. i y i (i=1 to 5, corresponding to five observation points) Substitute into the general equation of the ellipse, and the calculation result must satisfy... The geometric meaning of this constraint is that if the substitution result is equal to 0, it means that the observation point falls exactly on the boundary of the ellipse; if it is less than 0, it means that the observation point is inside the ellipse; only when all observation points satisfy this condition can the ellipse achieve the function of complete encirclement; during the calculation, the substitution result of each observation point needs to be verified one by one. If there is an observation point that does not satisfy the condition, the equation coefficients need to be adjusted until all points meet the constraint.

[0084] The gradient descent method is used to adjust the coefficients (a, b, c, d, e, f) of the ellipse equation. Under the premise of satisfying the constraints, the area of ​​the ellipse is gradually minimized. The specific steps are as follows: First, determine the range of values ​​for parameter k. First, analyze the distribution of the projected coordinates of the five observation points, find the maximum and minimum coordinate values ​​of all observation points in the x-direction, and calculate the difference between these two values ​​to obtain the coordinate span in the x-direction. At the same time, find the maximum and minimum coordinate values ​​of all observation points in the y-direction, and calculate the difference between these two values ​​to obtain the coordinate span in the y-direction. The parameter k needs to be determined based on these two coordinate spans, and its value range is from the square of the larger value of the x-direction and y-direction coordinate span multiplied by 1.2 to the square of the larger value multiplied by 1.5. Here, 1.2 and 1.5 are safety factors set to ensure that the initial ellipse can accurately and reasonably surround the observation points. 1.2 is used as a lower limit coefficient to avoid k being too small. If only the square of the larger value is used as k, the initial ellipse may just fit the edge of the span, while some observation points may be close to or slightly beyond the edge of the span. The initial ellipse cannot be fully covered due to insufficient coverage. Multiplying by 1.2 allows for a small amount of extra space for the initial ellipse, ensuring that all observation points are initially enclosed. 1.5 serves as an upper limit coefficient to prevent k from being too large. If k far exceeds the square of a larger value, the initial ellipse will excessively exceed the actual distribution range of the observation points. This not only contradicts the optimization direction of minimizing area but also increases the number of iterations for adjusting coefficients using gradient descent, reducing optimization efficiency. Multiplying by 1.5 controls the size of the initial ellipse within a reasonable range while ensuring coverage. This range setting avoids the initial ellipse failing to cover some observation points due to k being too small, and also prevents the initial ellipse from far exceeding the distribution range of observation points due to k being too large, ensuring that the initial ellipse roughly encloses all observation points. For example, if the x-axis coordinate span is 4 and the y-axis coordinate span is 3, then the larger of the two spans is 4, and its square is 16. In this case, the range of k is 16 × 1.2 = 19.2 to 16 × 1.5 = 24. A specific value can be selected from this range as k, such as 20.

[0085] Next, initial values ​​are assigned to the coefficients of the ellipse equation: a=1, b=0, c=1, d=0, e=0, f=-k (where k is a specific value selected within the above range). At this point, the initial ellipse equation can be simplified to: The graph represented by this equation is a perfect circle (a special form of an ellipse) centered at the origin and with a radius of √k. Considering the range of values ​​for k, this initial graph can roughly enclose all observation points. Finally, the validity of the initial coefficients is verified by substituting the initial coefficients a, b, and c into the equation. For the elliptical validity condition, if the calculation result is less than 0, it means that the initial coefficients satisfy the elliptical validity condition; if the calculation result is greater than or equal to 0, the initial values of a, b, and c need to be adjusted. For example, increase the value of a or c (such as adjusting a from 1 to 1.2, or c from 1 to 1.3), or decrease the value of b (if the initial value of b is not 0, it can be adjusted from 0.5 to 0.3) until the adjusted result is less than 0 to ensure that the initial figure is an ellipse meeting the requirements.

[0086] Substitute the initial coefficients into the area formula to obtain the initial elliptical area s0; calculate the partial derivatives (i.e., gradients) of the area s with respect to each coefficient (a, b, c, d, e, f). The positive or negative of the partial derivative represents the direction of the influence of the coefficient change on the area. For example, if the partial derivative of a certain coefficient is positive, it means that increasing this coefficient will increase the area, and this coefficient needs to be decreased to reduce the area; otherwise, this coefficient needs to be increased; adjust each coefficient along the opposite direction of the gradient (i.e., the gradient descent direction), and the adjustment amplitude is controlled by the step size (the step size is a relatively small positive number, such as 0.001, to avoid the ellipse exceeding the constraints due to excessive coefficient adjustment), to obtain a new coefficient combination (a1, b1, c1, d1, e1, f1); substitute the new coefficients into the constraint condition to verify whether the projected coordinates of all observation points satisfy the substitution result ≤ 0; if satisfied, calculate the elliptical area s1 corresponding to the new coefficients, and compare s1 with s0. If s1 < s0, retain the new coefficients; if the constraints are not satisfied or s1 ≥ s0, decrease the step size and readjust the coefficients; repeat and continuously iterate to adjust the coefficients until the area difference between two iterations is less than a preset threshold (such as 0.0001), indicating that the area has stabilized and reached the minimum value; when the area s reaches the minimum value, the figure represented by the corresponding elliptical equation is the minimum area ellipse, which is defined as the logical ellipse. This ellipse can not only completely cover the core and key links of the system, but also define a reasonable standard coverage range with the minimum area.

[0087] The geometric intersection of the logical polygon and the logical ellipse on the evaluation reference plane is calculated as the first region; the geometric difference between the logical polygon and the logical ellipse on the evaluation reference plane is calculated as the second region; the information from the first region and the second region is merged to generate standard distribution comparison data. Specifically, this includes: calculating the first region (geometric intersection), where the geometric intersection refers to the region that is simultaneously located within the boundary contour of the logical polygon and within the logical ellipse. The calculation method is as follows: traverse all projection points on the evaluation reference plane (including standard node projection points and observation point projection points). For each projection point P(x, y), first use the ray method to determine whether P is within the logical polygon, i.e., from P to x... Draw a ray along the positive axis and count the number of intersections between the ray and the boundary of the logical polygon. If the number of intersections is odd, P is determined to be inside the polygon; if it is even, it is determined to be outside the polygon. Then, substitute P(x, y) into the logical ellipse equation. If the result is ≤0, it is determined to be inside the ellipse; if the result is >0, it is determined to be outside the ellipse. If P satisfies both the condition of being inside the polygon and inside the ellipse, then the region where P is located is a component of the intersection region. Integrate all the projection points that meet the conditions and determine their coordinate range, such as x∈[1.2, 4.8], y∈[1.5, 3.9], to form the first region information (including the region coordinate range, the number of projection points in the region, and the corresponding standard node list).

[0088] The second region (geometric difference) is calculated. The geometric difference refers to the sum of two sub-regions: one inside a logical polygon but outside a logical ellipse, and the other inside a logical ellipse but outside a logical polygon. The calculation method is as follows: First, select projection points inside the polygon but outside the ellipse and determine their coordinate range, such as x∈[0.8, 1.1], y∈[2.0, 2.5]. Record the number of projection points in this region and the corresponding standard node list to form sub-region 1. Then, select projection points inside the ellipse but outside the polygon and determine their coordinate range, such as x∈[5.0, 5.2], y∈[3.6, 3.8]. Record the number of projection points in this region and the corresponding standard node list to form sub-region 2. Merge the information of sub-region 1 and sub-region 2 to form the second region information. Integrate the information of the first region and the second region, including the coordinate range of each region, the number of projection points contained in the region, the standard file number and logical coordinates corresponding to the projection points, and the region type (intersection / difference sub-region). Organize this information in a unified format (such as a tabular structure) to finally generate standard distribution comparison data.

[0089] This embodiment transforms the abstract standard node distribution into intuitive geometric regions through vertical projection and convex hull boundary calculation, avoiding the problem of ambiguous coverage in manual assessment. Simultaneously, the coordinate range of the geometric region and the number of projection points it contains provide quantitative evidence for standard distribution analysis, improving assessment accuracy. The minimum area logical ellipse generated based on five key observation points serves as an objective reference reflecting the reasonable standard coverage of the target industry system. This ellipse is jointly determined by the core and peripheral key links of the system, eliminating the need for subjective expert setting and avoiding assessment deviations caused by different teams using different reference standards, ensuring consistency in standard distribution comparison. Intersection and difference calculations clearly distinguish between areas where standard coverage matches the reasonable range and areas with redundant or incomplete standard coverage, addressing the pain point of difficulty in manually identifying subtle coverage issues. From projection calculation and convex hull construction to ellipse generation and region comparison, the entire process is automated based on algorithms, eliminating the need for manual point-by-point analysis or graphical drawing. Even with a large number of standard nodes, region calculation and data integration can be completed quickly, shortening the standard distribution assessment cycle.

[0090] In a preferred embodiment of the present invention, based on standard distribution comparison data, the standard dependency network mapped to the logical topology skeleton is corrected and integrated to generate a standard system model of the entire life cycle of the target industry system, including:

[0091] Analyze the second region in the standard distribution comparison data, identify projection points located within logical polygons but outside logical ellipses, and projection points located within logical ellipses but outside logical polygons. Mark the inter-standard dependency network nodes corresponding to the projection points as nodes to be corrected. Specifically, this includes: first, splitting the second region in the standard distribution comparison data, clarifying the projection point ranges of the two sub-regions, extracting projection points within logical polygons but outside logical ellipses (denoted as sub-region Q projection points) and projection points within logical ellipses but outside logical polygons (denoted as sub-region V projection points) from the second region information; and then using a table of correspondence between the coordinates (x, y, z) of the projection points and the original nodes (this table records which inter-standard dependency network node each projection point was projected from). The original network node corresponding to each projection point is located in reverse. The original network nodes corresponding to the projection points in sub-region Q and sub-region V are uniformly marked as nodes to be calibrated. At the same time, a sub-region identifier (marked as belonging to sub-region Q or V) is added to each node to be calibrated to ensure the accuracy of subsequent adjustment direction and avoid confusion. For example, if there is a projection point (5.1, 3.7, 6.2) in sub-region Q, and its corresponding original network node is found to be the environmental protection standard for power battery recycling through the correspondence table, then this standard node is marked as a node to be calibrated, and the sub-region identifier is Q; if there is a projection point (0.9, 2.1, 5.3) in sub-region V, and its corresponding original node is the basic standard for battery materials, then it is marked as a node to be calibrated, and the sub-region identifier is V.

[0092] For each node to be corrected, based on its position in the standard distribution comparison data, adjust the logical coordinates of the corresponding node in the logical topology skeleton. This moves the projection point of the corresponding node towards the boundary of the logical ellipse or logical polygon, completing the topology correction of the inter-standard dependency network to obtain the corrected inter-standard dependency network. Specifically, for each node to be corrected, determine the adjustment direction based on its sub-region identifier. By modifying the node's two-dimensional logical coordinates (business stage dimension X, technical level dimension Y), move the projection point towards the target boundary (ellipse or polygon boundary). The specific steps are as follows: If the node belongs to sub-region Q (inside the polygon, outside the ellipse), and the target boundary is the logical ellipse boundary, the boundary parameters of the logical ellipse (such as the maximum / minimum coordinates of the ellipse in the X direction) must be obtained first. , Maximum / minimum coordinates in the Y direction , If the node belongs to sub-region V (outside the polygon inside the ellipse), the target boundary is the logical polygon boundary. Obtain the boundary parameters of the logical polygon (e.g., , , , ); Set an adjustment coefficient (ranging from 0.3 to 0.6, balancing adjustment efficiency and stability; here, 0.5 is used); The calculation of the adjustment amount needs to be differentiated according to the sub-region type. The core logic is to make the node coordinates move closer to the target boundary. Specifically, for sub-region Q node (which needs to move closer to the ellipse boundary), since the node's current coordinates exceed the ellipse boundary, the adjustment amount = (node's current coordinates - ellipse boundary coordinates) × adjustment coefficient; the adjusted coordinates = node's current coordinates - adjustment amount. By subtracting, the node coordinates are moved closer to the ellipse boundary, reducing the distance to the ellipse. For sub-region V node (which needs to move closer to the polygon boundary), since the node's current coordinates have not reached the polygon boundary, the adjustment amount = (polygon boundary coordinates - node's current coordinates) × adjustment coefficient; the adjusted coordinates = node's current coordinates + adjustment amount. By adding, the node coordinates are moved closer to the polygon boundary, reducing the distance to the polygon.

[0093] The adjusted logical coordinates are combined with the original standard density attribute value Z, and the projection points are recalculated. It is verified whether the projection points have moved towards the target boundary (e.g., the distance from the new projection point of the sub-region Q node to the elliptical boundary is smaller than before the adjustment). If the movement effect is not as expected (e.g., the distance reduction is less than 0.05), the adjustment coefficient is increased (e.g., from 0.5 to 0.6), and the adjustment amount is recalculated until the projection points are close to the target boundary. After all the nodes to be corrected are adjusted, the inter-standard dependency network is updated, the old logical coordinates of the nodes are replaced with the new coordinates, and the directed edges between the nodes are re-verified. The verification logic here is consistent with the previous direct reference relationship and indirect support relationship. The weight values ​​(the first weight is based on the explicitness of the reference, and the second weight is based on the semantic similarity) remain unchanged. After the above updates are completed, the corrected inter-standard dependency network is finally formed.

[0094] The corrected inter-standard dependency network is mapped to the nodes in the working scope directory. Each node in the working scope directory is bound to a corresponding set of standards. The relationships between the standard sets are established based on the relational weights in the inter-standard dependency network. Specifically, this involves: traversing each node in the working scope directory, such as R&D phase - requirements analysis - user requirements survey, and testing phase - performance testing - energy density testing, and extracting the core descriptive keywords of the nodes, such as user requirements survey and energy density testing; simultaneously, traversing each node (standard document) in the corrected inter-standard dependency network and extracting the name and core keywords of the standard document, such as the keyword for the power battery user requirements survey specification being user requirements survey, and the keyword for the power battery energy density testing standard being energy density testing; and matching the two through keyword overlap, where keyword overlap = (number of common keywords ÷ total number of keywords in the working scope nodes) × 100%. If the overlap is ≥ 60%, the standard document is matched to the corresponding working scope node.

[0095] Each working scope node is bound to a standard set, which is to summarize all standard files that match that node to form a standard set specific to that node. For example, the standard set of the performance testing node in the testing phase includes energy density testing standards and cycle life testing standards. According to the relationship weights (first weight and second weight) between standard files in the corrected dependency network, if any standard file in the g standard set has a direct reference relationship with any standard file in the h standard set (first weight 1.0), then the g standard set and the h standard set are directly associated, and the association weight is the average of the weights of all related standard files (e.g., if two standards in g reference one standard in h, the weights are 1.0 and 0.9 respectively, with an average weight of 0.95); if there is an indirect support relationship (second weight 0.8), then an indirect association is established, and the weight is the average of the weights.

[0096] The process encapsulates the working scope directory, the corrected inter-standard dependency network, and the mapping relationship between them to generate a standard system model covering the entire lifecycle of the target industry system. Specifically, the encapsulated content includes three types of core information: the working scope directory, the complete hierarchical structure (first-level stage, second-level node, third-level sub-node) and the description information of each node; the corrected inter-standard dependency network, including all nodes (standard number, name, logical coordinates of standard files), directed edges (direct reference / indirect support relationship), and edge weights (first weight / second weight); and a mapping relationship table, showing the correspondence between nodes in the working scope directory and the corresponding standard sets, as well as the association relationships and weights between standard sets.

[0097] The above three types of information are encapsulated into a structured file according to the principles of hierarchy and queryability. For example, in JSON format, the working scope directory is stored in the form of nested dictionaries, the dependency network is stored in the form of node list and edge list, and the mapping relationship table is stored in the form of key-value pairs. The final generated file is the standard system model of the entire life cycle of the target industry system.

[0098] This embodiment eliminates the problems of redundant standard distribution and insufficient standard coverage in the original network by identifying and adjusting the coordinates of nodes to be calibrated. This makes the logical coordinates of network nodes more consistent with the reasonable range of the system, and the correlation between nodes more in line with the actual technical logic of the industry, thereby improving the accuracy of the network's representation of industry standards. The correlation mapping breaks the separation between work tasks and standard documents. Each work node has a clear standard set to support it, avoiding the problem of having no standard to follow or using standards in a chaotic manner when executing tasks. At the same time, the correlation between standard sets also provides clear guidance for the connection of standards in the task process. The generated standard system model covers the entire life cycle of the industry from R&D to recycling, integrating three core types of information: business processes, standard documents, and correlation relationships. It can be directly used for standard planning, evaluation, and updating. From the identification of nodes to be calibrated to model encapsulation, the entire process is automated based on data and algorithms. There is no need for manual matching of work nodes and standard documents one by one, which also avoids the subjective bias of manually setting correlation relationships, reduces the labor cost of standardization management, and ensures the objectivity and consistency of model results.

[0099] like Figure 2 As shown, embodiments of the present invention also provide a method for constructing a standard system model, comprising:

[0100] Step 1: Based on the knowledge of the target industry system, generate a work scope catalog covering the entire life cycle of the target industry system, and extract two dimensions, business stage and technology level, as logical baselines to construct a logical topology skeleton;

[0101] Step 2: Parse the content of the standard files in the standard file set to generate a dependency network between the standards;

[0102] Step 3: Within the two-dimensional logical space formed by the logical topology skeleton, determine that two system observation points are located in the core area where the business stage logical reference axis and the technical level logical reference axis intersect, and determine that three system observation points are located in the upper left, upper right and lower edge logical areas of the logical space, respectively, far from the core area where the intersection occurs.

[0103] Step 4: Based on the coordinates of the three system observation points and combined with the standard density attribute, solve for and determine an evaluation reference plane in the three-dimensional logical attribute space formed by expanding the two-dimensional logical space.

[0104] Step 5: Map the standard dependency network onto the evaluation reference plane to form a standard coverage area, and define a logical polygon based on the boundary of the standard coverage area; generate a logical ellipse based on the distribution of the five system observation points, detect and extract the intersection and difference areas of the logical polygon and the logical ellipse on the evaluation reference plane, and generate standard distribution comparison data.

[0105] Step 6: Based on the standard distribution comparison data, correct and integrate the standard dependency network mapped to the logical topology skeleton to generate a standard system model for the entire life cycle of the target industry system.

[0106] It should be noted that this method is a system corresponding to the above system. All implementation methods in the above system embodiments are applicable to this embodiment and can achieve the same technical effect.

[0107] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A system for constructing a standard system model, characterized in that, include: The modeling module is used to generate a work scope catalog covering the entire life cycle of the target industry system based on knowledge of the target industry system, and extract two dimensions, business stage and technology level, as logical baseline axes to construct a logical topology skeleton; The parsing module is used to parse the content of standard files in the standard file set and generate a dependency network between standards. The positioning module is used to determine, within the two-dimensional logical space formed by the logical topology skeleton, the intersection core area of ​​two system observation points located at the business stage logical reference axis and the technical level logical reference axis, and to determine the upper left, upper right and lower edge logical areas of three system observation points located in the logical space far from the intersection core area. The module is used to determine an evaluation reference plane in a three-dimensional logical attribute space formed by expanding a two-dimensional logical space, based on the coordinates of three system observation points and in combination with standard density attributes. The extraction module maps the standard dependency network to the evaluation reference plane to form a standard coverage area and defines a logical polygon based on the boundary of the standard coverage area; it generates a logical ellipse based on the distribution of the five system observation points, detects and extracts the intersection and difference areas of the logical polygon and the logical ellipse on the evaluation reference plane, and generates standard distribution comparison data. The model generation module is used to correct and integrate the standard dependency network mapped to the logical topology skeleton based on the standard distribution comparison data, and generate a standard system model of the entire life cycle of the target industry system.

2. The system for constructing a standard system model according to claim 1, characterized in that, The process of obtaining the standard file set is as follows: Access the standard bibliographic database, standard full-text database, and standard bulletin database to obtain published standard documents within the target industry system; Extract the standard number, standard name, publication date, and main text of each standard document; The standard documents are deduplicated according to their standard numbers and then integrated to form a set of standard documents.

3. The system for constructing a standard system model according to claim 2, characterized in that, Based on knowledge of the target industry system, a work scope catalog covering the entire lifecycle of the target industry system is generated, and two dimensions, business stage and technical level, are extracted as logical baselines to construct a logical topology skeleton, including: Extract thematic keywords describing the activities and technologies of the target industry system from the standard document set's standard names and text content; Based on the co-occurrence relationship and context of the keywords in the standard text, the themes describing the time sequence and process evolution are clustered into the stage sequence of the target industry system, forming the stage division of the work scope catalog; Based on the phase division, the main text of standard documents belonging to the same phase is aggregated to obtain aggregated text. Phrases describing specific tasks, outputs and technical components are identified and extracted from the aggregated text as sub-nodes of the corresponding phase, and combined to form a complete work scope directory. Perform attribute analysis on all sub-nodes in the work scope directory, classify phrase units with temporal attributes in the node description to the business stage dimension, and classify phrase units with hierarchical or technical abstract attributes in the node description to the technical level dimension. With the business stage dimension as the horizontal axis and the technical level dimension as the vertical axis, an orthogonal two-dimensional coordinate system is established, and the two-dimensional coordinate system constitutes the logical topological skeleton.

4. The system for constructing a standard system model according to claim 3, characterized in that, The standard documents in the standard document set are parsed to generate a dependency network between the standards, including: Based on the business stage dimension and technical level dimension of the logical topology skeleton, the standard files in the standard file set are annotated with content, and the logical coordinates of each standard file in the standard file set in the two-dimensional logical space are determined. The normative reference section of each standard document in the standard document set is parsed, the standard number of the referenced standard listed in the normative reference section is extracted, and a direct reference relationship from the current standard document to the referenced standard document is established. Analyze the technical content of the clauses in the main text of the standard documents in the standard document set, calculate the semantic similarity of different standard documents in the standard document set in terms of terminology definitions, technical parameters and functional requirements, and identify standard document pairs with semantic similarity exceeding a preset threshold as indirect support relationships; Assign a first weight value based on the explicitness of the reference to each direct reference relationship, and assign a second weight value based on the semantic similarity value to each indirect support relationship; Using each standard file in the standard file set as a node, and direct reference relationships with a first weight value and indirect support relationships with a second weight value as directed edges, a dependency network between standards is constructed. Each node in the dependency network between standards contains the logical coordinate attribute of the standard file corresponding to the node.

5. The system for constructing a standard system model according to claim 4, characterized in that, Within the two-dimensional logical space constituted by the logical topology skeleton, two system observation points are identified as being located in the core area where the business stage logical reference axis and the technical level logical reference axis intersect. Three system observation points are identified as being located in the upper left, upper right, and lower edge logical regions of the logical space, respectively, far from the core area. The origin of the coordinate system in the two-dimensional logical space is defined as the core area where the business stage logical reference axis and the technical level logical reference axis intersect, and the coordinates of the observation points of the two systems are set as the coordinates of the origin. Obtain the logical coordinates of all nodes in the inter-standard dependency network, determine the minimum and maximum logical coordinate values ​​for the business stage dimension, and the minimum and maximum logical coordinate values ​​for the technology level dimension; The region in the two-dimensional logical space where the business stage dimension has the minimum logical coordinate value and the technical level dimension has the maximum logical coordinate value is defined as the upper left edge logical region; the region where the business stage dimension has the maximum logical coordinate value and the technical level dimension has the maximum logical coordinate value is defined as the upper right edge logical region; and the region where the business stage dimension has the median logical coordinate value and the technical level dimension has the minimum logical coordinate value is defined as the lower edge logical region. Within the coordinate ranges corresponding to the upper left, upper right, and lower edge logic regions, the coordinates of the geometric center point of the upper left edge logic region are selected as the first system observation point coordinates, the coordinates of the geometric center point of the upper right edge logic region are selected as the second system observation point coordinates, and the coordinates of the geometric center point of the lower edge logic region are selected as the third system observation point coordinates.

6. The system for constructing a standard system model according to claim 5, characterized in that, Based on the coordinates of the three system observation points and combined with the standard density attribute, an evaluation reference plane is determined in the three-dimensional logical attribute space formed by expanding the two-dimensional logical space, including: For system observation points located in the upper left edge logic region, the upper right edge logic region, and the lower edge logic region, count the number of standard dependency network nodes contained in the circular logic region centered on the coordinates of the system observation point and with a preset logical distance as the radius. Use this number as the standard density attribute value of the system observation point. The first three-dimensional coordinate point is formed by combining the two-dimensional logical coordinates of the system observation point located in the upper left logical region with the standard density attribute value of the system observation point; the second three-dimensional coordinate point is formed by combining the two-dimensional logical coordinates of the system observation point located in the upper right logical region with the standard density attribute value of the system observation point; and the third three-dimensional coordinate point is formed by combining the two-dimensional logical coordinates of the system observation point located in the lower logical region with the standard density attribute value of the system observation point. Based on the first, second, and third three-dimensional coordinate points, a unique evaluation reference plane is solved and determined through spatial plane fitting calculation.

7. The system for constructing a standard system model according to claim 6, characterized in that, Based on the first, second, and third 3D coordinate points, a unique evaluation reference plane is solved and determined through spatial plane fitting calculations, including: The first three-dimensional coordinate point, the second three-dimensional coordinate point, and the third three-dimensional coordinate point are used as three non-collinear constraint points in space; Using the geometric principle of determining a spatial plane by three points, calculate the equation of the spatial plane passing through the first three-dimensional coordinate point, the second three-dimensional coordinate point, and the third three-dimensional coordinate point; The plane defined by the spatial plane equation is determined as the evaluation reference plane for mapping and evaluating the standard distribution.

8. The system for constructing a standard system model according to claim 7, characterized in that, The standard dependency network is mapped onto the evaluation reference plane to form a standard coverage area, and logical polygons are defined based on the boundaries of the standard coverage area. Logical ellipses are generated based on the distribution of the five system observation points. The intersection and difference regions between the logical polygons and logical ellipses on the evaluation reference plane are detected and extracted to generate standard distribution comparison data, including: The logical coordinates of each node in the inter-standard dependency network are vertically projected onto the evaluation reference plane to obtain the projection point of each node on the evaluation reference plane. All projection points constitute the standard coverage area. Calculate the convex hull boundary of all projected points in the standard coverage area, and define the convex hull boundary as the boundary contour of a logical polygon; Obtain the projected coordinates of two system observation points located in the intersection core area and three system observation points located in the upper left edge logical region, the upper right edge logical region and the lower edge logical region respectively on the evaluation reference plane; based on the projected coordinates of the five system observation points, calculate the minimum area ellipse that can enclose the five projected points, and define the ellipse as a logical ellipse. The geometric intersection of the logical polygon and the logical ellipse on the evaluation reference plane is calculated as the first region; the geometric difference between the logical polygon and the logical ellipse on the evaluation reference plane is calculated as the second region; the information from the first region and the second region is merged to generate standard distribution comparison data.

9. The system for constructing a standard system model according to claim 8, characterized in that, Based on the standard distribution comparison data, the standard dependency network mapped to the logical topology skeleton is corrected and integrated to generate a standard system model of the entire life cycle of the target industry system, including: Analyze the second region in the standard distribution comparison data, identify the projection points located inside the logical polygon but outside the logical ellipse, and the projection points located inside the logical ellipse but outside the logical polygon, and mark the inter-standard dependency network nodes corresponding to the projection points as nodes to be corrected. For each node to be corrected, the logical coordinates of the corresponding node in the logical topology skeleton are adjusted according to the position of the corresponding node in the standard distribution comparison data, so that the projection point of the corresponding node moves towards the boundary of the logical ellipse or logical polygon, thereby completing the topology correction of the inter-standard dependency network and obtaining the corrected inter-standard dependency network. The corrected inter-standard dependency network is mapped to the nodes in the working scope directory. Each node in the working scope directory is bound to a corresponding set of standards, and the association relationship between the sets of standards is established according to the relationship weight in the inter-standard dependency network. The encapsulated working scope directory, the corrected inter-standard dependency network, and the mapping relationship between them are used to generate a standard system model for the entire life cycle of the target industry system.

10. A method for constructing a standard system model, wherein the method implements the system as described in any one of claims 1 to 9, characterized in that, include: Step 1: Based on the knowledge of the target industry system, generate a work scope catalog covering the entire life cycle of the target industry system, and extract two dimensions, business stage and technology level, as logical baselines to construct a logical topology skeleton; Step 2: Parse the content of the standard files in the standard file set to generate a dependency network between the standards; Step 3: Within the two-dimensional logical space formed by the logical topology skeleton, determine that two system observation points are located in the core area where the business stage logical reference axis and the technical level logical reference axis intersect, and determine that three system observation points are located in the upper left, upper right and lower edge logical areas of the logical space, respectively, far from the core area where the intersection occurs. Step 4: Based on the coordinates of the three system observation points and combined with the standard density attribute, determine an evaluation reference plane in the three-dimensional logical attribute space formed by expanding the two-dimensional logical space. Step 5: Map the standard dependency network onto the evaluation reference plane to form a standard coverage area, and define a logical polygon based on the boundary of the standard coverage area; generate a logical ellipse based on the distribution of the five system observation points, detect and extract the intersection and difference areas of the logical polygon and the logical ellipse on the evaluation reference plane, and generate standard distribution comparison data. Step 6: Based on the standard distribution comparison data, correct and integrate the standard dependency network mapped to the logical topology skeleton to generate a standard system model for the entire life cycle of the target industry system.

Images