Substation monitoring information automatic generation and semantic enhancement verification method and system

Abstract

The invention discloses a transformer substation monitoring information automatic generation and semantic enhancement verification method and system, and belongs to the field of power monitoring information management.The method comprises the steps that S100, a power signal synonymous mapping library, a transformer substation knowledge graph, a scenarized adaptation rule library and a defect pre-judgment rule library are constructed; s200, constructing a self-adaptive template library; using the adaptive template library to generate standardized substation monitoring information; s300, preprocessing the data; s400, aiming at the structured fields, a structured matching algorithm is adopted to carry out comparison; s500, inputting the information to be subjected to semantic matching into the special semantic model, and calculating semantic similarity; performing auxiliary verification in combination with a power signal synonymous mapping library; s600, on the basis of the knowledge graph, querying an equipment affiliation relationship and interval association logic corresponding to the monitoring information to be verified; verifying the compliance of the incidence relation of the monitoring information; and S700, summarizing results. According to the invention, an efficient technical solution can be provided for transformer substation monitoring information processing.

Description

Technical Field

[0001] This invention relates to the field of power monitoring information management technology, and in particular to a method and system for automatic generation and semantic enhancement verification of substation monitoring information. Background Technology

[0002] The generation and verification of existing substation monitoring information mainly rely on manual compilation and fragmented tools. During the information generation stage, maintenance personnel need to manually enter information such as signal names, associated equipment, and bay numbers by referring to equipment manuals, engineering drawings, and historical ledgers. This is prone to problems such as inconsistent signal naming and missing fields due to differences in personnel experience. During the verification stage, single-dimensional local verification methods are mostly used, which can only achieve accurate matching of structured fields such as signal identifiers and voltage levels. They cannot handle semantic differences between synonyms with different names and lack systematic verification of the association logic between signals, equipment, and bays, relying on manual verification of engineering drawings to confirm the association relationship. Summary of the Invention

[0003] This invention aims to address at least one of the technical problems existing in the prior art. To this end, this invention proposes a method and system for automatically generating and semantically enhancing substation monitoring information, which can balance processing efficiency and result reliability, providing a systematic and efficient technical solution for substation monitoring information processing.

[0004] This invention provides a method for automatically generating and semantically enhancing substation monitoring information, comprising the following steps: S100, acquiring synonym mapping relationships for power signals and generating a power signal synonym mapping library; constructing a knowledge graph of directional associations of core elements of the substation, wherein the knowledge graph has signals, equipment, bays, and parameters as core nodes; obtaining a scenario-based adaptation rule library according to the engineering characteristics and naming conventions of the substation, and simultaneously constructing a defect prediction rule library based on high-frequency defect characteristics; S200, constructing an adaptive template library based on the scenario-based adaptation rule library, and achieving automatic adaptation of templates to different engineering characteristics through scenario-based rule mapping; integrating equipment ICDs. Documents, historical records, and engineering requirement data are used to generate standardized substation monitoring information that conforms to naming conventions and engineering characteristics using an adaptive template library; S300: The generated standardized monitoring information and preset standard templates are processed to unify field formats, aligning core comparison dimensions, including signal identifiers, voltage levels, and parameter types; redundant data and abnormal information are cleaned; S400: For structured fields including signal identifiers, voltage levels, and parameter types, an exact matching algorithm is used for comparison. The exact matching algorithm matches the field values ​​and formats of each structured field to be completely equal; based on the comparison results, the monitoring information is divided into three categories: complete match, obvious mismatch, and pending semantic matching; if all structured fields meet the exact matching requirement, it is determined to be a complete match; if any structured field has a substantial difference in field value content, it is determined to be an obvious mismatch; if all structured fields have the same format and no substantial difference in field value content, the core values ​​are consistent, only the description details are different, and not all meet the exact matching requirement, it is determined to be pending semantic matching; information pending semantic matching is obtained by filtering; S500: Sentence-BERT with parameter simplification and optimization is selected. The model, serving as a foundational semantic model, uses data from a power signal synonym mapping library as training samples to fine-tune the foundational semantic model, obtaining a dedicated semantic model adapted to power scenarios. The semantic matching information is input into the dedicated semantic model, and semantic similarity is calculated. Assisted verification is performed using the power signal synonym mapping library, and semantic matching results are output based on preset judgment criteria. S600: Relying on the knowledge graph of core elements of the substation, the equipment affiliation and interval association logic corresponding to the monitoring information to be verified are queried. The compliance of the monitoring information association is verified against the association logic standards defined in the scenario-based adaptation rule library, and the association verification results are output. S700: The results of structured data matching, semantically enhanced deep matching, and association compliance verification are summarized, generating a complete report containing verification conclusions and specific reasons for failure. For monitoring information with defects, targeted correction suggestions are pushed based on the defect prediction rule library.

[0005] The embodiments of this invention achieve at least the following beneficial effects: By constructing a synonym mapping library, a targeted association knowledge graph, and a scenario-based adaptation rule library, the embodiments of this invention transform the implicit knowledge of the power industry into structured data, providing convenience for automated processing; by integrating an adaptive template library and multi-source data, monitoring information can be generated with one click, replacing manual compilation work and greatly improving generation efficiency; addressing the limitations of traditional verification methods that can only cover the format dimension and rely on manual intervention, the semantic enhancement verification of the embodiments of this invention covers the three core dimensions of information verification: format, content, and logic, from structured fields and semantic connotations to association logic.

[0006] According to some embodiments of the present invention, the node association logic of the substation core element directed association knowledge graph includes the affiliation relationship between signals and equipment, the membership relationship between equipment and bays, and the correspondence relationship between signals and parameters. The knowledge graph is deployed through an open-source graph database and only stores the necessary association data for verification. When querying association relationships based on the knowledge graph, Cypher query statements are used to achieve fast retrieval. The association logic standards defined in the scenario-based adaptation rule base are that the same signal can only belong to one equipment, the same equipment can only belong to one bay, and the signal type matches the associated equipment type.

[0007] According to some embodiments of the present invention, when fine-tuning the basic semantic model, the underlying parameters are frozen and only the top-level classification head is fine-tuned; the preset judgment criteria are: when the semantic similarity is greater than or equal to the first threshold, it is judged as semantic matching passed; when the semantic similarity is less than the first threshold but greater than or equal to the second threshold, it is marked as pending manual review; when the semantic similarity is less than the second threshold, it is judged as semantic mismatch.

[0008] According to some embodiments of the present invention, the logic of using a power signal synonym mapping library for auxiliary verification is as follows: if the name combination of the information to be semantically matched has a clear association in the synonym mapping library, the semantic matching result is directly confirmed; when there is no corresponding association in the synonym mapping library, the calculation result of the dedicated semantic model shall prevail.

[0009] According to some embodiments of the present invention, the template fields of the adaptive template library include signal identifier, signal name, signal type, associated device, interval number, voltage level, parameter range and functional description. The template fields can be dynamically added or removed according to the engineering characteristics. The different engineering characteristics include new construction projects, renovation and expansion projects and intelligent transformation projects of conventional stations. The scenario-based rule mapping automatically calls the corresponding template configuration by matching the engineering type.

[0010] According to some embodiments of the present invention, the design includes a comprehensive reward function that incorporates water-saving effect and user satisfaction indicators, including: cleaning redundant data and abnormal information includes removing duplicate signal records, completing missing core fields, removing parameter values ​​that exceed reasonable ranges, and correcting field content with format errors, wherein the core fields are signal identifier, associated device, interval number, and voltage level.

[0011] According to some embodiments of the present invention, when performing the semantic enhancement deep matching in step S500 and the compliance verification of the association relationship in step S600, a bidirectional collaborative mechanism between the semantic knowledge graph and the deep learning model is adopted; the semantic similarity result output by the dedicated semantic model in step S500 is used as a pre-screening condition for the association relationship query of the knowledge graph, and the association relationship verification in step S600 is only performed on the monitoring information that has passed the semantic matching or is pending review; at the same time, the association logic result confirmed by the knowledge graph in step S600 is fed back to the dedicated semantic model to dynamically optimize the semantic similarity calculation weight of the model for closely related signals.

[0012] According to some embodiments of the present invention, after generating standardized substation monitoring information in step S200 and before performing pre-verification data preprocessing in step S300, the method further includes a machine learning prediction-assisted optimization step, which includes using a machine learning model to train the initially generated monitoring information and historical verification data to predict potential defects. The potential defects include ambiguous signal naming, abnormal correlation, and non-compliant parameter configuration. High-risk signals are marked and obvious defects are filtered out in advance.

[0013] According to some embodiments of the present invention, after outputting targeted correction suggestions in step S700, cross-version linkage editing is performed, high-risk defect signals predicted by machine learning are displayed first, and compliant modification directions are automatically recommended in combination with historical modification records. After editing is completed, the prediction results and correction data are synchronously fed back to the machine learning model.

[0014] Another embodiment of the present invention provides an automatic generation and semantic enhancement verification system for substation monitoring information, comprising: a basic resource construction module for acquiring synonym mapping relationships of power signals and generating a power signal synonym mapping library; constructing a directional association knowledge graph of core elements of the substation, wherein the knowledge graph has signals, equipment, bays, and parameters as core nodes; obtaining a scenario-based adaptation rule library according to the engineering characteristics and naming conventions of the substation, and simultaneously constructing a defect prediction rule library based on high-frequency defect characteristics; a monitoring information generation module for constructing an adaptive template library based on the scenario-based adaptation rule library, and realizing automatic adaptation of templates to different engineering characteristics through scenario-based rule mapping; and integrating equipment ICDs. Documents, historical records, and engineering requirements data are used to generate standardized substation monitoring information that conforms to naming conventions and engineering characteristics using an adaptive template library. A data preprocessing module unifies the field formats of the generated standardized monitoring information with preset standard templates, aligning core comparison dimensions, including signal identifiers, voltage levels, and parameter types; redundant data and abnormal information are cleaned. A structured data matching module compares structured fields, including signal identifiers, voltage levels, and parameter types, using an exact matching algorithm. This exact matching algorithm matches fields where both the content and format of the field values ​​are completely equal. Based on the comparison results, the monitoring information is divided into three categories: complete match, obvious mismatch, and pending semantic matching. If all structured fields satisfy exact matching, it is considered a complete match; if any structured field has a substantial difference in its content, it is considered an obvious mismatch; if all structured fields have the same format and no substantial differences in content, with consistent core values ​​but differences only in description details, and not all satisfy exact matching, it is considered pending semantic matching. Information pending semantic matching is then filtered out. A semantic enhancement matching module selects a parameter-optimized Sentence-BERT. The model serves as a foundational semantic model, using data from a power signal synonym mapping library as training samples. This foundational semantic model is then fine-tuned to obtain a dedicated semantic model adapted to power scenarios. The semantic matching information is input into this dedicated semantic model, and semantic similarity is calculated. Assisted verification is performed using the power signal synonym mapping library, and semantic matching results are output based on preset judgment criteria. A correlation verification module is used to query the equipment affiliation and interval correlation logic corresponding to the monitoring information to be verified, based on a knowledge graph of core elements of the substation. The compliance of the monitoring information correlation is verified against the correlation logic standards defined in the scenario-based adaptation rule library, and the correlation verification results are output. A verification result module summarizes the results of structured data matching, semantically enhanced deep matching, and correlation compliance verification, generating a complete report including verification conclusions and specific reasons for failure. For monitoring information with defects, targeted correction suggestions are pushed based on a defect prediction rule library.

[0015] The embodiments of this invention achieve at least the following beneficial effects: By constructing a synonym mapping library, a targeted association knowledge graph, and a scenario-based adaptation rule library, the embodiments of this invention transform the implicit knowledge of the power industry into structured data, providing convenience for automated processing; by integrating an adaptive template library and multi-source data, monitoring information can be generated with one click, replacing manual compilation work and greatly improving generation efficiency; addressing the limitations of traditional verification methods that can only cover the format dimension and rely on manual intervention, the semantic enhancement verification of the embodiments of this invention covers the three core dimensions of information verification: format, content, and logic, from structured fields and semantic connotations to association logic.

[0016] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0017] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which:

[0018] Figure 1 This is a flowchart illustrating the method according to an embodiment of the present invention.

[0019] Figure 2 This is a flowchart illustrating the bidirectional collaborative mechanism of an embodiment of the present invention.

[0020] Figure 3 This is a schematic diagram of the machine learning prediction-assisted optimization process according to an embodiment of the present invention.

[0021] Figure 4 This is a schematic diagram of the cross-version collaborative editing process according to an embodiment of the present invention.

[0022] Figure 5 This is a schematic block diagram of the system modules according to an embodiment of the present invention.

[0023] Figure label:

[0024] The module includes: basic resource construction module 100, monitoring information generation module 200, data preprocessing module 300, structured data matching module 400, semantic enhancement matching module 500, association verification module 600, and verification result module 700. Detailed Implementation

[0025] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0026] In the description of this invention, "several" means one or more, "multiple" means two or more, "greater than," "less than," and "exceeding" are understood to exclude the stated number, while "above," "below," and "within" are understood to include the stated number. The use of "first" and "second" in the description is merely for distinguishing technical features and should not be construed as indicating or implying relative importance, or implicitly indicating the number of indicated technical features, or implicitly indicating the order of the indicated technical features.

[0027] Definitions:

[0028] 1. Parameter-simplified and optimized Sentence-BERT model: A lightweight semantic representation model optimized from the original Sentence-BERT architecture. It achieves a lightweight design by reducing the network parameter size and simplifying redundant computation layers, reducing computational consumption and deployment costs while retaining core semantic encoding and similarity calculation capabilities. In this solution, the model specifically refers to a parameter-simplified version adapted for low-cost deployment in industrial scenarios. After targeted fine-tuning using a power signal synonym mapping library as training samples, it is specifically designed to capture the semantic associations of substation monitoring signal names, resolving semantic ambiguity caused by different naming conventions and abbreviation variations, and supporting the efficient execution of semantically enhanced deep matching.

[0029] 2. Power Signal Synonym Mapping Library: A semantic association resource specifically built for substation monitoring scenarios. Its core includes the correspondence between standard signal names and various synonym variants, covering naming conventions of different equipment manufacturers, commonly used expressions in engineering scenarios, simplified abbreviations, and other synonym forms. It is used to assist semantic models in verifying matching results and improve the accuracy of semantic comparison and scenario adaptability.

[0030] 3. Substation Core Element Targeted Association Knowledge Graph: A dedicated knowledge graph constructed with core elements of the substation monitoring scenario as nodes and necessary association logic as edges. Core nodes include signals, equipment, bays, and parameters. Association logic covers attribution, membership, and correspondence relationships. The purpose of this construction is to provide structured logical support for the compliance verification of the association relationships of monitoring information, retaining only the necessary data required for verification and avoiding redundant attributes from affecting query efficiency.

[0031] 4. Scenario-based Adaptation Rule Base: A set of rules built based on substation engineering characteristics, industry naming conventions, and equipment manufacturer standards. It covers adaptation logic for different engineering types (new construction, renovation and expansion, etc.), wiring methods, and equipment types. Its core purpose is to guide the dynamic adaptation of the adaptive template library and the standardized generation of monitoring information, ensuring that the generated monitoring information meets the compliance requirements of specific engineering scenarios.

[0032] 5. Defect Prediction Rule Base: This rule base integrates frequently occurring defect types and judgment conditions during the verification of substation monitoring information. It primarily covers defect types such as incorrect signal naming, abnormal correlation, and non-compliant parameter configuration. It is used to identify potential defects in advance, push targeted correction suggestions, and improve verification and optimization efficiency.

[0033] 6. Adaptive Template Library: A standardized collection of monitoring information templates built on a scenario-based adaptation rule base. The templates include core fields such as signal identifier, signal name, associated equipment, and bay number. They support dynamic adjustment of field composition and configuration logic according to different project characteristics. The core function is to automatically generate substation monitoring information that conforms to the standards, reducing the workload of manual compilation.

[0034] 7. Dedicated Semantic Model: This semantic model is formed by fine-tuning the Sentence-BERT model, which has been optimized with simplified parameters, in the context of power scenarios. By incorporating industry-specific semantic features from the power signal synonym mapping library, it optimizes the model's semantic understanding of power scenario signal names. It is primarily used for similarity calculation of signals to be semantically matched and is the core execution module for semantically enhanced deep matching.

[0035] 8. Cross-version collaborative editing: This function optimizes the collaborative editing of substation monitoring information across multiple versions. It supports the joint modification of monitoring information from different versions, prioritizes the display of high-risk defect signals, traces historical modification records, and recommends compliant modification directions. After editing, the relevant resource library is updated synchronously. Its core purpose is to improve the efficiency of defect correction and the consistency of information across multiple versions.

[0036] Reference Figure 1 This invention proposes a method for automatically generating and semantically enhancing substation monitoring information, comprising the following steps:

[0037] S100. Obtain synonym mapping relationships for power signals and generate a synonym mapping library for power signals; construct a knowledge graph of directional associations of core elements of substations, with signals, equipment, bays and parameters as core nodes; obtain a scenario-based adaptation rule library based on the engineering characteristics and naming conventions of substations, and construct a defect prediction rule library based on high-frequency defect characteristics.

[0038] S200: Based on a scenario-based adaptation rule base, an adaptive template library is built. Through scenario-based rule mapping, the templates are automatically adapted to different engineering characteristics. The system integrates equipment ICD files, historical records, and engineering requirement data, and uses the adaptive template library to generate standardized substation monitoring information that conforms to naming conventions and engineering characteristics.

[0039] S300 performs unified field format processing on the generated standardized monitoring information and preset standard templates, aligns the core comparison dimensions, including signal identification, voltage level and parameter type; and cleans up redundant data and abnormal information.

[0040] S400. For structured fields including signal identifiers, voltage levels, and parameter types, an exact matching algorithm is used for comparison. The exact matching algorithm matches fields where the content and format of the field values ​​are completely equal. Based on the comparison results, the monitoring information is divided into three categories: exact match, obvious mismatch, and pending semantic matching. If all structured fields meet the exact matching criteria, it is determined to be an exact match. If there is a substantial difference in the content of the field value of any structured field, it is determined to be an obvious mismatch. If the structured fields have the same format and no substantial difference in the content of the field values, the core values ​​are the same, and only the description details are different, and not all of them meet the exact matching criteria, it is determined to be pending semantic matching. The information pending semantic matching is obtained by filtering.

[0041] The S500 uses a mini version of Sentence-BERT as the basic semantic model. It uses data from the power signal synonym mapping library as training samples to fine-tune the basic semantic model and obtain a dedicated semantic model adapted to the power scenario. The semantic matching information to be matched is input into the dedicated semantic model to calculate the semantic similarity. The power signal synonym mapping library is used for auxiliary verification, and the semantic matching result is output according to the preset judgment criteria.

[0042] S600: Based on the knowledge graph of the core elements of the substation, query the equipment ownership relationship and bay association logic corresponding to the monitoring information to be verified; compare with the association logic standard defined in the scenario-based adaptation rule base, verify the compliance of the association relationship of the monitoring information, and output the association relationship verification result.

[0043] S700 summarizes the results of structured data matching, semantically enhanced deep matching, and compliance verification of relationships, and generates a complete report containing verification conclusions and specific reasons for failure; for monitoring information with defects, it pushes targeted correction suggestions based on the defect prediction rule base.

[0044] In some embodiments, the node association logic of the substation core element directed association knowledge graph includes the affiliation relationship between signals and equipment, the membership relationship between equipment and bays, and the correspondence relationship between signals and parameters. The knowledge graph is deployed through an open-source graph database and only stores the necessary association data for verification. When querying association relationships based on the knowledge graph, Cypher query statements are used to achieve fast retrieval. The association logic standards defined in the scenario-based adaptation rule base are that the same signal can only belong to one equipment, the same equipment can only belong to one bay, and the signal type matches the associated equipment type.

[0045] In a specific embodiment, when constructing a synonym mapping library for power signals, the naming of the same signal by different manufacturers is obtained from equipment manufacturer manuals. Taking 10kV circuit breaker signals as an example, manufacturer A calls it the 10kV circuit breaker open position signal, manufacturer B calls it the 10kV switch open position signal, and manufacturer C calls it the 10kV CB open position. Simplified names manually entered by maintenance personnel are obtained from historical monitoring system data, such as 10kV disconnection / open signal. The standard name 10kV circuit breaker open status signal is obtained from industry standard documents. All names related to the 10kV circuit breaker open status are extracted from the above data, deduplicated, and an initial candidate set is obtained. This set is then manually verified by experts and semantically assisted in verification. The semantically assisted verification is performed by calculating the semantic similarity between names using the Sentence-BERT model in this scheme. Then, using the industry standard name as the base word and the remaining names as synonym variants, a one-to-many mapping relationship is established.

[0046] In a specific embodiment, when constructing a knowledge graph of the core elements of a substation, the equipment list for 10kV bay 1 is obtained from the equipment ledger, including 10kV circuit breaker QF101, 10kV disconnector QS101, and 10kV current transformer TA101, etc.; the association between equipment and signals is obtained from the SCD file, such as the association between 10kV circuit breaker QF101 and the opening status signal, closing status signal, and overcurrent protection action signal; the attribution relationship between bay and equipment is obtained from the wiring diagram, where 10kV circuit breaker QF101 belongs to 10kV bay 1; and key parameters of the equipment are obtained from the parameter table, such as the rated current of 10kV circuit breaker QF101: 630A, and the opening coil voltage: DC220V. Four types of nodes—signals, equipment, bays, and parameters—are extracted from the data, and a unified naming format is adopted, such as the equipment node specification being [voltage level] [equipment type] [number]. Association relationships are defined, where the attribution relationship is between equipment and bay, the correspondence relationship is between signal and equipment, and the attribute relationship is between parameter and equipment. Use Neo4j to build a knowledge graph and store nodes and relationships in the form of triples, such as (10kV circuit breaker QF101, belonging to, 10kV bay 1).

[0047] In a specific implementation, when constructing a scenario-based adaptation rule base, the signal naming requirements for different project types in industry standards, as well as signal templates for newly built substations and renovated / expanded substations, are obtained. Substation projects are divided into different scenarios such as new construction and renovation / expansion. For each scenario, signal naming, field composition, and format requirements are extracted from the data. For example, in the new construction scenario, the mandatory naming is [voltage level][equipment type][number]_[status]. In the renovation / expansion scenario, existing signals retain their original names, and new signals must be marked with "-expansion". The rules are transformed into executable logic using regular expressions, conditional statements, and other methods.

[0048] In a specific implementation, when constructing the defect prediction rule base, the system retrieves monitoring signal defect reports from 10 substations over the past three years from historical defect records. These reports may include issues such as signal names lacking voltage levels leading to misjudgments, incorrect signal-equipment associations, and parameter values ​​exceeding equipment ratings. The system also retrieves common problems recorded by maintenance personnel from maintenance logs. The retrieved defects are categorized into three types: naming defects, association defects, and parameter defects. For each type of defect, judgment conditions are summarized based on the scenario-specific adaptation rule base. For example, a naming defect is defined as a signal name not containing fields required by the scenario-specific rule, such as a newly created scenario signal lacking a voltage level; an association defect is defined as a signal without a corresponding device in the knowledge graph, or an incorrect device and bay assignment, such as the QF101 trip signal being associated with 10kV bay 2; and a parameter defect is defined as parameter values ​​exceeding the equipment rating range. A judgment threshold is set for each rule.

[0049] In some embodiments, the template fields of the adaptive template library include signal identifier, signal name, signal type, associated device, interval number, voltage level, parameter range and functional description. The template fields can be dynamically added or removed according to the engineering characteristics. Different engineering characteristics include new construction projects, renovation and expansion projects and conventional station intelligent transformation projects. The scenario-based rule mapping automatically calls the corresponding template configuration by matching the engineering type.

[0050] In a specific embodiment, an adaptive template library is first constructed, with the signal point table template as the core. This library includes basic fields and scenario-specific extended fields. The fields are sourced to fully match the requirements of the scenario-specific adaptation rule library, ensuring the template can be dynamically adjusted. The basic fields are derived from industry-standard specifications and include: signal identifier, standard signal name, signal type (remote signaling / telemetry / remote control / remote adjustment), associated device number, interval number, voltage level, parameter range, and functional description. These are applicable to all scenarios. The scenario-specific extended fields, sourced from the scenario-specific adaptation rule library, can be optionally enabled based on different project characteristics. These include: expansion identifier (only for expansion / remodeling scenarios), existing signal ID (only for expansion / remodeling scenarios), and construction section (only for scenarios involving the creation of multiple construction sections).

[0051] In this embodiment, the core of scenario-based rule mapping is the identification of engineering characteristics, rule matching, and dynamic enabling / disabling of template fields, which relies entirely on the logical execution of the scenario-based adaptation rule base.

[0052] In some embodiments, step S200, integrating multi-source data, includes extracting device number, original signal name, and parameter value from the ICD file; extracting the correlation of existing signals from the historical ledger; extracting scenario parameters (such as section number, new bay number) from engineering requirements; removing redundant signals (such as manufacturer internal test signals) from the ICD file; supplementing missing voltage level fields in the historical ledger; correcting ambiguous parameter ranges in engineering requirements, such as correcting a current of approximately 600A to 630A; in the renovation and expansion scenario, the name of the original 10kV disconnector switch conflicts with the standard format, so the historical name is retained according to the rules (engineering requirements require that the original signal not be modified), and only the standard signal name field is added: 10kV_disconnector_QS101_closing; mapping the cleaned multi-source data one by one to the fields of the adapted template and filling in the corresponding content, such as mapping the opening status of the ICD file to the standard signal name of the template, and filling the section 1 of the engineering requirements into the construction section field of the template.

[0053] This embodiment avoids creating separate templates for different scenarios by using an adaptive template library with dynamically adjustable fields, thus reducing maintenance costs. The automatic adaptation of the adaptive template library is achieved through scenario-based rule mapping. By identifying the project type, the corresponding naming rules and field rules are directly called without manual intervention. The multi-source data integration in this embodiment ensures that the data conforms to equipment standards (ICD files) and historical continuity (historical ledgers). The standardized information generated in the end not only conforms to industry standards but also fits the specific characteristics of the project (newly built unified standards, modified and expanded compatible with history), and can be directly imported into the monitoring system for use, completely replacing manual compilation.

[0054] In some embodiments, cleaning redundant data and abnormal information includes removing duplicate signal records, completing missing core fields, removing parameter values ​​that are outside the reasonable range, and correcting field content with format errors. The core fields are signal identifier, associated device, interval number, and voltage level.

[0055] In one specific embodiment, the implementation of the exact matching algorithm includes: field-level exact matching, with the matching order prioritized by signal identifier, voltage level, and parameter type. The signal identifier is a unique identifier, and prioritizing matching allows for rapid filtering of most results. For signal identifier matching, a hash table lookup is used. The signal identifiers of the standard template are stored in the hash table, and the identifier of the signal to be verified is directly queried from the hash table; if it exists, a preliminary match is passed. For voltage level / parameter type matching, a string equality comparison is used. Only when all three fields satisfy the condition of string equality is it considered an exact match. The output results are categorized into three types: exact match, obvious mismatch, and pending semantic match. An exact match occurs when the signal identifier exists in the hash table and the voltage level and parameter type strings are completely identical. An obvious mismatch occurs when any field is inconsistent with the standard field (e.g., the signal identifier is not in the hash table, or the voltage level 10kV ≠ 35kV). A pending semantic match occurs when the three fields are substantially similar (i.e., the core values ​​are consistent but not completely equal, such as the signal identifier "10kV-J1-QF101-Open" vs. the standard "10kV-J1-QF101-Open Status"). The core values ​​represent the core focus of the power industry for structured fields.

[0056] In some embodiments, when fine-tuning the basic semantic model, the underlying parameters are frozen and only the top-level classification head is fine-tuned. The preset judgment criteria are: when the semantic similarity is greater than or equal to the first threshold, it is judged as semantic matching passed; when the semantic similarity is less than the first threshold but greater than or equal to the second threshold, it is marked as pending manual review; and when the semantic similarity is less than the second threshold, it is judged as semantic mismatch.

[0057] In some embodiments, the logic for auxiliary verification using a power signal synonym mapping library is as follows: if the name combination of the information to be semantically matched has a clear association in the synonym mapping library, the semantic matching result is directly confirmed; when there is no corresponding association in the synonym mapping library, the calculation result of the dedicated semantic model shall prevail.

[0058] In a specific embodiment, step S500 includes: combining base words and synonym variants from the synonym mapping library into training sample pairs, constructing a total of 5000 sets of power scenario-specific sample pairs as training data; and using a method of freezing the bottom semantic encoding layer and only fine-tuning the top classification head to allow the model to learn the semantic association rules of power signal naming, such as circuit breaker and switch, and open state and open position are synonymous associations. The resulting dedicated semantic model can accurately identify synonymous expressions of signal names in power scenarios and has anti-interference capability against semantic interference from non-power scenarios.

[0059] In a specific embodiment, step S500 includes: simultaneously inputting the semantic matching information to be verified, namely the signal name "10kV switch occupancy signal" and the standard template signal "10kV circuit breaker tripping status signal," into a dedicated semantic model. The model outputs a semantic similarity greater than a first threshold. Then, auxiliary verification is performed using a power signal synonym mapping library. The verification logic involves querying whether the signal name "10kV switch occupancy signal" exists in the synonym mapping library. The verification result indicates that the name is a synonym variant of "10kV circuit breaker tripping status signal," indicating a clear association. This directly confirms the validity of the semantic matching result without relying on model calculations for separate determination. Finally, it is determined that the semantic similarity is greater than the first threshold, and a clear association exists in the synonym mapping library; therefore, the match passes.

[0060] In another specific embodiment, step S500 includes: the name of the signal to be verified is "10kV CB open", and the standard template signal is "10kV circuit breaker tripping status signal". Through similarity calculation, the model identifies "CB" as a general abbreviation for "circuit breaker" in the power scenario and "open" as "tripping status", outputting a semantic similarity of less than a first threshold and greater than a second threshold. Assisted verification is performed using a power signal synonym mapping library. The system queries whether "10kV CB open" exists in the library. The result is that "10kV CB open position" (with the "position" suffix) exists, but "10kVCB open" does not, indicating no direct association. Therefore, the output result is: semantic similarity less than the first threshold and greater than the second threshold, and no direct association in the synonym mapping library, requiring manual review. In this embodiment, the power signal synonym mapping library-assisted verification can compensate for occasional model misjudgments, such as signals with similar names but different functions. Dual verification improves the reliability of the results.

[0061] In a specific embodiment, the association logic standards in the scenario-based adaptation rule base in step S600 include: the same signal can only belong to one device, and the signal type must match the device function; the device can only belong to one bay, and the bay associated with the signal must be consistent with the bay to which the device belongs; signals without a belonging device or a belonging bay are determined to have non-compliant association relationships. Step S600 includes: firstly, obtaining the monitoring information to be verified, including signal 1 to be verified: signal identifier "10kV-J1-QF101-01", signal name "10kV_circuit breaker_QF101_open", associated device "10kV circuit breaker QF101", associated bay "10kV bay 1"; signal 2 to be verified: signal identifier "10kV-J2-QS102-02", signal name "10kV_disconnecting switch_QS102_close", associated device "10kV circuit breaker QF101", associated bay "10kV bay 1".

[0062] In this embodiment, relying on the knowledge graph to query the association relationships, the system queries the device affiliation and interval association logic corresponding to the signal to be verified, using the signal name as the search keyword. It matches the corresponding affiliation device node from the knowledge graph, and then matches the affiliated interval node through the device node. The query results are: signal 1 belongs to 10kV circuit breaker QF101, which is consistent with the associated device marked in the information to be verified; device 10kV circuit breaker QF101 belongs to 10kV interval 1, which is consistent with the associated interval marked in the information to be verified. Then, compliance is verified by comparing with the scenario-based adaptation rule base: signal 1 belongs to only one device, and the signal type matches the device function, which conforms to the rule; device QF101 belongs only to 10kV interval 1, and the interval associated with signal 1 is consistent with the interval to which the device belongs, which conforms to the rule; signal 1 has a clearly defined affiliation device and a affiliated interval, which conforms to the rule. Finally, the association relationship is determined to be compliant, and the association verification is output as passed.

[0063] In this embodiment, relying on the knowledge graph to query the association relationship, the device affiliation relationship and interval association logic corresponding to the signal to be verified 2 are queried. Using the signal name as the search keyword, the corresponding affiliation device node is matched from the knowledge graph, and then the affiliation interval node is matched through the device node. The query results are as follows: signal 2 belongs to 10kV disconnector switch QS102, while the associated device marked in the information to be verified is 10kV circuit breaker QF101, which are inconsistent; device 10kV disconnector switch QS102 belongs to 10kV interval 2, while the associated interval marked in the information to be verified is 10kV interval 1, which are also inconsistent. The compliance was then verified by comparing the information with the scenario-based adaptation rule base: Signal 2's standard assigned device is QS102, but the information to be verified is labeled as QF101. One signal corresponds to two devices, and the signal type does not match the function of QF101, violating the rule. Device QS102 belongs to 10kV bay 2, but the information to be verified is labeled as associated with 10kV bay 1, causing a conflict between the signal's associated bay and the device's assigned bay, violating the rule. Signal 2 has an assigned device and bay, but the labeled information conflicts with the knowledge graph's standard association relationship, violating the core logic. The final judgment: the association relationship is non-compliant, the association verification fails, and the reason for the violation is recorded: the signal's assigned device labeling is incorrect, leading to a conflict in the bay association logic.

[0064] Reference Figure 2When performing semantic enhancement deep matching in step S500 and compliance verification of association in step S600, a two-way collaborative mechanism between semantic knowledge graph and deep learning model is adopted. The semantic similarity result output by the dedicated semantic model in step S500 is used as a pre-screening condition for the association query of knowledge graph. The association verification in step S600 is only performed on monitoring information that passes semantic matching or is pending review. At the same time, the association logic result confirmed by the knowledge graph in step S600 is fed back to the dedicated semantic model to dynamically optimize the semantic similarity calculation weight of closely related signals.

[0065] Reference Figure 3 After generating standardized substation monitoring information in step S200 and before performing pre-verification data preprocessing in step S300, the method of this embodiment of the invention further includes a machine learning prediction-assisted optimization step, which includes using a machine learning model to train the initially generated monitoring information and historical verification data, predicting potential defects, including signal naming ambiguity, abnormal correlation and non-compliant parameter configuration, marking high-risk signals and filtering obvious defects in advance.

[0066] Reference Figure 4 After the targeted correction suggestions are output in step S700, cross-version linkage editing is performed. High-risk defect signals predicted by machine learning are displayed first, and compliant modification directions are automatically recommended in combination with historical modification records. After the editing is completed, the prediction results and correction data are fed back to the machine learning model in a synchronous manner.

[0067] Reference Figure 5 This invention proposes an automatic generation and semantic enhancement verification system for substation monitoring information, comprising:

[0068] The basic resource construction module 100 is used to obtain the synonym mapping relationship of power signals and generate a power signal synonym mapping library; construct a knowledge graph of the core elements of the substation with signals, equipment, bays and parameters as core nodes; obtain a scenario-based adaptation rule library according to the engineering characteristics and naming conventions of the substation, and construct a defect prediction rule library based on high-frequency defect characteristics.

[0069] The monitoring information generation module 200 is used to build an adaptive template library based on a scenario-based adaptation rule library. Through scenario-based rule mapping, the templates are automatically adapted to different engineering characteristics. It integrates equipment ICD files, historical records and engineering requirement data, and uses the adaptive template library to generate standardized substation monitoring information that conforms to naming conventions and engineering characteristics.

[0070] The data preprocessing module 300 is used to unify the field format of the generated standardized monitoring information and the preset standard template, align the core comparison dimensions, including signal identification, voltage level and parameter type; and clean up redundant data and abnormal information.

[0071] The structured data matching module 400 is used to compare structured fields including signal identifiers, voltage levels, and parameter types using a structured matching algorithm. Based on the comparison results, the monitoring information is divided into three categories: complete match, obvious mismatch, and semantic matching to be determined, thus obtaining semantic matching information that needs further verification.

[0072] The semantic enhancement matching module 500 is used to select the Sentence-BERT model with simplified and optimized parameters as the basic semantic model, use data from the power signal synonym mapping library as training samples, fine-tune the basic semantic model in a targeted manner to obtain a dedicated semantic model adapted to the power scenario; input the information to be semantically matched into the dedicated semantic model to calculate the semantic similarity; combine the power signal synonym mapping library for auxiliary verification, and output the semantic matching result according to the preset judgment criteria.

[0073] The association verification module 600 is used to query the equipment affiliation and bay association logic corresponding to the monitoring information to be verified based on the knowledge graph of the core elements of the substation; it verifies the compliance of the association relationship of the monitoring information by comparing it with the association logic standard defined in the scenario-based adaptation rule base, and outputs the association verification result.

[0074] The inspection results module 700 is used to summarize the results of structured data matching, semantically enhanced deep matching, and compliance verification of related relationships, and generate a complete report that includes the verification conclusion and the specific reasons for failure; for monitoring information with defects, it pushes targeted correction suggestions based on the defect prediction rule base.

[0075] Although specific embodiments are described herein, those skilled in the art will recognize that many other modifications or alternative embodiments are also within the scope of this disclosure. For example, any of the functions and / or processing capabilities described in connection with a particular device or component can be performed by any other device or component. Furthermore, while various exemplary embodiments and architectures have been described according to embodiments of this disclosure, those skilled in the art will recognize that many other modifications to the exemplary embodiments and architectures described herein are also within the scope of this disclosure.

[0076] The foregoing description, with reference to block diagrams and flowcharts of systems, methods, systems, and / or computer program products according to exemplary embodiments, has described certain aspects of this disclosure. It should be understood that one or more blocks in the block diagrams and flowcharts, as well as combinations of blocks in the block diagrams and flowcharts, can be implemented by executing computer-executable program instructions, respectively. Similarly, according to some embodiments, some blocks in the block diagrams and flowcharts may not need to be executed in the order shown, or may not all need to be executed. Furthermore, additional components and / or operations beyond those shown in the blocks in the block diagrams and flowcharts may exist in some embodiments.

[0077] Therefore, blocks in block diagrams and flowcharts support combinations of means for performing a specified function, combinations of elements or steps for performing a specified function, and program instruction means for performing a specified function. It should also be understood that each block in a block diagram and flowchart, and combinations of blocks in block diagrams and flowcharts, can be implemented by a dedicated hardware computer system or a combination of dedicated hardware and computer instructions that performs a specific function, element, or step.

[0078] The program modules, applications, etc., described herein may include one or more software components, including, for example, software objects, methods, data structures, etc. Each such software component may include computer-executable instructions that, in response to execution, cause at least a portion of the functionality described herein (e.g., one or more operations of the exemplary methods described herein) to be performed.

[0079] Software components can be coded using any of a variety of programming languages. An exemplary programming language could be a low-level programming language, such as assembly language associated with a specific hardware architecture and / or operating system platform. Software components including assembly language instructions may need to be converted into executable machine code by an assembler before being executed by the hardware architecture and / or platform. Another exemplary programming language could be a higher-level programming language that is portable across multiple architectures. Software components including higher-level programming languages ​​may need to be converted into an intermediate representation by an interpreter or compiler before execution. Other examples of programming languages ​​include, but are not limited to, macro languages, shell or command languages, job control languages, scripting languages, database query or search languages, or report writing languages. In one or more exemplary embodiments, a software component containing instructions from one of the above-described programming language examples can be executed directly by the operating system or other software components without first being converted into another form.

[0080] Software components can be stored as files or other data storage structures. Software components of similar type or related function can be stored together in a specific directory, folder, or library. Software components can be static (e.g., pre-defined or fixed) or dynamic (e.g., created or modified at runtime).

[0081] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the above embodiments. Within the scope of knowledge possessed by those skilled in the art, various changes can be made without departing from the spirit of the present invention.

Claims

1. A method for automatically generating and semantically enhancing substation monitoring information, characterized in that, Includes the following steps: S100. Obtain the synonym mapping relationship of power signals and generate a power signal synonym mapping library; construct a knowledge graph of the core elements of the substation, with signals, equipment, bays and parameters as core nodes; obtain a scenario-based adaptation rule library according to the engineering characteristics and naming conventions of the substation, and construct a defect prediction rule library based on high-frequency defect characteristics. S200: Based on a scenario-based adaptation rule base, an adaptive template library is built. Through scenario-based rule mapping, the templates are automatically adapted to different engineering characteristics. The system integrates equipment ICD files, historical records, and engineering requirement data, and uses the adaptive template library to generate standardized substation monitoring information that conforms to naming conventions and engineering characteristics. S300 performs field format unification processing on the generated standardized monitoring information and preset standard templates, aligns core comparison dimensions, including signal identification, voltage level and parameter type; cleans redundant data and abnormal information; S400. For structured fields including signal identifiers, voltage levels, and parameter types, a precise matching algorithm is used for comparison. The precise matching algorithm matches all structured fields whose field values ​​and formats are completely equal. Based on the comparison results, the monitoring information is divided into three categories: complete match, obvious mismatch, and pending semantic match. If all structured fields meet the precise matching criteria, it is determined to be a complete match. If any structured field has a substantial difference in its field value, it is determined to be an obvious mismatch. If all structured fields have the same format and no substantial difference in their field values, with the core values ​​being the same but only the description details differing, and not all of them meet the precise matching criteria, it is determined to be pending semantic match. Information pending semantic match is then filtered out. S500. Select the parameter-simplified and optimized Sentence-BERT model as the basic semantic model, use the data in the power signal synonym mapping library as training samples, and fine-tune the basic semantic model in a targeted manner to obtain a special semantic model adapted to the power scenario; input the semantic matching information to be matched into the special semantic model and calculate the semantic similarity; combine the power signal synonym mapping library for auxiliary verification, and output the semantic matching result according to the preset judgment criteria. S600: Based on the knowledge graph of the core elements of the substation, query the equipment ownership relationship and bay association logic corresponding to the monitoring information to be verified; By comparing the correlation logic standards defined in the scenario-based adaptation rule base, the compliance of the correlation relationships of monitoring information is verified, and the correlation relationship verification results are output. S700 summarizes the results of structured data matching, semantically enhanced deep matching, and compliance verification of relationships, and generates a complete report containing verification conclusions and specific reasons for failure; for monitoring information with defects, it pushes targeted correction suggestions based on the defect prediction rule base.

2. The method for automatic generation and semantic enhancement verification of substation monitoring information according to claim 1, characterized in that, The node association logic of the substation core element directed association knowledge graph includes the affiliation relationship between signals and equipment, the membership relationship between equipment and bays, and the correspondence relationship between signals and parameters. The knowledge graph is deployed through an open-source graph database and only stores the necessary association data for verification. When querying association relationships based on the knowledge graph, Cypher query statements are used to achieve fast retrieval. The association logic standards defined in the scenario-based adaptation rule base are that the same signal can only belong to one device, the same device can only belong to one bay, and the signal type matches the associated device type.

3. The method for automatic generation and semantic enhancement verification of substation monitoring information according to claim 1, characterized in that, When fine-tuning the basic semantic model, the underlying parameters are frozen and only the top-level classification head is fine-tuned. The preset judgment criteria are: when the semantic similarity is greater than or equal to the first threshold, it is judged as semantic matching passed; when the semantic similarity is less than the first threshold but greater than or equal to the second threshold, it is marked as pending manual review; and when the semantic similarity is less than the second threshold, it is judged as semantic mismatch.

4. The method for automatic generation and semantic enhancement verification of substation monitoring information according to claim 1, characterized in that, The logic for auxiliary verification using the power signal synonym mapping library is as follows: if the name combination of the information to be semantically matched has a clear association in the synonym mapping library, the semantic matching result is directly confirmed; when there is no corresponding association in the synonym mapping library, the calculation result of the dedicated semantic model shall prevail.

5. The method for automatic generation and semantic enhancement verification of substation monitoring information according to claim 1, characterized in that, The template fields of the adaptive template library include signal identifier, signal name, signal type, associated device, interval number, voltage level, parameter range, and functional description. The template fields can be dynamically added or removed according to the engineering characteristics. The different engineering characteristics include new construction projects, renovation and expansion projects, and intelligent transformation projects of conventional stations. The scenario-based rule mapping automatically calls the corresponding template configuration by matching the engineering type.

6. The method for automatic generation and semantic enhancement verification of substation monitoring information according to claim 1, characterized in that, The process of cleaning redundant data and abnormal information includes removing duplicate signal records, completing missing core fields, removing parameter values ​​that are outside the reasonable range, and correcting field content with format errors. The core fields are signal identifier, associated device, interval number, and voltage level.

7. The method for automatic generation and semantic enhancement verification of substation monitoring information according to claim 1, characterized in that, When performing semantic enhancement deep matching in step S500 and compliance verification of association in step S600, a two-way collaborative mechanism between semantic knowledge graph and deep learning model is adopted. The semantic similarity result output by the dedicated semantic model in step S500 is used as a pre-screening condition for the association query of knowledge graph. The association verification in step S600 is only performed on monitoring information that passes semantic matching or is pending review. At the same time, the association logic result confirmed by the knowledge graph in step S600 is fed back to the dedicated semantic model to dynamically optimize the semantic similarity calculation weight of the model for closely related signals.

8. The method for automatic generation and semantic enhancement verification of substation monitoring information according to claim 1, characterized in that, After generating standardized substation monitoring information in step S200 and before performing pre-verification data preprocessing in step S300, the method further includes a machine learning prediction-assisted optimization step, which includes using a machine learning model to train the initially generated monitoring information and historical verification data to predict potential defects. The potential defects include ambiguous signal naming, abnormal correlation, and non-compliant parameter configuration. High-risk signals are marked and obvious defects are filtered out in advance.

9. The method for automatic generation and semantic enhancement verification of substation monitoring information according to claim 8, characterized in that, After the targeted correction suggestions are output in step S700, cross-version collaborative editing is performed. High-risk defect signals predicted by machine learning are displayed first, and compliant modification directions are automatically recommended in combination with historical modification records. After the editing is completed, the prediction results and correction data are fed back to the machine learning model in a synchronous manner.

10. A substation monitoring information automatic generation and semantic enhancement verification system, used to perform the method as described in any one of claims 1 to 9, characterized in that, include: The basic resource construction module is used to obtain the synonym mapping relationship of power signals and generate a power signal synonym mapping library; construct a knowledge graph of the core elements of the substation, with signals, equipment, bays and parameters as core nodes; obtain a scenario-based adaptation rule library according to the engineering characteristics and naming specifications of the substation, and construct a defect prediction rule library based on high-frequency defect characteristics. The monitoring information generation module is used to build an adaptive template library based on a scenario-based adaptation rule base. It achieves automatic adaptation of templates to different engineering characteristics through scenario-based rule mapping. It integrates equipment ICD files, historical ledgers and engineering requirement data, and uses the adaptive template library to generate standardized substation monitoring information that conforms to naming conventions and engineering characteristics. The data preprocessing module is used to unify the field format of the generated standardized monitoring information with the preset standard template, align the core comparison dimensions, including signal identification, voltage level and parameter type; and clean up redundant data and abnormal information. The structured data matching module is used to compare structured fields, including signal identifiers, voltage levels, and parameter types, using an exact matching algorithm. This exact matching algorithm ensures that the field values ​​and formats of all structured fields are completely identical. Based on the comparison results, the monitoring information is divided into three categories: complete match, obvious mismatch, and pending semantic matching. If all structured fields satisfy the exact matching criteria, it is determined to be a complete match. If any structured field has a substantial difference in its field value, it is determined to be an obvious mismatch. If all structured fields have the same format and no substantial differences in their field values, with consistent core values ​​but differences only in the details of their descriptions, and do not all satisfy the exact matching criteria, it is determined to be pending semantic matching. Information pending semantic matching is then filtered out. The semantic enhancement matching module is used to select the Sentence-BERT model with simplified and optimized parameters as the basic semantic model, use data from the power signal synonym mapping library as training samples, fine-tune the basic semantic model in a targeted manner to obtain a dedicated semantic model adapted to the power scenario; input the information to be semantically matched into the dedicated semantic model, calculate the semantic similarity; combine the power signal synonym mapping library for auxiliary verification, and output the semantic matching result according to the preset judgment criteria. The association verification module is used to query the equipment ownership relationship and bay association logic corresponding to the monitoring information to be verified by relying on the knowledge graph of the core elements of the substation to query the core elements of the substation; By comparing the correlation logic standards defined in the scenario-based adaptation rule base, the compliance of the correlation relationships of monitoring information is verified, and the correlation relationship verification results are output. The inspection results module is used to summarize the results of structured data matching, semantically enhanced deep matching, and compliance verification of relationships, and generate a complete report that includes the verification conclusion and the specific reasons for failure; for monitoring information with defects, it pushes targeted correction suggestions based on the defect prediction rule base.

Images