Power signal association method and system

Abstract

The invention is suitable for the technical field of smart power grids, and provides a power signal association method and system, and the method comprises the steps: obtaining a first signal set as a to-be-matched signal and a second signal set as a candidate signal; extracting a first key field used for rule matching and a first signal description text used for semantic matching from the first signal set, and extracting a second key field used for rule matching and a second signal description text used for semantic matching from the second signal set; for any to-be-matched signal in the first signal set, based on the first key field, the first signal description text, the second key field and the second signal description text, executing a progressive matching operation including rule matching, semantic matching and association constraint so as to determine a target matching signal from the second signal set; and generating a telecontrol configuration description file according to the target matching signal. Through the progressive strategy, complex conditions such as non-standard signal description can be effectively handled, and the matching accuracy and robustness are improved.

Description

Technical Field

[0001] This application belongs to the field of smart grid technology, and in particular relates to power signal correlation methods and systems. Background Technology

[0002] In the construction of smart grids, remote monitoring and automated control of substations are crucial. The telecontrol configuration description file (TCM) is the core of this functionality, recording the precise correspondence between the massive amounts of equipment monitoring signals within the substation and the monitoring system of the dispatch center. However, in existing technologies, the generation process of this file faces numerous challenges.

[0003] Currently, the generation of remote control configuration description files mainly relies on manual configuration by professional engineers. Engineers need to manually search for and establish the correspondence between signals one by one, based on the dispatch point table issued by the dispatch center and the local equipment information files of the substations. This process is not only tedious and time-consuming, but also suffers from significant arbitrariness in signal naming due to the lack of unified substation signal description standards. Synonyms, abbreviations, and reversed word order are common, making manual configuration prone to mismatches and omissions, seriously affecting configuration quality and the safety of power grid operation.

[0004] To address the drawbacks of manual configuration, some automated methods based on fixed keywords or naming rules have emerged. These methods compare specific fields in signal names using a pre-defined rule base, improving configuration efficiency to some extent. However, these methods have poor generalization ability, rigid matching logic, and cannot effectively handle complex textual description differences and deep semantic relationships. When encountering non-standard descriptions or new naming habits, their matching accuracy drops sharply, ultimately requiring significant manual intervention for proofreading and correction, failing to fundamentally solve the problem. Summary of the Invention

[0005] This application provides a method and system for power signal association, aiming to solve the technical problems in the prior art where power signal association mainly relies on manual or simple rule matching, resulting in low efficiency, low accuracy, and difficulty in adapting to non-standardized signal descriptions.

[0006] In a first aspect, embodiments of this application provide a power signal association method, including: Obtain a first set of signals as signals to be matched and a second set of signals as candidate signals; A first key field for rule matching and a first signal description text for semantic matching are extracted from the first signal set; and a second key field for rule matching and a second signal description text for semantic matching are extracted from the second signal set. For any signal to be matched in the first signal set, a progressive matching operation including rule matching, semantic matching and association constraints is performed based on the first key field, the first signal description text, the second key field and the second signal description text to determine the target matching signal from the second signal set. Based on the target matching signal, a telemetry configuration description file is generated.

[0007] In one possible implementation of the first aspect, obtaining the first signal set as the signal to be matched and the second signal set as the candidate signal includes: Load and parse the scheduling point table file, and extract signal entries to form the first signal set; The system configuration file is loaded and parsed to extract data object information to form the second signal set.

[0008] In one possible implementation of the first aspect, the progressive matching operation includes the following performed sequentially: Rule matching: Determine whether the first key field of the signal to be matched is complete; if complete, then based on the first key field and the second key field, filter out a subset of candidate signals from the second signal set; if incomplete, then use the entire second signal set as the subset of candidate signals. Semantic matching: Using a preset semantic representation model, calculate the semantic similarity score between the first signal description text of the signal to be matched and the second signal description text of each signal in the candidate signal subset; Association constraint: Based on the preset association relationship model, the reasonableness of the association between the signal to be matched and the candidate signal whose semantic similarity score is higher than the preset semantic threshold is judged, and the semantic similarity score is adjusted based on the association reasonableness judgment result. The candidate signal with the highest adjusted score is determined as the target matching signal.

[0009] In one possible implementation of the first aspect, the step of filtering candidate signal subsets from the second signal set based on the first key field and the second key field includes: The first key field of the signal to be matched is sequentially compared with the corresponding second key field of each signal in the second signal set using a whole-word comparison. The set of candidate signals that pass the full word comparison in the second signal set is taken as the candidate signal subset.

[0010] In one possible implementation of the first aspect, the step of extracting a first key field for rule matching and a first signal description text for semantic matching from the first signal set, and extracting a second key field for rule matching and a second signal description text for semantic matching from the second signal set, includes: The first original description text is extracted from the first signal set; Extract the first substring from the first original description text as the first key field; The remaining part after removing the first substring that has been extracted as the first key field from the first original description text is used as the first signal description text. The second original description text is extracted from the second signal set; Extract the second substring from the second original description text as the second key field; The remaining portion of the second original description text after removing the extracted second substring which is the second key field is used as the second signal description text.

[0011] In one possible implementation of the first aspect, the process of constructing the association model includes: The system configuration file is parsed to identify primary equipment topology association data that defines the electrical connection relationships and electrical bay divisions of primary equipment, as well as secondary equipment function definition data that defines the logical functions of secondary equipment. Traverse the topology information in the primary equipment topology association data to construct the hierarchical relationship between electrical bays and the primary equipment they contain, and clarify the electrical bay to which each primary equipment belongs; Traverse the secondary device function definition data to establish a mapping relationship between the secondary devices and their logical functions and the target devices; wherein, the target devices include the primary devices being measured, controlled or protected; By integrating the hierarchical relationships and the mapping relationships, an electrical topology association model that reflects the inherent relationship between the logical function of any of the secondary devices and the electrical bay to which they belong is generated, which serves as the association relationship model.

[0012] In one possible implementation of the first aspect, the result of the association rationality judgment includes an unreasonable judgment result and a reasonable judgment result; the association constraint includes: If the result of the association rationality judgment is the result of the unreasonable judgment, then the semantic similarity score is multiplied by a preset penalty factor less than 1; If the result of the association rationality judgment is the reasonable judgment result, then the semantic similarity score remains unchanged.

[0013] In one possible implementation of the first aspect, the method further includes: Receive user feedback information regarding the target matching signal; Labeled training samples are generated based on the feedback information, and the semantic representation model is incrementally trained using the training samples to optimize the model.

[0014] In one possible implementation of the first aspect, generating labeled training samples based on the feedback information and using the training samples to incrementally train the semantic representation model to optimize the model includes: When generating training samples, a weight factor corresponding to the signal type is assigned to the training samples; wherein, the signal types are pre-divided into multiple importance levels, and different importance levels of signal types correspond to different weight factors. During the incremental training of the semantic representation model, a weighted loss function is used to optimize the semantic representation model; wherein the weighted loss function is constructed based on the weight factors assigned in the training samples.

[0015] Secondly, embodiments of this application provide a power signal correlation system, including: The acquisition module is used to acquire a first set of signals as signals to be matched and a second set of signals as candidate signals; An extraction module is configured to extract a first key field for rule matching and a first signal description text for semantic matching from the first signal set, and to extract a second key field for rule matching and a second signal description text for semantic matching from the second signal set. The progressive matching module is used to perform a progressive matching operation, including rule matching, semantic matching and association constraints, on any signal to be matched in the first signal set, based on the first key field, the first signal description text, the second key field and the second signal description text, so as to determine the target matching signal from the second signal set. The generation module is used to generate a telemetry configuration description file based on the target matching signal.

[0016] Thirdly, embodiments of this application provide a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method as described in any of the first aspects.

[0017] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the method as described in any of the first aspects.

[0018] Fifthly, embodiments of this application provide a computer program product that, when run on a computer device, causes the computer device to perform the method described in any of the first aspects above.

[0019] The beneficial effects of the embodiments in this application compared with the prior art are: By adopting a progressive strategy of rule-based pre-screening, semantic matching deep analysis, and association constraint verification, this approach can comprehensively utilize the structured information of the signal, textual semantic information, and domain knowledge. It can effectively address complex situations such as non-standard signal descriptions, the presence of synonyms, abbreviations, and word order reversals, significantly improving the accuracy of matching and the ability to adapt to diverse data.

[0020] It is understood that the beneficial effects of the second to fifth aspects mentioned above can be found in the relevant descriptions in the first aspect mentioned above, and will not be repeated here. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 A flowchart illustrating a power signal association method provided in this application embodiment; Figure 2 An internal flowchart of the progressive matching operation provided in the embodiments of this application; Figure 3 This is a schematic diagram of the feedback and dynamic learning closed loop provided in the embodiments of this application; Figure 4 This is an example diagram of the data structure of the telemetry configuration description file generated for an embodiment of this application.

[0023] Figure 5 This application provides a schematic diagram of the architecture of a power signal correlation system. Figure 6 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0024] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0025] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0026] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0027] As used in this application specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if detected [the described condition or event]" may be interpreted, depending on the context, as meaning "once determined," "in response to determination," "once detected [the described condition or event]," or "in response to detection [the described condition or event]."

[0028] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0029] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0030] Example 1 This embodiment provides a specific implementation of a power signal association method. In the construction and operation of modern smart substations, it is necessary to accurately associate the remote control information point table issued by the superior dispatch center with the internal signal points of hundreds or thousands of intelligent electronic devices (IEDs) within the substation. This is a tedious and error-prone task. This method aims to automatically establish a precise correspondence between the signals to be configured issued by the dispatch center and the signals of local equipment in the substation, thereby efficiently and accurately generating remote control configuration description files and solving the aforementioned pain points. (Refer to...) Figure 1 This figure is an overall flowchart of a power signal association method provided in an embodiment of this application. The following will be combined with... Figure 1-4 The steps of this method will be explained.

[0031] Step S100: Obtain the first signal set as the signal to be matched and the second signal set as the candidate signal.

[0032] This step is the data preparation stage for the entire association process. Its purpose is to extract the original signal information that needs to be associated from data from different sources and in different formats.

[0033] In this embodiment, the scheduling point table file is loaded and parsed to extract signal entries to form the first signal set; the system configuration file is loaded and parsed to extract data object information to form the second signal set.

[0034] In this embodiment, the scheduling point table file can be a spreadsheet file, such as an Excel file (.xlsx or .xls), where each row represents a remote control signal that needs to be configured, typically containing columns such as SCADA point number, signal description, signal type, and associated bay. For example, a typical signal description is "#3 main transformer low 503 switch intelligent terminal B device receives protection GOOSE_B2 link failure alarm". The file is read line by line, and all signal entries are extracted, especially their descriptive text and associated metadata, to form the first signal set.

[0035] The second signal set can originate from a local configuration file in the substation containing a wealth of detailed technical information. In this embodiment, the second signal set can be obtained by loading and parsing a system configuration file conforming to the IEC 61850 standard. This file is typically a System Configuration Description (SCD) file in Extensible Markup Language (XML) format. The SCD file defines in detail the hierarchical structure and communication relationships of all intelligent electronic devices (IEDs), logical devices (LDevices), logical nodes (LNs), data objects (DOs), and data attributes (DAs) within the substation. An XML parser (such as lxml or Xerces) is used to traverse the tree structure of the SCD file, locating all signal points (usually specific data attributes such as stVal, q, t, etc.) related to the actual monitoring and protection functions, and extracting their complete IEC 61850 path information and associated descriptive text (usually stored in...). <dai>or <doi>(In the desc attribute of the tag). These extracted data object information together constitute the second signal set.

[0036] Step S200: Extract a first key field for rule matching and a first signal description text for semantic matching from the first signal set; and extract a second key field for rule matching and a second signal description text for semantic matching from the second signal set.

[0037] This step aims to separate the structured and unstructured information used in different matching stages from the raw, unstructured signal description, and to normalize the text to improve the robustness and accuracy of subsequent matching algorithms.

[0038] For each signal to be matched in the first signal set, the specific execution process of this step is as follows: extract a first original description text from the first signal set, extract a first substring as the first key field from the first original description text, and use the remaining part of the first original description text after removing the extracted first substring as the first key field as the first signal description text; and extract a second original description text from the second signal set, extract a second substring as the second key field from the second original description text, and use the remaining part of the second original description text after removing the extracted second substring as the second key field as the second signal description text.

[0039] To facilitate understanding, an example is given below. First, the complete description string is obtained from the signal entry as the first original description text, such as "#3 main transformer low 503 switch intelligent terminal B device receives protection GOOSE_B2 chain break alarm". A preset regular expression rule base containing prior knowledge of the power field is applied to extract substrings representing high-level electrical equipment or bay units from this first original description text, which are used as the first key field for rule matching. This regular expression can be designed to match patterns such as #[^transformer]+transformer, kV[^line]+line, kV[^bus]+busline, etc. Among them, the fixed prefix "#" locks the numbered device identifier, clearly indicating that the matching object is a main transformer device with a specific number; the reverse character class "[^transformer]+" means that [^transformer] matches any character other than transformer, and + requires that this type of character appear at least once; the fixed suffix "transformer" clearly indicates that the matching target is a main transformer type device, ensuring the consistency of the device type of the extracted results (avoiding confusion with other devices such as lines and busbars). The fixed prefix "kV" directly associates with voltage units (e.g., 110kV, 220kV), explicitly matching line equipment with voltage levels and excluding other equipment without voltage markings; the reverse character class "[^line]+" matches any character except "line," and "+" requires the character to appear at least once; the fixed suffix "line" limits the matching target to line equipment, ensuring consistent equipment types in the extracted results and avoiding confusion with transformers, busbars, etc. kV[^busbar]+busbar is not described. This regular expression is existing technology.

[0040] In this example, the rule (#\d{1,2}[^change]+change) successfully extracts "#3main changer" as the first key field. The use of regular expressions to determine the first key field is existing technology. Then, the remaining text after removing this first key field from the original description text, namely "changer low 503 switch smart terminal B device receive protection GOOSE_B2 disconnection alarm", is used as the first signal description text for semantic matching. Simultaneously, to eliminate noise interference in subsequent semantic matching, this first signal description text needs to be cleaned and standardized. This processing may include, but is not limited to: removing punctuation marks and other special characters, unifying the case of English letters, converting full-width characters to half-width characters, and converting Roman numerals (such as I, II, III) to Arabic numerals (1, 2, 3), etc.

[0041] Similarly, for each candidate signal in the second signal set, a similar operation is performed to extract the second key field and the cleaned second signal description text from its original description text. Through this step, the original, heterogeneous signal data is transformed into a unified format consisting of structured key fields and standardized description text, laying a solid data foundation for subsequent progressive matching.

[0042] Step S300: For any signal to be matched in the first signal set, a progressive matching operation including rule matching, semantic matching and association constraints is performed based on the first key field, the first signal description text, the second key field and the second signal description text, so as to determine the target matching signal from the second signal set.

[0043] The progressive matching operation includes the following steps executed sequentially: Rule matching: Determine whether the first key field of the signal to be matched is complete; if complete, then based on the first key field and the second key field, filter out a subset of candidate signals from the second signal set; if incomplete, then use the entire second signal set as the subset of candidate signals. Specifically, perform a full-word comparison between the first key field of the signal to be matched and the corresponding second key field of each signal in the second signal set, and use the set of candidate signals that pass the full-word comparison in the second signal set as the subset of candidate signals.

[0044] Semantic matching: Using a preset semantic representation model, calculate the semantic similarity score between the first signal description text of the signal to be matched and the second signal description text of each signal in the candidate signal subset.

[0045] Association constraint: Based on the preset association relationship model, the reasonableness of the association between the signal to be matched and the candidate signal whose semantic similarity score is higher than the preset semantic threshold is judged, and the semantic similarity score is adjusted based on the association reasonableness judgment result. The candidate signal with the highest adjusted score is determined as the target matching signal.

[0046] In the embodiments of this application, the core of the technical solution lies in performing a progressive matching operation that includes rule matching, semantic matching, and association constraints. This simulates the thought process of a domain expert: first, limiting the scope by device intervals; then, performing semantic understanding within a small scope; and finally, confirming the result by combining electrical logic. (Refer to...) Figure 2 For any signal to be matched in the first signal set, based on the fields and text extracted in step S200, the target matching signal will be accurately determined from the second signal set through a multi-stage funnel-shaped filtering process that includes rule matching, semantic matching and association constraints.

[0047] The first stage of the progressive matching is rule matching S310. This stage aims to use structured, highly discriminative key fields (usually electrical bays or main equipment names) for fast, coarse-grained filtering. Its core objective is to significantly reduce the computational scope of subsequent semantic matching, thereby improving overall matching efficiency without sacrificing recall. The specific process is as follows: First, it is determined whether the first key field of the signal to be matched is complete and valid. If the first key field is complete, for example, "#3 main transformer", then all candidate signals in the second signal set are traversed, and their first key field is compared with the second key field of each candidate signal. In this embodiment, the comparison adopts a full-word comparison strategy, that is, it requires that the strings of the two key fields are exactly the same. All candidate signals whose second key field completely matches "#3 main transformer" will be filtered out, forming a significantly reduced subset of candidate signals. If the first key field of the signal to be matched is incomplete or missing (for example, described as a public equipment alarm), then rule matching cannot be effectively performed. In this case, the entire second signal set is taken as a subset of candidate signals and directly proceeds to the next stage.

[0048] The second stage of the progressive matching is semantic matching (S320). After the initial screening by rule matching, this stage utilizes advanced Natural Language Processing (NLP) technology to perform deep semantic analysis on the candidate signal subset. This quantifies the similarity between signal descriptions at the semantic level, rather than merely the overlap of surface-level text. In this embodiment, this stage is implemented through a pre-defined semantic representation model. This model can be a Sentence-BERT model that has undergone domain-adaptive fine-tuning on professional corpora in the power industry (such as technical specifications, operating procedures, equipment manuals, fault reports, etc.). This model can deeply understand the professional terminology, abbreviations (e.g., differential refers to differential protection), and expression habits in the power industry, and efficiently map text of arbitrary length to a high-dimensional semantic vector space with a fixed dimension (e.g., 768 dimensions). Within this space, texts with similar semantics have corresponding vectors that are geometrically closer (e.g., angles).

[0049] The specific calculation process is as follows: First, the first signal description text of the signal to be matched (e.g., the low-voltage 503 switch smart terminal b device receives the protection goose_b2 disconnection alarm) is input into the semantic representation model to obtain its corresponding query semantic vector. Then, for each candidate signal in the subset of candidate signals, its second signal description text is also input into the model to obtain a series of candidate semantic vectors. Next, the query vector is calculated. With each candidate vector The cosine similarity between the two texts is used to quantify their semantic similarity score. The calculated cosine similarity value ranges between -1 and 1, with the value closer to 1 indicating greater semantic similarity. For example, differential protection action and differential tripping will have very high scores (e.g., >0.9), while voltage over-limit alarm will have very low scores (e.g., <0.3).

[0050] The third stage of the progressive matching process is the association constraint S330. It is understandable that relying solely on semantic matching might lead to seemingly reasonable but logically flawed matches due to coincidences or ambiguities in the textual descriptions (e.g., matching a main transformer signal to an outgoing line). Therefore, this stage introduces knowledge from the power system domain to perform a final rationality check and score adjustment on high-scoring semantic matching results, essentially adding an expert review step to the matching process. This relies on a pre-defined association model.

[0051] In this embodiment, the association model is an electrical topology association model. The construction process includes: parsing the system configuration file to identify primary equipment topology association data that defines the electrical connection relationships of primary equipment and the division of electrical bays, as well as secondary equipment function definition data that defines the logical functions of secondary equipment; traversing the topology information in the primary equipment topology association data to construct the hierarchical subordinate relationship between electrical bays and the included primary equipment, and clarifying the electrical bay to which each primary equipment belongs; traversing the secondary equipment function definition data to establish a mapping relationship between secondary equipment and its logical functions and target equipment; wherein, the target equipment includes the primary equipment being measured, controlled, or protected; integrating the hierarchical subordinate relationship and the mapping relationship to generate an electrical topology association model that reflects the inherent association between the logical function of any of the secondary equipment and its corresponding electrical bay, as the association model.

[0052] Specifically, the SCD file is first deeply parsed to identify the primary equipment topology association data that defines the physical connection relationships and electrical bay divisions of primary equipment (usually located in...). <voltagelevel> 、 <bay>(within the tags), and secondary equipment function definition data that defines the logic functions of secondary equipment (such as protection and control devices) (usually located in...). <ied> 、 <ln>(Within the tags). Secondly, the primary equipment topology association data is traversed to construct a complete hierarchical hierarchy tree from substation, voltage level, electrical bay to specific primary equipment, thus clarifying the exact electrical bay to which each primary equipment belongs. Thirdly, the secondary equipment functional definition data is traversed to establish the mapping relationship between logical nodes in the secondary equipment (such as PDIF logical nodes representing differential protection, PTOC logical nodes representing overcurrent protection) and the target primary equipment they protect or control. Finally, by integrating the above hierarchical hierarchy and mapping relationships, a knowledge graph or database that reflects the inherent relationship between any secondary equipment and its signals and the electrical bays to which they are physically and functionally associated is generated.

[0053] Based on this electrical topology association model, the rationality of the association can be judged. Specifically, candidate signals with semantic similarity scores higher than a preset semantic threshold (e.g., 0.8) are verified. For example, if the first key field of the signal to be matched is "#1 main transformer," it indicates that it belongs to the #1 main transformer electrical bay in terms of its business. A high-scoring candidate signal, upon tracing, is found to originate from a monitoring and control device serving a 220kV bus bay in the SCD file. Although their descriptive texts may be semantically similar due to the inclusion of common terms such as "device" and "alarm," the #1 main transformer bay and the 220kV bus bay are topologically independent units, and this match will be deemed not to have reasonable association. Conversely, if the high-scoring candidate signal originates from a protection device serving the #1 main transformer, then it is deemed to have reasonable association.

[0054] Based on the results of the association rationality judgment, the semantic similarity score is adjusted. If the association rationality judgment result is unreasonable, the original semantic similarity score of the candidate signal is multiplied by a preset penalty factor less than 1 (e.g., 0.1), significantly reducing its final score and thus placing it lower in the ranking. If the association rationality judgment result is reasonable, its semantic similarity score remains unchanged, or it can be multiplied by a reward factor greater than 1 (e.g., 1.1) to enhance its weight. After this adjustment, the scores of all candidate signals are re-ranked, and the candidate signal with the highest score is determined as the target matching signal for the signal to be matched.

[0055] As an optional way to enhance interactivity and reliability, after determining the target matching signal, step S400 is executed to receive feedback information from the user regarding the target matching signal.

[0056] In this embodiment of the application, through the user interface (such as...) Figure 5 As shown (which may be a graphical interface), the interface displays the recommended best match result to the operations engineer, possibly with several options that have the next highest scores. The interface clearly lists the description of the signal to be matched, the IEC61850 path and description of the recommended match signal, and the match score. The engineer can review the result; if correct, they can confirm it with one click; if incorrect, they can select the correct match from the secondary list or manually specify it using the search function. These user actions (confirmation, correction) are recorded, forming structured feedback information 401, such as... Figure 3 As shown.

[0057] Accordingly, to enable self-evolution, an optional step S500 can be executed: generating labeled training samples based on the feedback information, and using the training samples to incrementally train the semantic representation model to achieve dynamic learning and model optimization. This step aims to utilize user feedback to build a continuous learning loop, enabling the model to continuously learn from practical applications, thereby achieving continuous model optimization. (Refer to...) Figure 3 The diagram illustrates the closed-loop process of feedback and dynamic learning. In step 402, labeled training samples are automatically generated based on the received feedback information 401. For example, a matching pair confirmed by the user (signal to be matched A, target matching signal B) is constructed as a positive sample with a target similarity of 1.0; an incorrect matching pair corrected by the user (signal to be matched A, system-recommended incorrect signal C) is constructed as a negative sample, with a target similarity that can be set to 0 or -1.0.

[0058] To ensure the model's learning is focused and prioritizes the accuracy of matching key signals, this embodiment assigns weight factors ω corresponding to signal types when generating training samples. Specifically, signal types are pre-classified into multiple importance levels, with different weight factors corresponding to different importance levels. During incremental training of the semantic representation model, a weighted loss function is used to optimize the model; this weighted loss function is constructed based on the weight factors assigned to the training samples.

[0059] In this embodiment of the application, the signal type (which can be obtained from the scheduling point table) can be divided into multiple importance levels in advance according to the importance of the business. For example, protection action signals (such as tripping and reclosing) are the highest level with a weighting factor ω=2.0; key remote signals (such as switch position and disconnector position) are the second highest level with a weighting factor ω=1.5; and general telemetry (such as voltage, current, and power) are the ordinary level with a weighting factor ω=1.0.

[0060] Subsequently, in step 403, the semantic representation model is incrementally trained or fine-tuned using these weighted training samples. During training, a weighted loss function is used to optimize the model parameters. In this embodiment, a weighted cosine similarity loss function can be used, the specific form of which is: .in, It is the vector of the i-th correct match determined based on user feedback. It is the vector predicted by the model. This refers to the weight factor corresponding to the sample. In this way, when updating parameters, the model will pay more attention to the loss caused by important signal samples with high weights, thus prioritizing the learning of how to correctly match these key signals. After training, the optimized semantic representation model 404 is obtained. This model will replace the old model for subsequent matching tasks, thereby achieving continuous self-optimization and adaptation to new data patterns and naming conventions.

[0061] Step S600: Generate a telecontrol configuration description file. Based on the target matching signal finally confirmed by the user or automatically determined by the system, the relevant information in the first signal set and the second signal set is automatically integrated to generate a structured telecontrol configuration description file that can be directly used by downstream systems.

[0062] Reference Figure 4 The data structure example in this file clearly demonstrates its contents. This file is typically a table (such as a CSV or Excel file), with each row representing a complete signal association record. It contains several key fields, such as: the SCADA point number field from the dispatch point table, the complete IEC61850 path field for the target signal from the SCD file (this is the core information for the final configuration), the detailed Chinese description field for the plant / substation end from the SCD file, and the equipment category field parsed from the SCD file. This generated file has a standardized format and complete information, and can be directly imported by subsequent configuration tools or SCADA / EMS systems. This reduces manual configuration work that would have taken days or even weeks to just a few hours, and significantly lowers the safety risks caused by human error.

[0063] Example 2 This embodiment is a variant of embodiment 1. Its core idea is consistent with that of embodiment 1. The main difference lies in the specific type and implementation method of the association relationship model used in the association constraint S330, which aims to verify the matching results from different dimensions.

[0064] In Example 1, the correlation model was an electrical topology-based correlation model that focused on verifying the physical location of the signal source and the rationality of the electrical connections. In this example, the correlation model is replaced by a device functional logic constraint model. This model is not based on the physical topology of the devices, but focuses on the logical correspondence between device types and their functional outputs, i.e., what devices can generate what signals.

[0065] The construction process of the functional logic constraint model for this equipment is as follows: First, a knowledge base is predefined by power system experts. This knowledge base, in the form of rules or mapping tables, explicitly defines the types of signals (such as protection action signals, remote signaling, remote measurement, SOE events, synchronization status, etc.) that different types of secondary equipment (identified by their logical node classes in the SCD file, such as PDIF, PTOC, MMXU, CSWI, PTTR, etc.) can generate. For example, the model defines the following rules: equipment with a logical node class of PDIF (differential protection) or PTOC (overcurrent protection) can generate signals such as differential protection action and overcurrent protection action; equipment with a logical node class of CSWI (switch controller) can generate switch position signals; and equipment with a logical node class of PTTR (clock synchronization) can only generate time-related signals such as clock synchronization anomaly and time synchronization status.

[0066] When executing the association constraint S330, the working method changes accordingly: For a candidate signal with a semantic similarity score higher than the threshold, firstly, the logical node type of its source is parsed (this information can be obtained from the SCD file). Then, the descriptive text of the signal to be matched is analyzed to identify the function or event type it contains (e.g., identified by keywords such as differential protection, switch position, etc.). Suppose the signal to be matched is "#1 main transformer differential protection action", and its function type is differential protection. Suppose that in the semantic matching stage S320, a candidate signal from the #1 main transformer temperature online monitoring device is found, and the logical node in its SCD file is TTMP (temperature monitoring). Although its description may also contain words such as #1 main transformer and action, resulting in a high semantic score, in the association constraint stage, querying the equipment functional logic constraint model reveals that the functional scope of the TTMP logical node is limited to temperature measurement and alarm, and does not include differential protection function. Therefore, this match is determined to violate the equipment functional logic, is an unreasonable match, and a significant penalty factor is applied to its score.

[0067] In this way, this embodiment can effectively avoid erroneous matches caused by similar text but completely inconsistent functional logic, thus improving the logical rigor of the matching results. This method is particularly suitable for handling signals generated by devices with overlapping functions or ambiguous naming, and can constrain and verify the matching results from a functional perspective. In some implementations, topological constraints and functional logic constraints can even be combined to form a stronger multi-dimensional constraint system.

[0068] Example 3 This embodiment is another variant of Embodiment 1, which optimizes the specific implementation of semantic matching S320. It aims to further balance the computational efficiency and final accuracy of matching through a hierarchical strategy, especially when dealing with highly similar and easily confused signal pairs.

[0069] In Example 1, the semantic matching stage S320 uses a single Sentence-BERT model to calculate the similarity of all candidate signals. It is understandable that although the Sentence-BERT model with a dual-tower structure is relatively fast, its ability to distinguish between extremely similar text pairs with only subtle differences may be limited, because the vectors of the two sentences are generated independently and lack deep interactive information fusion.

[0070] To address this issue, this embodiment employs a hybrid model strategy of coarse screening + fine ranking (Recall + Rank) in semantic matching S320. This strategy combines the high efficiency of the dual-tower model with the high accuracy of the cross-encoder model. The specific process adjustments are as follows: The first step is semantic coarse screening (Recall). Similar to Example 1, an efficient dual-tower model (such as Sentence-BERT) is first used to quickly compute the subset of candidate signals filtered by rule matching S310. This involves comparing the cosine similarity of the vector of the signal to be matched with the pre-computed vectors of all candidate signals. The purpose of this step is not to directly determine the final result, but to quickly filter out the top-K semantically most relevant candidates from potentially hundreds or thousands of candidate signals, for example, K=5 or 10. This reduces the number of signals that need subsequent fine processing, ensuring the efficiency of the overall process.

[0071] The second step is semantic ranking. The signal to be matched is paired one by one with the K candidate signals selected in the initial screening stage. Each text pair is then fed into a more computationally complex but more accurate cross-encoder model (such as standard BERT, RoBERTa, etc.) for secondary ranking. Unlike the dual-tower model, which encodes two sentences separately, the cross-encoder model concatenates the two sentences into a single input sequence (e.g., [CLS]Signal to be matched description [SEP]Candidate signal description [SEP]), and uses a multi-layered self-attention mechanism within the model to fully interact and context-awarely perceive the two sentences at the lexical level. Finally, the model typically predicts a single similarity score based on the final output vector labeled [CLS] through a linear layer. Due to this deep interactive analysis, the cross-encoder is more sensitive to subtle differences between texts, negation relationships, or key discriminative words.

[0072] For example, assuming the signal to be matched is the A-phase current on the high-voltage side of transformer #1, after coarse screening, the Top-2 candidate signals might be the A-phase current on the high-voltage side of transformer #1 and the A-phase current on the medium-voltage side of transformer #1. Their semantic vectors might be very similar in the dual-tower model. In the fine-tuning stage, the cross-encoder model fully understands the strong correlation between the high-voltage side and high voltage when processing the first pairing; while when processing the second pairing, it clearly identifies the high-voltage side and medium-voltage side as key distinguishing terms describing different objects, even though other terms are the same. Therefore, the cross-encoder can assign a higher similarity score (e.g., 0.98) to the first pairing, while assigning a significantly lower score (e.g., 0.45) to the second pairing.

[0073] Finally, the subsequent association constraint S330 will operate based on this more accurate similarity score obtained after fine-ranking. This embodiment significantly improves the ability to distinguish easily confused signals by adding a fine-ranking step, thereby increasing the upper limit of the overall matching accuracy and making the final result more reliable.

[0074] Example 4 This embodiment is another variant of Embodiment 1. The improvement is in the first stage of progressive matching, namely rule matching S310, which aims to improve the fault tolerance and recall rate of this stage.

[0075] In Example 1, rule matching S310 employs a strict full-word comparison strategy, requiring that the first key field of the signal to be matched be completely identical to the second key field of the candidate signal. In practical applications, due to different engineers' input habits, the key fields themselves may contain minor inconsistencies or errors, such as spelling errors, character substitutions (e.g., i and 1, 0 and o), mixed use of Chinese and English punctuation (e.g., interval _A and interval -A), and inconsistent abbreviations. Strict matching can cause these signals that should match to be incorrectly filtered out in the first stage, preventing them from entering the subsequent semantic matching stage, thus reducing the overall recall rate of the system.

[0076] To enhance the robustness of the system, this embodiment replaces the full-word comparison in rule matching S310 with a fuzzy matching algorithm based on edit distance. Specifically, when comparing the first key field of the signal to be matched with the second key field of the candidate signal, the strings are no longer required to be completely equal; instead, the normalized edit distance between them is calculated. In this embodiment, the classic Levenstein distance can be used, which represents the minimum number of single-character edits (insertion, deletion, or replacement) required to transform one string into another.

[0077] To ensure that the distance metric is unaffected by string length, a uniform threshold is set, and the normalized edit distance is calculated using the following formula: .in, It is a string and Lewinstein distance, This is the string length. The normalized distance value is between 0 and 1, where 0 represents completely identical and 1 represents completely different.

[0078] A preset edit distance threshold is set, for example, 0.2. During rule matching, two key fields are considered a match only if their normalized edit distance is less than or equal to this threshold, and the corresponding candidate signals are included in the initial candidate signal subset.

[0079] For a specific example: Suppose the key field of a signal to be matched is extracted as "110kV test line i", while the corresponding key field in the plant's SCD file is recorded as "110kV test line 1". Under the strict matching in Example 1, this candidate signal would be filtered out because i and 1 are different. However, in this example, the Levinstein distance between the two strings is calculated to be 1, the maximum string length is 10, and the normalized edit distance is... Since 0.1 is less than the preset threshold of 0.2, the two key fields are determined to be a successful match. Therefore, this candidate signal that should have matched will be successfully retained and enter the subsequent semantic matching S320, thereby increasing the probability that the candidate signal will be correctly associated.

[0080] This embodiment improves the fault tolerance and recall rate of the rule matching stage by introducing fuzzy matching, enabling the entire matching process to better cope with problems such as non-standard data and minor input errors, and significantly enhancing overall robustness.

[0081] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0082] Corresponding to the power signal association method described in the above embodiments, Figure 5 A structural block diagram of a power signal association system provided in an embodiment of this application is shown. For ease of explanation, only the parts related to the embodiments of this application are shown.

[0083] Reference Figure 5 A power signal correlation system, comprising: The acquisition module is used to acquire a first set of signals as signals to be matched and a second set of signals as candidate signals.

[0084] As the system's data entry point, it is configured to execute step S100 as described in Embodiment 1. Internally, it integrates a file reading interface (for processing Excel, XML, and other files) and a database connection driver (for connecting to SQL or NoSQL databases). It can automatically load and parse the scheduling point table file and SCD file from a specified path based on preset configurations or real-time user commands. The parsed data is then structured to form a first signal set and a second signal set. This module can also receive files manually uploaded by the user through a user interface.

[0085] The extraction module is configured to extract a first key field for rule matching and a first signal description text for semantic matching from the first signal set, and to extract a second key field for rule matching and a second signal description text for semantic matching from the second signal set.

[0086] The extraction module can internally include a text processing engine. This engine utilizes a configurable regular expression library to process the descriptive text of each signal to extract key fields and the main descriptive text. It also includes a suite of text cleaning tools for denoising and standardizing the descriptive text.

[0087] The progressive matching module is used to perform a progressive matching operation, including rule matching, semantic matching and association constraints, on any signal to be matched in the first signal set, based on the first key field, the first signal description text, the second key field and the second signal description text, so as to determine the target matching signal from the second signal set.

[0088] The generation module is used to generate a telemetry configuration description file based on the target matching signal.

[0089] This module receives the final matching result pairs determined by the progressive matching module, and integrates and formats the information in each matching pair (such as dispatch point number, IEC61850 path, Chinese description, signal type, etc.) according to the preset output template, ultimately generating a complete and compliant telecontrol configuration description file (such as...). Figure 4 (As shown). The generated file can be presented to the user for download through the user interface, or directly output to a specified storage location for automatic use by other systems.

[0090] In addition, the system includes a database / model library and a user interface. The database / model library is a persistent storage system used to centrally store and manage semantic model files, relational model data, user feedback logs, and intermediate data generated during training. The user interface is the medium for human-computer interaction; it is a graphical user interface (GUI) application installed on the client computer, enabling operations such as starting tasks, monitoring progress, reviewing results, and providing feedback.

[0091] Optionally, the acquisition module is used for: Load and parse the scheduling point table file, and extract signal entries to form the first signal set; The system configuration file is loaded and parsed to extract data object information to form the second signal set.

[0092] Optionally, the progressive matching module is used for: Rule matching: Determine whether the first key field of the signal to be matched is complete; if complete, then based on the first key field and the second key field, filter out a subset of candidate signals from the second signal set; if incomplete, then use the entire second signal set as the subset of candidate signals. Semantic matching: Using a preset semantic representation model, calculate the semantic similarity score between the first signal description text of the signal to be matched and the second signal description text of each signal in the candidate signal subset; Association constraint: Based on the preset association relationship model, the reasonableness of the association between the signal to be matched and the candidate signal whose semantic similarity score is higher than the preset semantic threshold is judged, and the semantic similarity score is adjusted based on the association reasonableness judgment result. The candidate signal with the highest adjusted score is determined as the target matching signal.

[0093] Optionally, the progressive matching module is used for: The first key field of the signal to be matched is sequentially compared with the corresponding second key field of each signal in the second signal set using a whole-word comparison. The set of candidate signals that pass the full word comparison in the second signal set is taken as the candidate signal subset.

[0094] Optionally, the extraction module is used for: The first original description text is extracted from the first signal set; Extract the first substring from the first original description text as the first key field; The remaining part after removing the first substring that has been extracted as the first key field from the first original description text is used as the first signal description text. The second original description text is extracted from the second signal set; Extract the second substring from the second original description text as the second key field; The remaining portion of the second original description text after removing the extracted second substring which is the second key field is used as the second signal description text.

[0095] Optionally, the process of constructing the association model includes: The system configuration file is parsed to identify primary equipment topology association data that defines the electrical connection relationships and electrical bay divisions of primary equipment, as well as secondary equipment function definition data that defines the logical functions of secondary equipment. Traverse the topology information in the primary equipment topology association data to construct the hierarchical relationship between electrical bays and the primary equipment they contain, and clarify the electrical bay to which each primary equipment belongs; Traverse the secondary device function definition data to establish a mapping relationship between the secondary devices and their logical functions and the target devices; wherein, the target devices include the primary devices being measured, controlled or protected; By integrating the hierarchical relationships and the mapping relationships, an electrical topology association model that reflects the inherent relationship between the logical function of any of the secondary devices and the electrical bay to which they belong is generated, which serves as the association relationship model.

[0096] Optionally, the result of the association rationality judgment includes an unreasonable judgment result and a reasonable judgment result; the association constraint includes: If the result of the association rationality judgment is the result of the unreasonable judgment, then the semantic similarity score is multiplied by a preset penalty factor less than 1; If the result of the association rationality judgment is the reasonable judgment result, then the semantic similarity score remains unchanged.

[0097] Optionally, the power signal correlation system also includes a feedback module for: Receive user feedback information regarding the target matching signal; Labeled training samples are generated based on the feedback information, and the semantic representation model is incrementally trained using the training samples to optimize the model.

[0098] Optionally, the feedback module is used for: When generating training samples, a weight factor corresponding to the signal type is assigned to the training samples; wherein, the signal types are pre-divided into multiple importance levels, and different importance levels of signal types correspond to different weight factors. During the incremental training of the semantic representation model, a weighted loss function is used to optimize the semantic representation model; wherein the weighted loss function is constructed based on the weight factors assigned in the training samples.

[0099] It should be noted that the information interaction and execution process between the above-mentioned devices / units / modules are based on the same concept as the method embodiments of this application. For details on their specific functions and technical effects, please refer to the method embodiments section, and they will not be repeated here.

[0100] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0101] This application also provides a computer device, which includes: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the processor executes the computer program to implement the steps in any of the above method embodiments.

[0102] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, can implement the steps in the above-described method embodiments.

[0103] This application provides a computer program product. When the computer program product is run on a computer device, the computer device can implement the steps in the above-described method embodiments.

[0104] Figure 6 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Figure 6 As shown, the computer device of this embodiment includes: at least one processor 60 ( Figure 6 (Only one is shown in the diagram), memory 61, and computer program 62 stored in the memory 61 and executable on the at least one processor 60, wherein the processor 60 executes the computer program 62 to implement the steps in the various visual programming method embodiments described above.

[0105] The computer device may include, but is not limited to, a processor 60 and a memory 61. Those skilled in the art will understand that... Figure 6 The examples of computer devices are merely examples and do not constitute a limitation on computer devices. They may include more or fewer components than shown in the illustration, or combinations of certain components, or different components. For example, they may also include input / output devices, network access devices, etc.

[0106] The processor 60 may be a Central Processing Unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0107] In some embodiments, the memory 61 may be an internal storage unit of the computer device, such as a hard drive or memory. In other embodiments, the memory 61 may be an external storage device of the computer device, such as a plug-in hard drive, smart media card (SMC), secure digital card (SD), flash card, etc. Furthermore, the memory 61 may include both internal and external storage units of the computer device. The memory 61 is used to store the operating system, applications, bootloader, data, and other programs, such as the program code of the computer program. The memory 61 can also be used to temporarily store data that has been output or will be output.

[0108] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of this application can be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include at least: any entity or device capable of carrying computer program code to a device / computer equipment, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium. Examples include USB flash drives, portable hard drives, magnetic disks, or optical disks.

[0109] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0110] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0111] In the embodiments provided in this application, it should be understood that the disclosed apparatus / computer devices and methods can be implemented in other ways. For example, the apparatus / computer device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0112] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0113] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.< / ln> < / ied> < / bay> < / voltagelevel> < / doi> < / dai>

Claims

1. A method for associating power signals, characterized in that, include: Obtain a first set of signals as signals to be matched and a second set of signals as candidate signals; A first key field for rule matching and a first signal description text for semantic matching are extracted from the first signal set; and a second key field for rule matching and a second signal description text for semantic matching are extracted from the second signal set. For any signal to be matched in the first signal set, a progressive matching operation including rule matching, semantic matching and association constraints is performed based on the first key field, the first signal description text, the second key field and the second signal description text to determine the target matching signal from the second signal set. Based on the target matching signal, a telemetry configuration description file is generated.

2. The method according to claim 1, characterized in that, The acquisition of a first signal set as the signal to be matched and a second signal set as the candidate signal includes: Load and parse the scheduling point table file, and extract signal entries to form the first signal set; The system configuration file is loaded and parsed to extract data object information to form the second signal set.

3. The method according to claim 2, characterized in that, The progressive matching operation includes the following steps, performed sequentially: Rule matching: Determine whether the first key field of the signal to be matched is complete; if complete, then based on the first key field and the second key field, filter out a subset of candidate signals from the second signal set; if incomplete, then use the entire second signal set as the subset of candidate signals. Semantic matching: Using a preset semantic representation model, calculate the semantic similarity score between the first signal description text of the signal to be matched and the second signal description text of each signal in the candidate signal subset; Association constraint: Based on the preset association relationship model, the reasonableness of the association between the signal to be matched and the candidate signal whose semantic similarity score is higher than the preset semantic threshold is judged, and the semantic similarity score is adjusted based on the association reasonableness judgment result. The candidate signal with the highest adjusted score is determined as the target matching signal.

4. The method according to claim 3, characterized in that, The step of filtering candidate signal subsets from the second signal set based on the first key field and the second key field includes: The first key field of the signal to be matched is sequentially compared with the corresponding second key field of each signal in the second signal set using a whole-word comparison. The set of candidate signals that pass the full word comparison in the second signal set is taken as the candidate signal subset.

5. The method according to claim 1, characterized in that, The step of extracting a first key field for rule matching and a first signal description text for semantic matching from the first signal set, and extracting a second key field for rule matching and a second signal description text for semantic matching from the second signal set, includes: The first original description text is extracted from the first signal set; Extract the first substring from the first original description text as the first key field; The remaining part after removing the first substring that has been extracted as the first key field from the first original description text is used as the first signal description text. The second original description text is extracted from the second signal set; Extract the second substring from the second original description text as the second key field; The remaining portion of the second original description text after removing the extracted second substring which is the second key field is used as the second signal description text.

6. The method according to claim 3, characterized in that, The process of constructing the relationship model includes: The system configuration file is parsed to identify primary equipment topology association data that defines the electrical connection relationships and electrical bay divisions of primary equipment, as well as secondary equipment function definition data that defines the logical functions of secondary equipment. Traverse the topology information in the primary equipment topology association data to construct the hierarchical relationship between electrical bays and the primary equipment they contain, and clarify the electrical bay to which each primary equipment belongs; Traverse the secondary device function definition data to establish a mapping relationship between the secondary devices and their logical functions and the target devices; wherein, the target devices include the primary devices being measured, controlled or protected; By integrating the hierarchical relationships and the mapping relationships, an electrical topology association model that reflects the inherent relationship between the logical function of any of the secondary devices and the electrical bay to which they belong is generated, which serves as the association relationship model.

7. The method according to claim 6, characterized in that, The results of the association rationality judgment include unreasonable judgment results and reasonable judgment results; the association constraints include: If the result of the association rationality judgment is the result of the unreasonable judgment, then the semantic similarity score is multiplied by a preset penalty factor less than 1; If the result of the association rationality judgment is the reasonable judgment result, then the semantic similarity score remains unchanged.

8. The method according to claim 3, characterized in that, The method further includes: Receive user feedback information regarding the target matching signal; Labeled training samples are generated based on the feedback information, and the semantic representation model is incrementally trained using the training samples to optimize the model.

9. The method according to claim 8, characterized in that, The step of generating labeled training samples based on the feedback information, and using the training samples to incrementally train the semantic representation model to optimize the model, includes: When generating training samples, a weight factor corresponding to the signal type is assigned to the training samples; wherein, the signal types are pre-divided into multiple importance levels, and different importance levels of signal types correspond to different weight factors. During the incremental training of the semantic representation model, a weighted loss function is used to optimize the semantic representation model; wherein the weighted loss function is constructed based on the weight factors assigned in the training samples.

10. A power signal correlation system, characterized in that, include: The acquisition module is used to acquire a first set of signals as signals to be matched and a second set of signals as candidate signals; An extraction module is configured to extract a first key field for rule matching and a first signal description text for semantic matching from the first signal set, and to extract a second key field for rule matching and a second signal description text for semantic matching from the second signal set. The progressive matching module is used to perform a progressive matching operation, including rule matching, semantic matching and association constraints, on any signal to be matched in the first signal set, based on the first key field, the first signal description text, the second key field and the second signal description text, so as to determine the target matching signal from the second signal set. The generation module is used to generate a telemetry configuration description file based on the target matching signal.

Images