A method and system for generating a remote forwarding point table, and a computer device

Abstract

This application discloses a method and system for generating a remote control forwarding point table, as well as a computer device, belonging to the field of power system technology. The method for generating the remote control forwarding point table includes: standardizing the full point table by inputting the standard signal description into an embedding model to obtain a full vector; semantically decomposing the signal description of the master station table to obtain interval information and filter out candidate devices; searching the full vector corresponding to the candidate device based on the master station vector to obtain candidate signals; and inputting the candidate signals into a matching model, which determines the target signal matching the master station table from the candidate signals. This application converts text signals into high-dimensional vectors through an embedding model and achieves cross-semantic matching through vector similarity calculation, solving the problem of mismatch between non-standard manufacturer descriptions and standard master station descriptions, significantly reducing the omission rate and mismatch rate. By determining the target signal through the matching model, the configuration efficiency of the remote control forwarding point table is improved.

Description

Technical Field

[0001] This application relates to the field of power system technology, specifically to a method and system for generating a remote switching point table, and computer equipment. Background Technology

[0002] In the field of power system dispatch automation, the configuration of substation telecontrol communication is a crucial link in enabling the dispatch master station to monitor the substation. The telecontrol forwarding point table is used to ensure accurate signal transmission between the substation and the dispatch master station, establishing a one-to-one correspondence between the full range of signals acquired at the substation and the required monitoring signals at the dispatch master station. The generation of the telecontrol forwarding point table is primarily done manually. The specific process is as follows: dispatchers first issue the master station table, and substation maintenance personnel compare the master station table with the full range of points from all devices in the substation, filtering out matching signals one by one to finally form the forwarding point table. However, the signal descriptions in the full range of points in the substation are generated by devices from different manufacturers, exhibiting obvious manufacturer characteristics, abbreviations, or non-standard naming, resulting in a significant semantic gap with the dispatch master station table. Manual matching is prone to omissions and mismatches. Furthermore, the full range of points in a single substation contains tens of thousands of signals, making the manual comparison with the master station table extremely labor-intensive. Moreover, with the expansion of the power grid and the increasing frequency of new substation commissioning and old substation upgrades, manual configuration can no longer meet the timeliness requirements. Summary of the Invention

[0003] This invention provides a method, system, and computer device for generating a telemetry forwarding point table, aiming to solve the technical problems of frequent omissions and mismatches in telemetry forwarding point tables, as well as low matching efficiency.

[0004] In a first aspect, embodiments of this application provide a method for generating a telemetry forwarding point table, comprising the following steps: Obtain the full point table of the substation, perform signal standardization processing on the full point table to obtain a standard signal description, and input the standard signal description and the device description of the full point table into the embedding model to obtain the full vector. Semantic segmentation is performed on the signal descriptions in the main station table to obtain interval information; candidate devices are selected from the substation devices based on the interval information and semantic similarity. The main station signal in the main station table is vectorized using the embedding model to obtain the main station vector; based on the main station vector, the candidate signal is obtained by searching the full vector corresponding to the candidate device. Based on the substation parameter information, a prompt message is constructed, and the candidate signals are input into the matching model. The matching model then determines the target signal that matches the main station table from the candidate signals based on the prompt message.

[0005] In one embodiment of this application, the signal standardization processing of the full point table includes: A thesaurus for the power industry is constructed, and the non-standard signal descriptions in the full-point table are replaced with the standard signal descriptions using the thesaurus.

[0006] In one embodiment of this application, the method of inputting the device description of the standard signal description and the full point table into the embedding model includes: Extract the bay number corresponding to the substation device from the full point table, and establish a device bay mapping table corresponding to the substation device and the bay number; The device description and the standard signal description are input into the embedding model, converted into the full vector by the embedding model, and stored in the vector database.

[0007] In one embodiment of this application, the semantic decomposition of the signal description of the main site table includes: parsing the signal description of the main site table using the matching model, and semantically decomposing interval information and signal information; The step of selecting candidate devices from substation devices based on the interval information and semantic similarity includes: When the signal in the main station table is a conventional signal, prompt information is generated according to the substation voltage level to guide the matching model to initially screen the substation devices, and then according to the device interval mapping table, devices in the same interval, common measurement and control devices, and composite signal virtual devices are added as the candidate devices. When the signal in the main station table is a device fault signal; if it is determined to be a protection device fault, select the measurement and control device with the same interval number, the common measurement and control device, the bus measurement and control device, and the composite virtual device as the alternative devices; if it is determined to be a measurement and control device fault, select the common measurement and control device, the bus measurement and control device, and the composite virtual device as the alternative devices.

[0008] In one embodiment of this application, the master station signal includes the signal description of the master station table, the signal description after synonym rewriting, and the signal information.

[0009] In one embodiment of this application, the step of searching the full vector corresponding to the candidate devices based on the main station vector includes: Semantic retrieval and full-text retrieval matching are performed on the full vector corresponding to the candidate device to obtain the candidate signal.

[0010] In one embodiment of this application, the telemetry forwarding point table generation method further includes: After obtaining the candidate signals, a rearrangement model is used to score the correlation of the candidate signals and obtain the scoring results. The candidate signals are sorted according to the scoring results, and a preset number of candidate signals with the highest scores are retained.

[0011] In one embodiment of this application, the step of constructing prompting information based on substation parameter information, inputting the candidate signals into a matching model, and the matching model determining the target signal matching the main station table from the candidate signals based on the prompting information includes: Extract preset keywords from the main site table and generate device priority rules; Construct a typical example library, and retrieve similar examples from the typical example library by semantics and full text. The substation parameter information, the device priority rules, the similar examples, and the alternative signals are combined to form the structured prompt information. The prompt information is input into the matching model, and the matching model determines the target signal that matches the main site table from the candidate signals based on the prompt information.

[0012] Secondly, embodiments of this application also provide a telemetry forwarding point table generation system, comprising: The preprocessing module is used to obtain the full point table of the substation, perform signal standardization processing on the full point table to obtain a standard signal description, and input the standard signal description and the device description of the full point table into the embedding model to obtain the full vector. The filtering module is used to perform semantic segmentation on the signal descriptions in the master site table to obtain interval information; and to filter out candidate devices from the substation devices based on the interval information and semantic similarity. The recall module is used to vectorize the master station signals in the master station table through the embedding model to obtain the master station vector; and to search in the full vector corresponding to the candidate device according to the master station vector to obtain the candidate signal. The confirmation module is used to construct prompt information based on substation parameter information, input the candidate signals into the matching model, and the matching model determines the target signal that matches the main station table from the candidate signals according to the prompt information.

[0013] Thirdly, embodiments of this application also 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 above-described method for generating a telemetry forwarding point table.

[0014] The beneficial effects of this application are as follows: This application utilizes an embedding model to convert text signals into high-dimensional vectors and achieves cross-semantic matching through vector similarity calculation. Semantic retrieval is performed on the full set of vectors to obtain candidate signals, and a matching model is used to match the target signal from the candidate signals. This solves the problem of mismatch between non-standard descriptions from manufacturers and standard descriptions from the main station, significantly reducing the omission and mismatch rates. This application eliminates the need for manual comparison of each item; the target signal is determined through a matching model, significantly improving the configuration efficiency of the telemetry forwarding point table. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying 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.

[0016] Figure 1 This is a schematic diagram illustrating the steps of the telemetry forwarding point table generation method according to an embodiment of this application; Figure 2 This is a schematic diagram of step S1 in an embodiment of this application; Figure 3 This is a flowchart illustrating step S2 in an embodiment of this application; Figure 4 This is a schematic diagram of step S3 in an embodiment of this application; Figure 5 This is a schematic diagram of the signal processing in step S3 of an embodiment of this application; Figure 6 This is a schematic diagram of step S4 in an embodiment of this application; Figure 7 This is an architecture diagram of the remote telemetry forwarding point table generation system according to an embodiment of this application.

[0017] Explanation of reference numerals in the attached figures: 1. Preprocessing module; 2. Filtering module; 3. Recall module; 4. Confirmation module. Detailed Implementation

[0018] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0019] In the description of this application, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships based on the orientation or positional relationships shown in the accompanying drawings, are used only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this application. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include one or more of the stated features. In the description of this application, "a plurality of" means two or more, unless otherwise explicitly specified.

[0020] like Figure 1 As shown, an embodiment of this application provides a method for generating a telemetry forwarding point table, including the following steps: Step S1: Obtain the full point table of the substation, perform signal standardization processing on the full point table to obtain the standard signal description, and input the standard signal description and the device description of the full point table into the embedding model to obtain the full vector.

[0021] Step S2: Semantically decompose the signal descriptions in the master station table to obtain interval information; select candidate devices from the substation devices based on the interval information and semantic similarity.

[0022] Step S3: Vectorize the master station signal in the master station table using the embedding model to obtain the master station vector; based on the master station vector, search the full vector corresponding to the candidate device to obtain the candidate signal.

[0023] Step S4: Construct prompt information based on substation parameter information, input candidate signals into the matching model, and the matching model determines the target signal that matches the main station table from the candidate signals according to the prompt information.

[0024] This application standardizes the entire point table to unify the description of non-standard signals; it uses an embedding model to convert text signals into high-dimensional vectors and achieves cross-semantic matching through vector similarity calculation, solving the problem of mismatch between the manufacturer's non-standard description and the master station's standard description, and significantly reducing the omission rate and mismatch rate. This application eliminates the need for manual comparison of each item, determining the target signal through a matching model, and significantly improves the efficiency of telemetry forwarding point table configuration.

[0025] In some embodiments, such as Figure 2 As shown, step S1 includes: Step S11: Construct a thesaurus for the power industry and use the thesaurus to convert non-standard signal descriptions in the full-point table into standard signal descriptions.

[0026] Step S12: Extract the bay number corresponding to the substation device from the full data point table, and establish a device bay mapping table corresponding to the substation device and the bay number.

[0027] Step S13: Input the device description and standard signal description into the embedding model, convert them into a full vector through the embedding model, and store them in the vector database.

[0028] Because the total number of signals in the full-scale data point table is very large, and the signal descriptions vary across different manufacturers, this embodiment utilizes a pre-built thesaurus of the power industry to standardize and rewrite the signal descriptions in the substation full-scale data point table. For example, "Cos" is uniformly rewritten as "power factor cos", and "reclosing" is uniformly described as "reclosing action", eliminating the differences in descriptions between different manufacturers.

[0029] By leveraging the extraction capabilities of the Large Language Model (LLM), interval numbers for all devices are extracted from the full point table, and a device-interval mapping table is established to obtain the device-interval mapping table. When subsequently screening candidate devices, the interval information derived from the signal breakdown in the main site table can be used to directly match all devices in the same interval from the device-interval mapping table, improving screening efficiency.

[0030] Using a pre-trained embedding model, such as the bilingual-global-embedded model (Bilingual-Global-Embedding M3, bge-m3) or Conan, the device descriptions and signal texts from the full point table are converted into high-dimensional full vectors, respectively. These full vectors are then stored in a vector database, such as the PostgreSQL vector storage extension module (pgVector) or the Milvus vector database.

[0031] This application effectively eliminates the semantic gap between different substations and the main site table by signal preprocessing and standardized rewriting of the thesaurus, combined with vectorized retrieval of the embedding model, thereby improving the accuracy of signal recall.

[0032] In some embodiments, step S2 involves semantically decomposing the signal description of the master site table, including: using a matching model to parse the signal description of the master site table and semantically decomposing it into interval information and signal information.

[0033] like Figure 3 As shown, step S2 involves selecting candidate devices from the substation equipment based on interval information and semantic similarity, including: When the signal in the main station table is a conventional signal, prompt information is generated based on the substation voltage level to guide the matching model to initially screen substation devices. Then, based on the device bay mapping table, devices in the same bay, common measurement and control devices, and composite signal virtual devices are added as alternative devices.

[0034] When the signal in the main station table is a device fault signal; if it is determined to be a protection device fault, select the measurement and control device with the same interval number, the common measurement and control device, the bus measurement and control device, and the composite virtual device as alternative devices; if it is determined to be a measurement and control device fault, select the common measurement and control device, the bus measurement and control device, and the composite virtual device as alternative devices.

[0035] In step S2, for each signal in the master station table, the description is parsed using a matching model and split into interval information, such as "#1 main transformer" and signal information, such as "non-electrical protection action".

[0036] Routine signals refer to non-fault signals monitored during the daily operation of a substation. When the main station meter signals are routine signals, the extracted bay and signal information, combined with the substation voltage level, generates prompt information, which is then input into the matching model. Based on the prompt information, the matching model initially filters out functionally matched devices from the full range of devices.

[0037] To prevent model illusions or omissions, based on the device interval mapping table established in step S1, this application also supplements the set of physical devices with the same interval number, as well as station-wide common measurement and control devices and composite signal virtual devices as alternative device sets.

[0038] The supplementary devices are combined with the initially screened devices to form a set of candidate devices corresponding to conventional signals.

[0039] Device fault signals refer to alarm signals indicating that the device itself is faulty. When the signal information indicates a fault in the protection device, the protection device itself is unable to upload the fault signal. It is necessary to screen other devices that can collect the fault signal. Specifically, this includes measurement and control devices with the same interval number, common measurement and control devices, bus measurement and control devices, and composite virtual devices.

[0040] When the signal information indicates a fault in the monitoring and control device, the monitoring and control device itself cannot upload the fault signal, and other monitoring and control devices in the same interval are likely to be affected. Therefore, there is no need to supplement the monitoring and control devices in the same interval; only common monitoring and control devices, bus monitoring and control devices, and composite virtual devices are selected.

[0041] This application combines hard rules of interval numbering with soft rules of semantic similarity to screen candidate devices, and selects corresponding candidate device sets for regular signals and fault signals respectively, which significantly improves matching efficiency and reduces interference from irrelevant signals.

[0042] like Figure 4 ,Figure 5 As shown, in step S3, the master station signal of the master station table is vectorized by the embedding model to obtain the master station vector; the master station signal includes the signal description of the master station table (i.e., the original signal), the signal description after synonym rewriting, and the signal information.

[0043] In some embodiments, step S3 includes: Step S31: Perform semantic search and full-text search matching on the full vector corresponding to the candidate device to obtain the candidate signal.

[0044] Step S32: After obtaining the candidate signals, the reordering model is used to score the correlation of the candidate signals and obtain the scoring results.

[0045] Step S33: Sort the candidate signals according to the scoring results and retain a preset number of candidate signals that are ranked first.

[0046] The signal range retrieved in step S3 of this application is not the signal of the full-scale point table of the substation, but the full-scale vector corresponding to the candidate device selected in step S2. By filtering the device range in step S2, the search range is narrowed and the matching efficiency is improved.

[0047] This application utilizes a matching model to perform multi-dimensional processing and feature extraction on the signal descriptions of the scheduling master site table. First, semantic expansion and rewriting are performed based on a thesaurus. The matching model generalizes the original signals using synonyms, for example, rewriting "protection power supply disappears" as "protection power supply disappears (circuit breaker trips / power failure)" to increase the coverage of semantic matching.

[0048] Step S3 uses a matching model to extract device descriptions from the main site table description and extract the core signal description. For example, for "110kV Tongke A Line Measurement and Control Device Time Synchronization Anomaly Alarm," the matching model identifies and removes equipment information such as "110kV Tongke A Line Measurement and Control Device," extracting "Time Synchronization Anomaly Alarm" as the pure signal part. The purpose of this step is that the full site table of a substation often contains a large amount of device prefix information, such as voltage level, bay name, and device type. During vector retrieval, the high-frequency device information can cause weight shifts, obscuring the characteristics of core signals, such as "Time Synchronization Anomaly Alarm," leading to recall failure. This application can significantly correct the attention mechanism of vector retrieval by extracting the pure signal part.

[0049] Finally, the original master site table signal, the synonym rewritten signal, and the extracted pure signal description are input into the embedding model for vectorization processing.

[0050] Using the aforementioned vectors, semantic search and keyword full-text search are performed within the selected candidate devices. A reranking model is then used to score the relevance of the results and perform a secondary ranking. The retrieved results are then merged, and a predetermined number of the top-ranked results are selected as candidate signals.

[0051] In some embodiments, the preset number of items ranked first is, for example, 10, 20, 30, 40, 50, 60, 70, 80, 90, etc.

[0052] The signal flow direction in step S3 of this application is as follows: Figure 5 As shown, the master station table signals issued by the master station are preprocessed, and the master station signals are split into original signals, synonym-rewritten signals, and pure signal information. The signals corresponding to the candidate devices obtained in step S2 are used as candidate signals. Semantic retrieval and full-text retrieval are used to obtain candidate signals, and relevance scoring and secondary sorting are performed. The candidate signals are sorted, and a preset number of candidate signals with the highest ranking are retained.

[0053] The candidate signals obtained in step S3 are input into the matching model. The target signal with the highest matching degree with the main station signal is confirmed through the matching model, thus completing the confirmation of the candidate signal to the target signal.

[0054] like Figure 6 As shown, in some embodiments, step S4 includes: Step S41: Extract preset keywords from the main site table and generate device priority rules.

[0055] Step S42: Construct a typical example library, and retrieve similar examples from the typical example library by semantics and full text.

[0056] Step S43: The substation parameter information, device priority rules, similar examples, and alternative signals are spliced ​​together to form a structured prompt information.

[0057] Step S44: Input the prompt information into the matching model. Based on the prompt information, the matching model determines the target signal that matches the main site table from the candidate signals.

[0058] The system generates priority rules for generating devices based on keywords in the main site table. For example, "maintenance input" prioritizes "maintenance hard plate". The system then combines substation parameter information, priority rules, and the list of recalled alternative signals into a structured prompt message.

[0059] The prompt information is input into the large model, which is constrained to perform logical reasoning. The semantics of the main station signal requirements and the candidate signals are compared, and the signal with the highest matching degree is selected first. The final point table mapping result is then output.

[0060] The matching model in this application is a large language model optimized for the field of power telemetry, such as the DeepSeek large model. Furthermore, this embodiment fine-tunes the existing model by injecting professional corpora from fields such as power dispatching, substation signals, and telemetry communication, which enables the matching model to have a stronger ability to understand power professional terms and match signals, allowing it to more accurately perform steps such as semantic decomposition of master station table signal descriptions and matching target signals.

[0061] Embodiments of this application also provide a telemetry forwarding point table generation system for implementing the above-described telemetry forwarding point table generation method, such as... Figure 7 As shown, the telemetry forwarding point table generation system includes: Preprocessing module 1 is used to obtain the full point table of the substation, perform signal standardization processing on the full point table to obtain standard signal descriptions, and input the standard signal descriptions and device descriptions of the full point table into the embedding model to obtain the full vector.

[0062] The filtering module 2 is used to perform semantic segmentation on the signal descriptions in the master station table to obtain interval information; and to filter out candidate devices from the substation devices based on the interval information and semantic similarity.

[0063] The recall module 3 is used to vectorize the master station signals in the master station table through the embedding model to obtain the master station vector; based on the master station vector, it searches in the full vector corresponding to the candidate device to obtain the candidate signal.

[0064] The confirmation module 4 is used to construct prompt information based on the substation parameter information, input the candidate signals into the matching model, and the matching model determines the target signal that matches the main station table from the candidate signals according to the prompt information.

[0065] Embodiments of this application also provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the telemetry forwarding point table generation method of the above embodiments.

[0066] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0067] The foregoing has provided a detailed description of a method and system for generating a telemetry forwarding point table and a computer device provided in the embodiments of this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for generating a remote telemetry forwarding point table, characterized in that, Includes the following steps: Obtain the full point table of the substation, perform signal standardization processing on the full point table to obtain a standard signal description, and input the standard signal description and the device description of the full point table into the embedding model to obtain the full vector. Semantic segmentation is performed on the signal descriptions in the main station table to obtain interval information; candidate devices are selected from the substation devices based on the interval information and semantic similarity. The main station signal in the main station table is vectorized using the embedding model to obtain the main station vector; based on the main station vector, the candidate signal is obtained by searching the full vector corresponding to the candidate device. Based on the substation parameter information, a prompt message is constructed, and the candidate signals are input into the matching model. The matching model then determines the target signal that matches the main station table from the candidate signals based on the prompt message.

2. The method for generating a telemetry forwarding point table according to claim 1, characterized in that, The signal standardization process for the full point table includes: A thesaurus for the power industry is constructed, and the non-standard signal descriptions in the full-point table are replaced with the standard signal descriptions using the thesaurus.

3. The method for generating a telemetry forwarding point table according to claim 1, characterized in that, The method of inputting the device description of the standard signal description and the full point table into the embedding model includes: Extract the bay number corresponding to the substation device from the full point table, and establish a device bay mapping table corresponding to the substation device and the bay number; The device description and the standard signal description are input into the embedding model, converted into the full vector by the embedding model, and stored in the vector database.

4. The method for generating a telemetry forwarding point table according to claim 3, characterized in that, The semantic decomposition of the signal description in the main site table includes: parsing the signal description in the main site table using the matching model, and semantically decomposing it into interval information and signal information; The step of selecting candidate devices from substation devices based on the interval information and semantic similarity includes: When the signal in the main station table is a conventional signal, prompt information is generated according to the substation voltage level to guide the matching model to initially screen the substation devices, and then according to the device interval mapping table, devices in the same interval, common measurement and control devices, and composite signal virtual devices are added as the candidate devices. When the signal in the main station table is a device fault signal; if it is determined to be a protection device fault, select the measurement and control device with the same interval number, the common measurement and control device, the bus measurement and control device, and the composite virtual device as the alternative devices; if it is determined to be a measurement and control device fault, select the common measurement and control device, the bus measurement and control device, and the composite virtual device as the alternative devices.

5. The method for generating a telemetry forwarding point table according to claim 4, characterized in that, The master station signal includes the signal description in the master station table, the signal description after synonym rewriting, and the signal information.

6. The method for generating a telemetry forwarding point table according to claim 5, characterized in that, The step of searching the full vector corresponding to the candidate devices based on the main station vector includes: Semantic retrieval and full-text retrieval matching are performed on the full vector corresponding to the candidate device to obtain the candidate signal.

7. The method for generating a telemetry forwarding point table according to claim 1, characterized in that, The method for generating the telemetry forwarding point table also includes: After obtaining the candidate signals, a rearrangement model is used to score the correlation of the candidate signals and obtain the scoring results. The candidate signals are sorted according to the scoring results, and a preset number of candidate signals with the highest scores are retained.

8. The method for generating a telemetry forwarding point table according to claim 1, characterized in that, The step of constructing prompt information based on substation parameter information, inputting the candidate signals into a matching model, and the matching model determining the target signal matching the main station table from the candidate signals based on the prompt information includes: Extract preset keywords from the main site table and generate device priority rules; Construct a typical example library, and retrieve similar examples from the typical example library by semantics and full text. The substation parameter information, the device priority rules, the similar examples, and the alternative signals are combined to form the structured prompt information. The prompt information is input into the matching model, and the matching model determines the target signal that matches the main site table from the candidate signals based on the prompt information.

9. A remote telemetry forwarding point table generation system, characterized in that, include: The preprocessing module is used to obtain the full point table of the substation, perform signal standardization processing on the full point table to obtain a standard signal description, and input the standard signal description and the device description of the full point table into the embedding model to obtain the full vector. The filtering module is used to perform semantic segmentation on the signal descriptions in the master site table to obtain interval information; and to filter out candidate devices from the substation devices based on the interval information and semantic similarity. The recall module is used to vectorize the master station signals in the master station table through the embedding model to obtain the master station vector; and to search in the full vector corresponding to the candidate device according to the master station vector to obtain the candidate signal. The confirmation module is used to construct prompt information based on substation parameter information, input the candidate signals into the matching model, and the matching model determines the target signal that matches the main station table from the candidate signals according to the prompt information.

10. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method as described in any one of claims 1 to 8.

Images