CN-122017551-A - State monitoring method and system for gas-insulated high-voltage switch of power distribution system

CN122017551ACN 122017551 ACN122017551 ACN 122017551ACN-122017551-A

Abstract

The application discloses a state monitoring method and system of a gas-insulated high-voltage switch of a power distribution system, wherein the method comprises the steps of collecting state data which is related to the gas-insulated high-voltage switch and comprises multiple pieces of data through a sensor, determining multiple state categories corresponding to the state data, determining multiple pieces of semantic information corresponding to the multiple state categories based on the state data and determining association information among the multiple state categories, determining at least one piece of data corresponding to the semantic information from the multiple pieces of data based on the association information, the semantic information and the multiple pieces of data aiming at each piece of semantic information, carrying out fusion processing on the semantic information and the at least one piece of data to obtain at least one piece of fused data, generating fusion data based on the fused data and the state data which are respectively corresponding to the semantic information, and calling a neural network model to determine a state monitoring result based on the association information and the fusion data, so that the state monitoring accuracy of the gas-insulated high-voltage switch can be improved, and the safety of the power distribution system can be improved.

Inventors

WU XIAOPING
HUANG YAOJIE

Assignees

广州市顺承电气设备有限公司

Dates

Publication Date: 20260512
Application Date: 20260226

Claims (10)

1. A method for monitoring the state of a gas-insulated high-voltage switch of an electrical distribution system, comprising: Collecting state data related to the gas-insulated high-voltage switch through a sensor, and determining a plurality of state categories corresponding to the state data, wherein the state data comprises a plurality of pieces of data; determining a plurality of pieces of semantic information corresponding to the plurality of state categories one by one based on the state data, and determining association information among the plurality of state categories; Determining at least one piece of data corresponding to the semantic information from the multiple pieces of data based on the associated information, the semantic information and the multiple pieces of data aiming at each piece of semantic information, and carrying out fusion processing on the semantic information and the at least one piece of data to obtain at least one piece of fused data corresponding to the semantic information; generating fusion data based on the fused data and the state data which correspond to the semantic information respectively; And calling a neural network model to determine a state monitoring result matched with the gas-insulated high-voltage switch based on the association information and the fusion data.
2. The method of claim 1, wherein determining at least one piece of data corresponding to the semantic information from the plurality of pieces of data based on the association information, the semantic information and the plurality of pieces of data, and performing fusion processing on the semantic information and the at least one piece of data to obtain at least one piece of fused data corresponding to the semantic information, comprises: Determining first cross-attention information based on the semantic information and the association information; For each piece of data in the plurality of pieces of data, determining second cross attention information corresponding to the piece of data based on the semantic information and the piece of data, and determining weight corresponding to the piece of data based on the first cross attention information and the second cross attention information; Determining the at least one piece of data from the plurality of pieces of data based on a preset weight threshold and weights corresponding to the plurality of pieces of data respectively; And based on the weight corresponding to each of the at least one piece of data, carrying out fusion processing on the semantic information and the at least one piece of data to obtain at least one piece of fused data corresponding to the semantic information.
3. The method of claim 2, wherein the step of determining the position of the substrate comprises, The determining the first cross-attention information based on the semantic information and the association information includes: Based on semantic understanding characteristics of the semantic information and associated characteristics of the associated information, a first attention characteristic pair is obtained by using cross attention calculation, and the first attention characteristic pair is input into a denoising model based on a gating mechanism to obtain the first cross attention information output by the denoising model based on the gating mechanism; And/or the number of the groups of groups, The determining, based on the semantic information and the data, second cross attention information corresponding to the data includes: Based on the semantic understanding characteristics of the semantic information and the data characteristics of the data, a second attention characteristic pair is obtained by using cross attention calculation, and the second attention characteristic pair is input into a denoising model based on a gating mechanism to obtain second cross attention information corresponding to the data output by the denoising model based on the gating mechanism.
4. The method of claim 3, wherein the gating mechanism based denoising model comprises a gating network, wherein the gating mechanism based denoising model is configured to: determining, in response to an input pair of attention features, raw difference information between the input pair of attention features; Splicing the input attention characteristic pairs to obtain attention splicing characteristics; Inputting the attention splicing characteristics into the gating network to obtain gating vectors output by the gating network; Element-wise multiplication is performed on the gating vector and the original difference information, and cross attention information corresponding to the input attention feature pair is generated and output.
5. The method of claim 1, wherein the neural network model comprises a first expert model and a second expert model, model parameters of the second expert model being more complex than model parameters of the first expert model; and based on the association information and the fusion data, invoking a neural network model to determine a state monitoring result matched with the gas-insulated high-voltage switch, wherein the method comprises the following steps of: Inputting the association information and the fusion data into the first expert model to obtain first state prediction information output by the first expert model; Inputting the first state prediction information into the second expert model to obtain second state prediction information output by the second expert model; and determining a state monitoring result matched with the gas-insulated high-voltage switch according to the second state prediction information.
6. The method of claim 5, wherein the first expert model comprises a model encoder, an input information construction unit, a first output layer, and P first expert units; The step of inputting the association information and the fusion data into the first expert model to obtain first state prediction information output by the first expert model, including: respectively inputting the association information and the fusion data into the model encoder to obtain association characterization information corresponding to the association information and fusion characterization information corresponding to the fusion data output by the model encoder; Inputting the association characterization information and the fusion characterization information into the input information construction unit to obtain expert unit input information output by the input information construction unit; Inputting the expert unit input information into t first expert units to obtain the output of the t first expert units, wherein the t first expert units are first expert units matched with the expert unit input information in the P first expert units, t is less than or equal to P, and t and P are positive integers respectively; And inputting the output of the t first expert units into the first output layer to obtain the first state prediction information output by the first output layer.
7. The method of claim 6, wherein the second expert model comprises a second output layer and K second expert units, P < K, each of the first expert units corresponding to at least one of the second expert units, and each of the second expert units corresponding to only one of the first expert units; The step of inputting the first state prediction information into the second expert model to obtain second state prediction information output by the second expert model includes: inputting the first state prediction information into s second expert units to obtain output of the s second expert units, wherein the s second expert units are second expert units corresponding to the first state prediction information in T second expert units, the T second expert units are second expert units corresponding to the T first expert units in the K second expert units, and T is less than or equal to T, s is less than or equal to T, and K, s and T are positive integers respectively; And inputting the output of the s second expert units into the second output layer to obtain the second state prediction information output by the second output layer.
8. The method according to any one of claims 5-7, wherein determining a status monitoring result matching the gas-insulated high-voltage switch according to the second status prediction information comprises: matching in a state monitoring auxiliary database associated with the gas-insulated high-voltage switch based on the second state prediction information to obtain at least one state monitoring auxiliary information; Invoking a first pre-training model to determine a state prediction result based on the at least one state monitoring auxiliary information and the second state prediction information; Invoking a second pre-training model to analyze the content of the effective information of the at least one state monitoring auxiliary information based on the at least one state monitoring auxiliary information, the second state prediction information and the state prediction result, so as to obtain an analysis result of the content of the effective information; And determining the state monitoring result based on a preset effective information content condition, the effective information content analysis result, the state monitoring auxiliary database and the state prediction result.
9. The method of claim 8, wherein the determining the status monitoring result based on the preset effective information content condition, the effective information content analysis result, the status monitoring assistance database, and the status prediction result comprises: Updating the at least one state monitoring auxiliary information based on the effective information content analysis result or based on the effective information content analysis result and the state monitoring auxiliary database under the condition that the effective information content analysis result does not meet the effective information content condition, and returning the step of calling a first pre-training model to determine a state prediction result based on the at least one state monitoring auxiliary information and the second state prediction information; And determining the state monitoring result based on the state prediction result under the condition that the effective information content analysis result meets the effective information content condition.
10. A condition monitoring system for a gas-insulated high-voltage switch of a power distribution system, comprising: The data acquisition and processing module is used for acquiring state data related to the gas-insulated high-voltage switch through the sensor and determining a plurality of state categories corresponding to the state data, wherein the state data comprises a plurality of pieces of data; a state data processing module, configured to determine a plurality of pieces of semantic information corresponding to the plurality of state categories one to one based on the state data, and determine association information between the plurality of state categories; The first fusion module is used for determining at least one piece of data corresponding to the semantic information from the multiple pieces of data based on the associated information, the semantic information and the multiple pieces of data aiming at each piece of semantic information, and carrying out fusion processing on the semantic information and the at least one piece of data to obtain at least one piece of fused data corresponding to the semantic information; The second fusion module is used for generating fusion data based on the fused data and the state data which correspond to the semantic information respectively; and the state prediction module is used for calling a neural network model to determine a state monitoring result matched with the gas-insulated high-voltage switch based on the association information and the fusion data.

Description

State monitoring method and system for gas-insulated high-voltage switch of power distribution system Technical Field The application relates to the technical field of state monitoring, in particular to a state monitoring method and system of a gas-insulated high-voltage switch of a power distribution system. Background The gas-insulated high-voltage switchgear is a modular switchgear suitable for a power distribution system (e.g., a 12KV power distribution system), and can use insulating gas (e.g., SF6 gas) as an insulating medium to improve the safety performance of the power distribution system. However, when the gas-insulated high-voltage switchgear itself is abnormal and is difficult to operate normally, the safety performance of the related power distribution system is easily reduced. Therefore, how to accurately monitor whether the state of the gas-insulated high-voltage switchgear is abnormal is important to improve the safety performance of the power distribution system. Disclosure of Invention In order to solve the technical problems, the embodiment of the application provides a state monitoring method and a state monitoring system for a gas-insulated high-voltage switch of a power distribution system, which can improve the state monitoring accuracy of the gas-insulated high-voltage switch so as to improve the safety performance of the power distribution system. In a first aspect, an embodiment of the present application provides a method for monitoring a state of a gas-insulated high-voltage switch of a power distribution system, including: Collecting state data related to the gas-insulated high-voltage switch through a sensor, and determining a plurality of state categories corresponding to the state data, wherein the state data comprises a plurality of pieces of data; determining a plurality of pieces of semantic information corresponding to the plurality of state categories one by one based on the state data, and determining association information among the plurality of state categories; Determining at least one piece of data corresponding to the semantic information from the multiple pieces of data based on the associated information, the semantic information and the multiple pieces of data aiming at each piece of semantic information, and carrying out fusion processing on the semantic information and the at least one piece of data to obtain at least one piece of fused data corresponding to the semantic information; generating fusion data based on the fused data and the state data which correspond to the semantic information respectively; And calling a neural network model to determine a state monitoring result matched with the gas-insulated high-voltage switch based on the association information and the fusion data. Optionally, the determining at least one piece of data corresponding to the semantic information from the multiple pieces of data based on the association information, the semantic information and the multiple pieces of data, and performing fusion processing on the semantic information and the at least one piece of data to obtain at least one piece of fused data corresponding to the semantic information includes: Determining first cross-attention information based on the semantic information and the association information; For each piece of data in the plurality of pieces of data, determining second cross attention information corresponding to the piece of data based on the semantic information and the piece of data, and determining weight corresponding to the piece of data based on the first cross attention information and the second cross attention information; Determining the at least one piece of data from the plurality of pieces of data based on a preset weight threshold and weights corresponding to the plurality of pieces of data respectively; And based on the weight corresponding to each of the at least one piece of data, carrying out fusion processing on the semantic information and the at least one piece of data to obtain at least one piece of fused data corresponding to the semantic information. Optionally, the determining the first cross-attention information based on the semantic information and the association information includes: Based on semantic understanding characteristics of the semantic information and associated characteristics of the associated information, a first attention characteristic pair is obtained by using cross attention calculation, and the first attention characteristic pair is input into a denoising model based on a gating mechanism to obtain the first cross attention information output by the denoising model based on the gating mechanism; And/or the number of the groups of groups, The determining, based on the semantic information and the data, second cross attention information corresponding to the data includes: Based on the semantic understanding characteristics of the semantic information and the data characteristics of the data, a seco