CN-121995273-A - Fault identification method based on ultrahigh frequency characteristic change

CN121995273ACN 121995273 ACN121995273 ACN 121995273ACN-121995273-A

Abstract

The application discloses a fault identification method based on ultrahigh frequency characteristic change, which relates to the technical field of transformer fault identification, and comprises the steps of predicting the fault probability of a target transformer and determining a fault identification aging coefficient; the method comprises the steps of carrying out data value evaluation on a sensor deployed on a target transformer according to working operation information and regional environment information, carrying out value evaluation on spectrum characteristic parameters to obtain a key sensor group and a key spectrum characteristic parameter group, screening an ultrahigh frequency electromagnetic wave signal sequence set to obtain a key ultrahigh frequency electromagnetic wave signal sequence set, returning a comprehensive ultrahigh frequency electromagnetic wave signal sequence set through a UHF sensor to carry out fault fine diagnosis, and outputting a fault identification result. The method solves the problems of insufficient timeliness and accuracy of transformer fault identification caused by the fact that an adaptive signal monitoring identification scheme cannot be set according to a transformer operation scene in the method for identifying transformer faults based on the ultrahigh frequency characteristics in the prior art.

Inventors

WANG QING
WANG GUOCHUN
XU LINGLING
CHEN ZHENGGUANG
TIAN HUIDONG
WANG TONGLEI
DING CHENG
WANG XIAOHUA
SU HAOYU

Assignees

国家电网有限公司
四川鸿华睿橙智能电气有限公司

Dates

Publication Date: 20260508
Application Date: 20251230

Claims (8)

1. The fault identification method based on the characteristic change of the ultrahigh frequency is characterized by comprising the following steps of: Predicting the fault probability of the target transformer according to the working operation information and the regional environment information of the target transformer in the historical time window, performing fault identification timeliness analysis based on the predicted fault probability in a preset time zone, and determining a fault identification timeliness coefficient; performing data value evaluation on a plurality of UHF sensors deployed on a target transformer according to the working operation information and the regional environment information, performing parameter value evaluation on preset spectrum characteristic parameters, and screening to obtain a key sensor group and a key spectrum characteristic parameter group based on the fault identification aging coefficient; in the preset time zone, the ultrahigh frequency electromagnetic wave signal is acquired according to the key sensor group and the key frequency spectrum characteristic parameter group, a key data screening scheme is set based on the fault identification aging coefficient, and the ultrahigh frequency electromagnetic wave signal sequence set is screened to obtain a key ultrahigh frequency electromagnetic wave signal sequence set; And performing fault primary diagnosis according to the key ultrahigh frequency electromagnetic wave signal sequence set, returning a comprehensive ultrahigh frequency electromagnetic wave signal sequence set through the UHF sensors to perform fault fine diagnosis if the target transformer is abnormal, and outputting a fault recognition result.
2. The method for recognizing faults based on characteristic change of ultrahigh frequency according to claim 1, wherein the method for predicting the fault probability of the target transformer according to the working operation information and the regional environment information of the target transformer in the historical time window, performing the fault recognition timeliness analysis based on the predicted fault probability in the preset time zone, and determining the fault recognition timeliness coefficient comprises the following steps: Monitoring and acquiring a working operation data sequence set and a regional environment data sequence set of a target transformer in a historical time window, and uploading the working operation data sequence set and the regional environment data sequence set to a cloud server to serve as working operation information and regional environment information, wherein the working operation data at least comprise load current, voltage, oil temperature and winding temperature, and the regional environment data at least comprise electromagnetic interference intensity, relative temperature and environmental humidity; In the cloud server, predicting the fault probability of the target transformer in a preset time zone according to the working operation information and the regional environment information by using a fault probability predictor, and outputting a predicted fault probability; and setting the ratio of the preset fault probability scalar of the target transformer to the predicted fault probability as a fault identification aging coefficient.
3. The fault identification method based on the characteristic change of the ultrahigh frequency according to claim 2, wherein the construction method of the fault probability predictor comprises the following steps: Based on the historical operation records of similar transformers of the target transformer, collecting a sample working operation information set and a sample area environment information set, and collecting the fault event duty ratio of different sample working operation information and sample area environment information in a historical time zone, wherein the fault event duty ratio is used as sample fault probability, and a sample fault probability set is obtained; And training a long-time and short-time memory network until convergence by taking the sample working operation information set and the sample area environment information set as inputs and taking the sample fault probability set as supervision to obtain a fault probability predictor, and deploying the fault probability predictor on a cloud server.
4. The fault identification method based on the characteristic change of the ultrahigh frequency according to claim 1, wherein the data value evaluation of a plurality of UHF sensors deployed on a target transformer and the parameter value evaluation of a preset frequency spectrum characteristic parameter are performed according to the working operation information and the regional environment information, and the fault identification method comprises the following steps: Disposing a plurality of UHF sensors at a plurality of key positions of the target transformer, wherein the UHF sensors are built-in UHF sensors or external UHF sensors, and the number of the sensors is not less than 5; Based on the historical operation records of similar transformers, respectively analyzing the relevance between the key positions and the faults of the transformers, and outputting a plurality of fault relevance coefficients; based on a plurality of key positions and a plurality of sensor attribute information of the plurality of UHF sensors, respectively carrying out data interference intensity simulation on the plurality of UHF sensors according to the working operation information and the regional environment information, and outputting a plurality of data credible coefficients; Respectively evaluating the data value of the UHF sensors according to the fault correlation coefficients and the data credibility coefficients to obtain data value coefficients, wherein the data value coefficients are positively correlated with the fault correlation coefficients and the data credibility coefficients; Based on the historical operation records of similar transformers, respectively analyzing the relevance of a plurality of frequency spectrum characteristic parameters in the preset frequency spectrum characteristic parameters and transformer fault identification to obtain a plurality of parameter relevance coefficients; respectively carrying out parameter reliability evaluation on the plurality of spectrum characteristic parameters according to the working operation information and the regional environment information, and outputting a plurality of parameter reliability coefficients; And carrying out parameter value evaluation based on the plurality of parameter association coefficients and the plurality of parameter credible coefficients, and outputting a plurality of parameter value coefficients, wherein the parameter value coefficients are positively correlated with the parameter association coefficients and the parameter credible coefficients.
5. The method of claim 4, wherein the predetermined spectral feature parameters include dominant frequency location, bandwidth range, power spectral density, energy distribution characteristics, and spectral centroid.
6. The method for recognizing faults based on the characteristic change of the ultrahigh frequency according to claim 4, wherein the key sensor group and the key frequency spectrum characteristic parameter group are obtained based on the fault recognition aging coefficient screening, and the method comprises the following steps: Multiplying and rounding the fault identification aging coefficient with the initial sensor number and the initial frequency spectrum characteristic parameter number respectively to obtain the key sensor number and the key characteristic parameter number, wherein the initial sensor number is one half of the deployment sensor number, the initial frequency spectrum characteristic parameter number is 3, the key sensor number is more than or equal to 2, and the key characteristic parameter number is more than or equal to 2; Screening from large to small according to the number of the key sensors based on the data value coefficients to obtain a plurality of key sensors to construct a key sensor group; And screening according to the number of the key characteristic parameters from large to small based on the plurality of parameter value coefficients to obtain a key frequency spectrum characteristic parameter set.
7. The method for recognizing faults based on characteristic change of ultrahigh frequency according to claim 1, wherein the steps of collecting ultrahigh frequency electromagnetic wave signals according to the key sensor group and the key frequency spectrum characteristic parameter group, setting a key data screening scheme based on the fault recognition aging coefficient, and screening the ultrahigh frequency electromagnetic wave signal sequence set to obtain the key ultrahigh frequency electromagnetic wave signal sequence set comprise the following steps: activating a corresponding UHF sensor according to the key sensor group to collect the UHF electromagnetic wave signals in the preset time zone, and screening the collection result according to the key frequency spectrum characteristic parameter group at an edge processing unit of a target transformer to obtain an UHF electromagnetic wave signal sequence set; Taking the product of the fault identification aging coefficient and a preset data extraction ratio as an adaptive data extraction ratio, wherein the preset data extraction ratio is 50%; And performing data similarity dimension reduction on the ultrahigh frequency electromagnetic wave signal sequence set according to the adaptive data extraction proportion to obtain a key ultrahigh frequency electromagnetic wave signal sequence set, and uploading the key ultrahigh frequency electromagnetic wave signal sequence set to a cloud server.
8. The fault identification method based on the characteristic change of the ultrahigh frequency according to claim 1, wherein the fault primary diagnosis is performed according to the key ultrahigh frequency electromagnetic wave signal sequence set, if the target transformer is abnormal, the fault fine diagnosis is performed by returning the comprehensive ultrahigh frequency electromagnetic wave signal sequence set through the plurality of UHF sensors, and the fault identification result is output, comprising: Traversing and judging the key ultrahigh frequency electromagnetic wave signal sequence set based on a preset signal early warning threshold, and continuously monitoring according to the key sensor group and the key spectrum characteristic parameter group if the target transformer is not abnormal; If the target transformer is abnormal, activating the UHF sensors to acquire the UHF electromagnetic wave signals, acquiring a comprehensive UHF electromagnetic wave signal sequence set, and uploading the comprehensive UHF electromagnetic wave signal sequence set to the cloud server; and in the cloud server, performing fault fine diagnosis on the comprehensive ultrahigh frequency electromagnetic wave signal sequence set by using a transformer fault recognition model, and outputting a fault recognition result, wherein the transformer fault recognition model is constructed based on a transformer fault mechanism knowledge graph and a graph neural network.

Description

Fault identification method based on ultrahigh frequency characteristic change Technical Field The application relates to the technical field of transformer fault identification, in particular to a fault identification method based on ultrahigh frequency characteristic change. Background With the continuous development of power systems, transformers are used as core equipment for power transmission and distribution, and the stability and reliability of operation of transformers are of great importance. However, the traditional transformer fault identification method mainly relies on periodic manual inspection and off-line test, has the problems of long detection period, poor real-time performance and the like, and is difficult to meet the requirements of a modern power system on real-time monitoring of the running state of the transformer and rapid fault diagnosis. Disclosure of Invention The embodiment of the application solves the technical problems of insufficient timeliness and accuracy of transformer fault identification caused by the fact that an adaptive signal monitoring identification scheme cannot be set according to a transformer operation scene in the method for identifying transformer faults based on the ultrahigh frequency characteristics in the prior art by providing the fault identification method based on the ultrahigh frequency characteristics. The technical scheme for solving the technical problems is as follows: In a first aspect, the present application provides a fault identification method based on characteristic change of a ultrahigh frequency, the method comprising: Predicting the fault probability of the target transformer according to the working operation information and the regional environment information of the target transformer in the historical time window, performing fault identification timeliness analysis based on the predicted fault probability in a preset time zone, and determining a fault identification timeliness coefficient; performing data value evaluation on a plurality of UHF sensors deployed on a target transformer according to the working operation information and the regional environment information, performing parameter value evaluation on preset spectrum characteristic parameters, and screening to obtain a key sensor group and a key spectrum characteristic parameter group based on the fault identification aging coefficient; in the preset time zone, the ultrahigh frequency electromagnetic wave signal is acquired according to the key sensor group and the key frequency spectrum characteristic parameter group, a key data screening scheme is set based on the fault identification aging coefficient, and the ultrahigh frequency electromagnetic wave signal sequence set is screened to obtain a key ultrahigh frequency electromagnetic wave signal sequence set; And performing fault primary diagnosis according to the key ultrahigh frequency electromagnetic wave signal sequence set, returning a comprehensive ultrahigh frequency electromagnetic wave signal sequence set through the UHF sensors to perform fault fine diagnosis if the target transformer is abnormal, and outputting a fault recognition result. The application provides one or more technical schemes, which at least have the following technical effects or advantages: According to the fault identification method based on the ultrahigh frequency characteristic change, firstly, fault probability prediction and timeliness analysis are carried out according to the working operation and the regional environment information of the transformer, and the timeliness coefficient of fault identification is determined, so that the follow-up fault identification is more attached to the actual operation scene of the transformer, and the timeliness of identification is improved. Secondly, the UHF sensor and the frequency spectrum characteristic parameters are subjected to value evaluation, key groups are screened out, key data can be acquired pertinently, interference of invalid data is avoided, and the accuracy of the data is improved. And then, a key sequence set is obtained by screening the ultrahigh frequency electromagnetic wave signal sequence set, so that the data processing amount is reduced, and the fault recognition speed is increased. Finally, the primary fault diagnosis is carried out, and if the fault exists, the fine diagnosis is carried out, so that the comprehensiveness of fault identification is ensured, and the accuracy of the identification is improved. Through the technical scheme, the aging coefficient is determined through predicting the fault probability, the key sensor group and the key frequency spectrum characteristic parameter group are screened out, and the collected signals are reasonably screened and analyzed, so that the transformer fault can be identified efficiently and accurately. According to the actual operation scene of the transformer, the signal monitoring and identifying scheme is flexibly adj