CN117131786A - Voltage transformer insulation fault online identification method - Google Patents
Voltage transformer insulation fault online identification method Download PDFInfo
- Publication number
- CN117131786A CN117131786A CN202311396755.3A CN202311396755A CN117131786A CN 117131786 A CN117131786 A CN 117131786A CN 202311396755 A CN202311396755 A CN 202311396755A CN 117131786 A CN117131786 A CN 117131786A
- Authority
- CN
- China
- Prior art keywords
- fault
- data set
- target domain
- insulation
- voltage
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000009413 insulation Methods 0.000 title claims abstract description 98
- 238000000034 method Methods 0.000 title claims abstract description 59
- 239000000203 mixture Substances 0.000 claims abstract description 34
- 230000003321 amplification Effects 0.000 claims abstract description 14
- 238000003199 nucleic acid amplification method Methods 0.000 claims abstract description 14
- 238000004088 simulation Methods 0.000 claims abstract description 12
- 238000012549 training Methods 0.000 claims abstract description 7
- 230000015556 catabolic process Effects 0.000 claims description 34
- 239000003990 capacitor Substances 0.000 claims description 23
- 238000009826 distribution Methods 0.000 claims description 21
- 230000006870 function Effects 0.000 claims description 15
- 238000013508 migration Methods 0.000 claims description 15
- 230000005012 migration Effects 0.000 claims description 13
- 238000000513 principal component analysis Methods 0.000 claims description 11
- 238000005070 sampling Methods 0.000 claims description 10
- 239000010410 layer Substances 0.000 claims description 9
- 238000000605 extraction Methods 0.000 claims description 6
- 238000012847 principal component analysis method Methods 0.000 claims description 4
- 239000011229 interlayer Substances 0.000 claims description 3
- 238000013459 approach Methods 0.000 claims 1
- 238000013480 data collection Methods 0.000 claims 1
- 238000009795 derivation Methods 0.000 claims 1
- 238000012423 maintenance Methods 0.000 abstract description 3
- 238000010586 diagram Methods 0.000 description 12
- 238000004590 computer program Methods 0.000 description 7
- 238000012545 processing Methods 0.000 description 7
- 230000008569 process Effects 0.000 description 5
- 238000012986 modification Methods 0.000 description 4
- 230000004048 modification Effects 0.000 description 4
- 238000011161 development Methods 0.000 description 3
- 230000018109 developmental process Effects 0.000 description 3
- 238000009421 internal insulation Methods 0.000 description 3
- 230000007774 longterm Effects 0.000 description 3
- 241000894007 species Species 0.000 description 3
- 238000003860 storage Methods 0.000 description 3
- 230000004075 alteration Effects 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 244000132059 Carica parviflora Species 0.000 description 1
- 235000014653 Carica parviflora Nutrition 0.000 description 1
- 230000032683 aging Effects 0.000 description 1
- 238000007635 classification algorithm Methods 0.000 description 1
- 230000008094 contradictory effect Effects 0.000 description 1
- 230000006866 deterioration Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000005284 excitation Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000011430 maximum method Methods 0.000 description 1
- 238000012806 monitoring device Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000013526 transfer learning Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R35/00—Testing or calibrating of apparatus covered by the other groups of this subclass
- G01R35/02—Testing or calibrating of apparatus covered by the other groups of this subclass of auxiliary devices, e.g. of instrument transformers according to prescribed transformation ratio, phase angle, or wattage rating
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/12—Testing dielectric strength or breakdown voltage ; Testing or monitoring effectiveness or level of insulation, e.g. of a cable or of an apparatus, for example using partial discharge measurements; Electrostatic testing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- Economics (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Power Engineering (AREA)
- Software Systems (AREA)
- Medical Informatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computer Hardware Design (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention provides an online identification method for insulation faults of a voltage transformer, which is used for collecting data of all-class fault states of an insulation fault simulation platform as a source domain fault data set; collecting incomplete data of the insulating state fault type in the transformer substation site as a target domain few-label fault data set; based on the connection between the insulation fault of the voltage transformer and the characteristic data of the voltage transformer, the category label completion of the target domain few-label characteristic data set is realized, and the data amplification of the full category fault state is carried out on the target domain few-label characteristic data set based on a Gaussian mixture model; and migrating the two data sets, training based on the migrated source domain characteristic data sets, and identifying the insulation faults of the transformer substation data by using a trained identification model. According to the invention, power failure maintenance is not needed, the insulation faults of all types of voltage transformers of the transformer substation can be identified with high accuracy only by using a sample data set with incomplete insulation fault types of the transformer substation, the running state of the voltage transformers is mastered in real time, and the guarantee is provided for the stable running of a power system.
Description
Technical Field
The invention relates to the field of power equipment fault state evaluation, in particular to an online identification method for insulation faults of a voltage transformer.
Background
The running stability and the development reliability of the power system are related to national lives, the running stability and the development reliability are important bases for the national economy development, and the voltage transformer is used as high-voltage measuring equipment widely applied in the power system, and the output voltage is an important application basis for the electric energy metering and the state monitoring of the power system.
The voltage transformer exposes important problems in the long-term operation process, and factors such as long-term electrothermal aging, primary overvoltage and the like can cause the deterioration of the internal insulation performance of the voltage transformer, so that the metering accuracy and the long-term operation stability of the voltage transformer are seriously affected.
The traditional voltage transformer insulation state detection method is fixed-period off-line detection, and the insulation operation state of the voltage transformer cannot be mastered in time due to the fact that planned power failure is needed. Therefore, in order to get rid of the constraint of the planned power outage cooperation, patent CN207832951U proposes an insulation on-line monitoring device for a capacitive voltage transformer, and the insulation states of all parts of the capacitive voltage transformer are obtained by comparing waveform changes through an external waveform analyzer, but are easily affected by electromagnetic interference. Because the secondary output voltage of the transformer substation voltage transformer is low in quality and has less fault data with labels, the requirement of applying a machine learning intelligent classification algorithm is difficult to meet, and therefore, the patent CN113447783B provides a voltage transformer insulation fault identification model which realizes insulation state identification of the voltage transformer by simulating the insulation fault of the voltage transformer in a transformer substation environment in a laboratory through a transfer learning and supervision learning method, but the types of faults which can be acquired in the actual operation process of the transformer substation are less, only the insulation faults which have occurred and are acquired in a transformer substation site can be identified, and the application range is narrow.
Disclosure of Invention
The invention provides an on-line identification method for insulation faults of a voltage transformer, aiming at the technical problems in the prior art, comprising the following steps:
step S1, taking voltage transformers VT of the same voltage class as a group, and taking secondary voltage data as a source domain fault data set when the insulating state of a VT insulating fault simulation platform is normal and taking secondary voltage data as a source domain fault data set when the insulating state of the whole class is fault; collecting secondary voltage data in a transformer substation site when the insulation state is normal as a fault-free data set of a target domain, and collecting secondary voltage data with incomplete insulation state fault types in the transformer substation site as a fault data set of fewer labels of the target domain;
step S2, carrying out feature extraction on a source domain non-fault data set and a source domain fault data set to obtain a source domain feature data set, and carrying out feature extraction on a target domain non-fault data set and a target domain less-label fault data set to obtain a target domain feature data set;
step S3, calculating expectations and variances of characteristic data under various insulation state faults of a target domain based on knowledge deduction of the relation between the insulation faults of the voltage transformer and the characteristic data of the voltage transformer, and realizing class label completion of a few-label characteristic data set of the target domain;
s4, establishing a Gaussian mixture model of the target domain few-label feature data set by using an Expectation Maximization (EM), obtaining Gaussian mixture distribution of the target domain full-class labels according to expectations and variances of feature data under each class of insulation state faults in the step S3, carrying out simple random sampling on the Gaussian mixture distribution, and amplifying the target domain few-label fault data set to form a target domain full-label feature data set;
and S5, migrating the source domain characteristic dataset and the target domain full-label characteristic dataset by using a migration learning method to obtain a migrated source domain characteristic dataset, training the migrated source domain characteristic dataset by using a supervision learning method, and identifying the insulation fault of the migrated target domain characteristic dataset by using a trained identification model.
According to the online identification method for the insulation faults of the voltage transformer, provided by the invention, the insulation faults of all types of the voltage transformer of the transformer substation can be identified with high accuracy only by using the sample data set with incomplete types of the insulation faults of the transformer substation without power failure maintenance, the running state of the voltage transformer is mastered in real time, and the guarantee is provided for the stable running of a power system.
Drawings
Fig. 1 is a schematic flow chart of a voltage transformer insulation fault identification method according to an embodiment of the present invention;
fig. 2 is a schematic diagram of an overall flow of a voltage transformer insulation fault identification method provided by the invention;
FIG. 3a is a topology diagram of a wiring structure of an L-group voltage transformer simulated by a voltage transformer fault simulation platform;
fig. 3b is a topology diagram of the wiring structure of the L-group voltage transformer at the substation site;
fig. 4 is a schematic diagram of the equivalent circuit model structure of the voltage transformer CVT.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention. In addition, the technical features of each embodiment or the single embodiment provided by the invention can be combined with each other at will to form a feasible technical scheme, and the combination is not limited by the sequence of steps and/or the structural composition mode, but is necessarily based on the fact that a person of ordinary skill in the art can realize the combination, and when the technical scheme is contradictory or can not realize, the combination of the technical scheme is not considered to exist and is not within the protection scope of the invention claimed.
The invention provides a voltage transformer insulation fault on-line identification method flow chart, as shown in fig. 1 and 2, comprising the following steps:
step S1, taking voltage transformers VT of the same voltage class as a group, and collecting a large amount of secondary voltage data as a source domain fault-free data set when the insulating state of a VT insulating fault simulation platform is normal, and collecting a large amount of secondary voltage data as a source domain fault data set when the insulating state of the full class is fault; and collecting a large amount of secondary voltage data in the transformer substation site when the insulation state is normal as a fault-free data set of the target domain, and a small amount of secondary voltage data with incomplete insulation state fault types in the transformer substation site as a fault data set of the target domain and a small number of labels.
The step S1 specifically includes:
s101: the three-phase secondary side voltage amplitude signals of the VT insulation fault simulation platform L groups VT are acquired in real time through the data acquisition device, wherein the topological diagram of the wiring structure of the voltage transformer of the VT insulation fault simulation platform can be seen in fig. 3a, the acquired data comprise secondary voltage data of normal insulation state and insulation state faults, a source domain fault-free data set and a source domain fault data set are respectively formed, the insulation faults in the source domain fault data set are multiple types of faults, and the voltage transformer comprisesInterlayer breakdown failure>Turn-to-turn breakdown fault and capacitive divider high and medium voltage +.>And the capacitor breakdown faults of the capacitor units are comprehensive in fault type, and L, n is a positive integer.
S102: the three-phase secondary side voltage amplitude signals of the L-group VT of the transformer substation site are acquired in real time through the data acquisition device, the topological diagram of the wiring structure of the voltage transformer of the transformer substation site can be seen in fig. 3b, the data acquired at the transformer substation site comprises a large amount of secondary voltage data with normal insulation state and a small amount of secondary voltage data with incomplete edge state fault types, a target domain non-fault data set and a target domain few-label fault data set are respectively formed, and the insulation faults in the target domain few-label fault data set are few-type faults and comprise the voltage transformerLayer by layerBreakdown failure between(s),>turn-to-turn breakdown fault and high-medium voltage of capacitive voltage dividerIndividual capacitor cell breakdown failure etc +.>Can take->Any one of the values) that is, the target domain few-label fault dataset includes a small amount of secondary voltage data complemented by the fault category label.
And S2, carrying out feature extraction on the source domain non-fault data set and the source domain fault data set to obtain a source domain feature data set, and carrying out feature extraction on the target domain non-fault data set and the target domain less-label fault data set to obtain a target domain feature data set.
The step S2 specifically includes:
s201, taking three-phase secondary voltage data of the source domain fault-free data set in S1A principal component analysis model M1 is respectively established by using a principal component analysis method to obtain a source domain fault-free residual error information data set +.>The method comprises the steps of carrying out a first treatment on the surface of the Three-phase data of a source domain fault data set in S1 are taken>Substituting M1 to obtain source domain fault residual information data setCombining the source domain fault-free residual information data set and the source domain fault residual information data set into a source domain feature data set +.>。
S202, taking three-phase data in the fault-free data set of the target domain in S1A principal component analysis method is used for respectively establishing principal component analysis models M2 to obtain a target domain fault-free residual error information data set +.>The method comprises the steps of carrying out a first treatment on the surface of the Three-phase data of a target domain few-label fault data set in S1 are taken +.>Substituting M2 to obtain target domain fault residual information data setCombining the target domain fault-free residual information data set and the target domain few-label fault residual information data set into a target domain characteristic data set +.>。
And step S3, calculating expectations and variances of the characteristic data under each type of insulation state faults of the target domain based on knowledge deduction of the relation between the insulation faults of the voltage transformer and the characteristic data, and realizing class label completion of the few-label characteristic data set of the target domain.
The step S3 specifically includes:
s301, recording source domain characteristic data set(in this case, for example, phase A data, the data processing of the other phases is identical to phase A) contains ∈>Failure of species category, representing L groups VT occurring +.>Species category breakdown->Individual (layer),Turns), ith group VT occurs +.>The source domain feature data set corresponding to the insulation fault and breakdown c (layers, turns) is recorded as +.>If the 1 st insulating fault occurs in the 1 st VT and 1 (layer, turn) corresponding source domain characteristic data set is punctured, the data set is recorded as +.>If no fault is recorded as。
S302, recording a target domain feature data set(for example, phase A data) together include +.>Category faults, representing a certain group of VT occurrences +.>Breakdown of species category->（/>Can take->Any one of the values) are (layers, turns) the ith group VT occurs +.>Insulation fault and breakdown->The corresponding target domain feature data set is denoted +.>No fault is marked as +.>。
S303, calculating a characteristic data set containing the insulation fault type in the target domain by using the statistical relationshipIs +.>And variance->. Based on the calculated expected value +.>And variance->The physical characteristics of the insulation faults of the voltage transformer are combined, and the function relation between expected values and variances of the target domain characteristic data set is obtainedAnd->And calculating expected values and variances of the characteristic data sets of the other various types of insulation faults which are not included in the target domain according to the functional relation. The above description is given by taking the phase a data as an example, and the processing procedure of the phase B and the phase C data is the same as that of the phase a. And (3) obtaining expected values and variances of other data feature sets without fault class labels in the target domain through the processing of the step (S3), and realizing the amplification of the fault class labels in the feature data set with few labels in the target domain.
And S4, establishing a Gaussian mixture model of the target domain few-tag feature data set by using an Expectation Maximization (EM), obtaining Gaussian mixture distribution of the target domain full-class tag according to the expectation and variance of the feature data under each type of insulation state fault in the step S3, performing simple random sampling on the Gaussian mixture distribution, and amplifying the target domain few-tag fault data set to form the target domain full-tag feature data set.
The step S4 specifically includes:
s401, calculating insulation fault categories contained in target domain feature data set by using expected maximum method EMThe gaussian mixture model probability density function of (c) is:
then there are:
wherein k is the number of normal distribution models forming the Gaussian mixture model,is the kth normal distribution model in the Gaussian mixture model, +.>Is the expectation of the kth normal distribution model, +.>Variance of kth normal distribution model, +.>Is the weight of the kth normal distribution model.
From characteristic datasetIs calculated by combining the expected value and variance of formula (19)>And->。
Then for the AND that the target domain feature dataset does not contain a failure category labelFeature data sets of the same fault group as the fault category but different breakdown numbers +.>The gaussian mixture model probability density function of (c) is:
then there are:
where v=1,..n, n is a positive integer and v+.m.
For the AND without fault class labels in the target domain feature datasetCharacteristic data set with same fault category and breakdown number but different fault groups +.>The gaussian mixture model probability density function of (c) is:
then there are:
wherein the method comprises the steps of。/>Are all a K x L matrix.
S402, combining expected values and standard deviations of feature data sets of each insulation fault class of the target domain after the fault class amplification in S3, obtaining:
（13）
（15）
wherein the method comprises the steps of。
Equation (13) givesAnd->And +.>Relation between them, according to formula (1)>Can obtain +.>. Formula (15) gives +.>And->And->Etc., according to the formula (1)>Can obtain +.>. Find->And->According to formula (5), the +.>。
S403, combining expected values and standard deviations of feature data sets of each insulation fault class of the target domain after the fault class amplification in S3, obtaining:
(16)
(17)
(18)
wherein the method comprises the steps of。
S404, recordThere are r samples, with +.>Simple random sampling of r samples is done +.>Data amplification of (2) with ∈ ->Simple random sampling of r samples is done +.>Is a data amplification of (a). After data amplification, a new target domain characteristic data set is formed>Comprises->A class of faults. The other phases repeat the operation S4 described above.
Wherein, for example, the data already comprising fault class labels in the target domain feature data set comprises r, then the following is utilizedSimple random sampling is also performed to obtain +.>Of the fault classr samples, complete->Is a data amplification of (a). Likewise, use->Simple random sampling of r samples is done +.>Is a data amplification of (a). />Post-amplification->Andtarget Domain feature dataset comprising a New full class Label +.>。
And S5, migrating the source domain characteristic dataset and the target domain full-label characteristic dataset by using a migration learning method to obtain a migrated source domain characteristic dataset, training the migrated source domain characteristic dataset by using a supervision learning method, and identifying the insulation fault of the migrated target domain characteristic dataset by using a trained identification model.
The step S5 specifically includes:
s501, using a migration learning method to perform source domain feature data setAnd a new target domain feature dataset +.>Performing migration learning to obtain a migrated source domain feature data set +.>And the target domain characteristic data set after migration +.>. The other phases repeat the above steps.
S502, training the migrated source domain feature data set by using a supervised learning methodAnd extracting residual information from newly collected fault data in the transformer substation site through principal component analysis, substituting the residual information into the migration learning model constructed in the step S501, and substituting the migrated data into a trained supervised learning method to finish identification of the insulation fault.
The present example will now be described in connection with specific application examples.
The topological structures of the capacitive voltage transformers of the 6 groups of the 110kV insulation fault simulation platform and the capacitive voltage transformers of the 6 groups of the same bus of 220kV of a certain transformer substation are shown in fig. 3a and 3b.
The implementation steps of the method are shown in fig. 2:
1) And collecting CVT secondary voltage data of each group of the insulation fault simulation platform and the transformer substation, wherein the CVT secondary voltage data comprises fault-free data and insulation fault data. The fault-free data are secondary voltage data when the insulating state of each group of CVT is normal. The insulation fault simulation platform totally comprises 2 multiplied by 6 insulation fault types, whereinRepresenting that each group of CVTA phases respectively generate 1 breakdown of the high-voltage capacitor, 2 breakdown of the high-voltage capacitor, 1 breakdown of the medium-voltage capacitor and 2 breakdown of the medium-voltage capacitor. The substation contains 2 insulation fault types in total, wherein a first group of CVTA phases respectively generate 1 breakdown of a high-voltage capacitor and 1 breakdown of a medium-voltage capacitor.
The number of source domain secondary voltage samples collected and the corresponding labels are shown in table 1.
TABLE 1 Source Domain Secondary Voltage sample count and corresponding Label
The number of target domain secondary voltage samples collected and the corresponding labels are shown in table 2.
TABLE 2 number of target domain secondary voltage samples and corresponding tags
2) Taking phase a using 500 fault-free operational data samplesEstablishing a 110kV principal component analysis model M1 to obtain a source domain fault-free residual error information data set +.>Source Domain failure dataset +.>Substituting M1 to obtain source domain fault residual information data set +.>Will->And->Combination is source domain feature dataset +.>. Wherein the number of principal components is selected to be 1 in consideration of the uniformity of primary voltages of the in-phase CVT.
Taking phase a using 500 fault-free operational data samplesEstablishing a 220kV principal component analysis model M2 to obtain a target domain fault-free residual error information data set +.>Taking the +.f in the few-label fault data set of the target domain>Substituting M2 to obtain target domain fault residual information data set +.>Will->And->Combination is target domain feature dataset +.>. Wherein the number of principal components is selected to be 1 in consideration of the uniformity of primary voltages of the in-phase CVT.
The truncated source domain part feature dataset is shown in table 3 and the target domain part feature dataset is shown in table 4.
TABLE 3 Source Domain partial feature dataset
TABLE 4 target Domain partial feature dataset
3) And searching the physical characteristics of the CVT capacitor breakdown fault, and deducing the relation between the CVT insulation fault and the characteristics of the CVT insulation fault. The dielectric losses of the high-voltage capacitor and the medium-voltage capacitor are respectively equivalent to parallel resistors, the intermediate transformer is represented by a T-shaped equivalent circuit, and a CVT equivalent circuit model considering internal insulation parameters can be obtained, as shown in fig. 4.
In FIG. 4Equivalent capacitive reactance for a single capacitive element, +.>Equivalent resistance of a single capacitive element, +.>To compensate the reactance of the reactor +.>Is an equivalent parameter of the intermediate transformer, +.>Equivalent impedance for the excitation branch of the intermediate transformer, < >>Is the load of the CVT. The high-voltage side has m capacitance units, and the medium-voltage side has n capacitance units. Then for a single capacitive element its equivalent resistance +.>The usable dielectric loss tangent tan delta is expressed as:
（19）
wherein the method comprises the steps ofFor angular velocity of vector rotation, +.>For the capacitance of a single capacitance unit, let，/>The CVT secondary side output voltage taking into account the internal insulation parameters at this time is:
then there is
Therein is provided with
Variable ratio ofThe method comprises the following steps:
when the CVT is considered to have a capacitor breakdown fault, only the secondary ammeter and other loads are connected to the secondary side of the CVTRemain unchanged. The transformation ratio k satisfies +.>When the capacitor cells are suddenly shorted, only the number m or n of the high-voltage and medium-voltage capacitor cells is changed. Difference of->Can be expressed as:
then when the high pressure sideCVT ratio when breakdown of the capacitor cells occurs>The process is as follows:
when the medium pressure sideCVT ratio when breakdown of the capacitor unit occurs>The process is as follows:
from equations (53) (54) (55), the CVT parameters are taken in to obtain a high voltage breakdown of 2 times the difference in ratio of one capacitor for high voltage breakdown and a medium voltage breakdown of 2 times the difference in ratio of one capacitor for medium voltage breakdown.
Let the secondary voltage normal sample before normalization be:
(38)
then there isAnd (3) standardization to obtain:
(39)
the mean and variance in the formula (39) are shown as formula (40):
，/>(40)
the normalized secondary voltage normal samples are:
(41)
the residual information of the normalized secondary voltage normal sample is as follows:
(42)
then there is an expected value:
(43)
when the j-th group CVT at the moment t has the capacitance breakdown fault and causesIs the ratio difference between the two voltage samples before normalization:
(44)
the standardized secondary voltage fault samples are:
(45)
order the
(46)
(47)
Theoretically there isAnd->Satisfying the same distribution, there is->
The residual information of the secondary voltage fault sample at the time t after standardization is as follows:
(48)/>
theoretically there is an expected value, variance:
(49)
order the
(50)
Wherein the method comprises the steps ofFrom principal component analysis models
The expectations of the fault labels of the available target domains are shown in table 5 and the variances are shown in table 6.
TABLE 5 target Domain failure tag expectations
TABLE 6 target Domain failure tag variances
4) And calculating the probability density distribution of the Gaussian mixture model of the target domain fault tag [1] by using an EM method, wherein the probability density distribution is as follows:
then there is
/>
Similarly, the probability density distribution of the Gaussian mixture model of the fault tag [3] can be calculated.
The partial gaussian mixture model parameters for the fault types available are shown in tables 7, 8, 9, 10.
TABLE 7 Gaussian mixture parameters for each failure of the target Domain mu 1
TABLE 8 Gaussian mixture parameters for each failure of the target domain mu 2
TABLE 9 Gaussian mixture parameters for each failure in the target region w 1
TABLE 10 Gaussian mixture parameters for each failure in the target Domain w 2
The standard deviation can be calculated from formulas (15) and (18).
The fault types and the sample numbers of the target domain feature data set after data amplification are obtained by simple random sampling through Gaussian mixture probability density distribution are shown in table 11.
TABLE 11 failure type and sample count of target Domain feature dataset after sample amplification
5) And migrating the source domain characteristic data set and the target domain characteristic data set after sample amplification by using a Coral migration learning method, wherein partial data after migration are shown in Table 12.
Table 12 post-migration source domain part feature dataset
The truncated target domain partial secondary voltage dataset is shown in table and the partial feature dataset is shown in table 13.
TABLE 13 post-migration target Domain partial feature dataset
Modeling the migrated source domain characteristic data set by using a random forest method, and testing by using newly collected capacitor breakdown insulation fault data in a transformer substation site. The new collected data of the transformer substation contains fault types and sample numbers as shown in table 14.
Table 14 new collected data of substation contains fault type and sample number
The model identification accuracy can reach 100%.
The embodiment of the invention provides an online identification method for insulation faults of a voltage transformer, which is used for collecting a large amount of data of all-class fault states of an insulation fault simulation platform as a source domain fault data set; collecting incomplete data of the insulating state fault type in the transformer substation site as a target domain few-label fault data set; based on knowledge deduction of the relation between the insulation fault of the voltage transformer and the characteristic data thereof, the category label completion of the target domain few-label characteristic data set is realized, and the data amplification of the full category fault state is carried out on the target domain few-label characteristic data set based on a Gaussian mixture model; and migrating the two data sets, training based on the migrated source domain characteristic data sets, and identifying the insulation faults of the transformer substation data by using a trained identification model. According to the invention, power failure maintenance is not needed, the insulation faults of all types of voltage transformers of the transformer substation can be identified with high accuracy only by using a sample data set with incomplete insulation fault types of the transformer substation, the running state of the voltage transformers is mastered in real time, and the guarantee is provided for the stable running of a power system.
In the foregoing embodiments, the descriptions of the embodiments are focused on, and for those portions of one embodiment that are not described in detail, reference may be made to the related descriptions of other embodiments.
It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
While preferred embodiments of the present invention have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. It is therefore intended that the following claims be interpreted as including the preferred embodiments and all such alterations and modifications as fall within the scope of the invention.
It will be apparent to those skilled in the art that various modifications and variations can be made to the present invention without departing from the spirit or scope of the invention. Thus, it is intended that the present invention also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.
Claims (7)
1. The online identification method for the insulation fault of the voltage transformer is characterized by comprising the following steps of:
step S1, taking voltage transformers VT of the same voltage class as a group, and taking secondary voltage data as a source domain fault data set when the insulating state of a VT insulating fault simulation platform is normal and taking secondary voltage data as a source domain fault data set when the insulating state of the whole class is fault; collecting secondary voltage data in a transformer substation site when the insulation state is normal as a fault-free data set of a target domain, and collecting secondary voltage data with incomplete insulation state fault types in the transformer substation site as a fault data set of fewer labels of the target domain;
step S2, carrying out feature extraction on a source domain non-fault data set and a source domain fault data set to obtain a source domain feature data set, and carrying out feature extraction on a target domain non-fault data set and a target domain less-label fault data set to obtain a target domain feature data set;
step S3, calculating expectations and variances of characteristic data under various insulation state faults of a target domain based on knowledge deduction of the relation between the insulation faults of the voltage transformer and the characteristic data of the voltage transformer, and realizing class label completion of a few-label characteristic data set of the target domain;
s4, establishing a Gaussian mixture model of the target domain few-label feature data set by using an Expectation Maximization (EM), obtaining Gaussian mixture distribution of the target domain full-class labels according to expectations and variances of feature data under each class of insulation state faults in the step S3, carrying out simple random sampling on the Gaussian mixture distribution, and amplifying the target domain few-label fault data set to form a target domain full-label feature data set;
and S5, migrating the source domain characteristic dataset and the target domain full-label characteristic dataset by using a migration learning method to obtain a migrated source domain characteristic dataset, training the migrated source domain characteristic dataset by using a supervision learning method, and identifying the insulation fault of the migrated target domain characteristic dataset by using a trained identification model.
2. The method for on-line identification of insulation faults of voltage transformers according to claim 1, wherein the step S1 comprises:
the three-phase secondary side voltage amplitude signals of the VT of the L groups of VT insulation fault simulation platforms are collected in real time through the data collection device, the three-phase secondary side voltage amplitude signals comprise secondary voltage data of normal insulation state and insulation state faults, a source domain fault-free data set and a source domain fault data set are respectively formed, wherein the insulation faults are multiple types of faults, and the insulation faults at least comprise voltage transformersInterlayer breakdown failure>Turn-to-turn breakdown fault and capacitive divider high and medium voltage +.>The capacitor breakdown faults of the capacitor units are positive integers, and L and n are all positive integers;
the three-phase secondary side voltage amplitude signals of the L groups VT of the transformer substation are collected in real time through a data collecting device, and the three-phase secondary side voltage amplitude signals comprise secondary voltage data with normal insulation state and incomplete insulation state fault types, and a target domain non-fault data set and a target domain few-label fault data set are respectively formed, wherein the insulation faults are few types of faults and comprise voltage transformersInterlayer breakdown failure>Turn-to-turn breakdown fault and capacitive divider high and medium voltage +.>And the capacitor units break down and fail, m is a positive integer, and m is smaller than n.
3. The method for on-line identification of insulation faults of voltage transformers according to claim 1, wherein the step 2 comprises:
source field fault-free data set three-phase dataA principal component analysis model M1 is respectively established by using a principal component analysis method to obtain a source domain fault-free residual error information data set +.>The method comprises the steps of carrying out a first treatment on the surface of the Source takingDomain fault data set three-phase dataSubstituting into principal component analysis model M1 to obtain source domain fault residual information data set +.>Combining the source domain fault-free residual information data set and the source domain fault residual information data set into a source domain characteristic data set；
Taking three-phase data in fault-free data set of target domainA principal component analysis method is used for respectively establishing principal component analysis models M2 to obtain a fault-free residual error information data set of a target domain +.>The method comprises the steps of carrying out a first treatment on the surface of the Three-phase data in few-label fault data set of target domain are taken>Substituting the target domain fault residual error information data set into a principal component analysis model M2 to obtain a target domain fault residual error information data setCombining the target domain fault-free residual information data set and the target domain few-label fault residual information data set into a target domain characteristic data set +.>。
4. The method for on-line identification of insulation faults of voltage transformers according to claim 3, wherein the step S3 of calculating expectations and variances of feature data under each type of insulation state faults of a target domain based on knowledge derivation of connection between the insulation faults of the voltage transformers and feature data thereof, and realizing type label completion of a few-label feature data set of the target domain comprises the following steps:
target domain feature datasetCo-inclusion->Category faults, representing a certain group of VT occurrences +.>Breakdown of species category->The ith layer or turn, the ith group VT occurs +.>Insulation fault and breakdown->The target domain feature data set corresponding to the individual layers or turns is denoted +.>；
Calculating a target domain feature dataset using statistical relationshipsIs +.>And variance->Combining the physical characteristics of the insulation fault of the voltage transformer to obtain a target domain characteristic data set +.>Is present with the expected value and variance of (a)Functional relation->And->Wherein->Represents i, m and->The functional relation between ∈>Indicating (I)>Representing i, m andthe functional relation between them is used for g' b The expected values and variances of other types of insulation faults not included in the target domain are calculated by using the functional relation.
5. The method for on-line identification of insulation faults of voltage transformers according to claim 4, wherein the step S4 comprises:
source record domain feature datasetCo-inclusion->Failure of species category, representing L groups VT occurring +.>Species category breakdown->The ith layer or turn, the ith group VT occurs +.>The source domain feature data set corresponding to the insulation fault and breakdown c layers or turns is recorded as +.>；
Calculating insulation fault categories contained in target domain feature data set using Expectation Maximization (EM) methodIs a Gaussian mixture model probability density function +.>；
Based onIs a Gaussian mixture model probability density function +.>Calculating the sum +.>Feature data sets of the same fault group as the fault category but different breakdown numbers +.>Is a Gaussian mixture model probability density function +.>And calculating a sum of the target domain feature data set and the data set which does not contain the insulation fault categoryThe fault category is the same as the breakdown number but the feature number of the fault group is differentData set->Is a Gaussian mixture model probability density function +.>；
Recording deviceThere are r samples, with +.>Simple random sampling of r samples is done +.>Data amplification of (2) usingSimple random sampling of r samples is done +.>Is amplified to form a target domain full-label characteristic data set +.>Comprises->One category of insulation faults.
6. The method for on-line identification of insulation faults of a voltage transformer according to claim 5, wherein,is a Gaussian mixture model probability density function +.>The method comprises the following steps:
；
the method comprises the following steps:
；
；
；
k is the number of normal distribution models constituting the Gaussian mixture model,is the kth normal distribution model in the Gaussian mixture model, +.>Is the expectation of the kth normal distribution model, +.>Variance of kth normal distribution model, +.>The weight of the kth normal distribution model;
is a Gaussian mixture model probability density function +.>The method comprises the following steps:
；
the method comprises the following steps:
；
；
；
wherein v=1,..n, n is a positive integer and v+.m;
is a Gaussian mixture model probability density function +.>The method comprises the following steps:
；
then there are:
；
；
；
wherein the method comprises the steps ofAll are KXL matrices;
wherein,
；
；
；
；
；
；
wherein the method comprises the steps of。
7. The method for identifying the insulation fault of the voltage transformer on line according to claim 1, wherein the step S5 comprises:
source domain feature dataset using a migration learning methodFull tag feature data set with target field->Performing migration learning to obtain a migrated source domain feature data set +.>；
Training a migrated source domain feature dataset using a supervised learning approachObtaining a migration learning model, extracting residual information from newly collected fault data in a transformer substation site through principal component analysis, substituting the residual information into the constructed migration learning model, and substituting the migrated data into a trained supervised learning method to finish identification of an insulation fault.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311396755.3A CN117131786B (en) | 2023-10-26 | 2023-10-26 | Voltage transformer insulation fault online identification method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311396755.3A CN117131786B (en) | 2023-10-26 | 2023-10-26 | Voltage transformer insulation fault online identification method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117131786A true CN117131786A (en) | 2023-11-28 |
| CN117131786B CN117131786B (en) | 2024-01-26 |
Family
ID=88858627
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311396755.3A Active CN117131786B (en) | 2023-10-26 | 2023-10-26 | Voltage transformer insulation fault online identification method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117131786B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117849558A (en) * | 2024-03-06 | 2024-04-09 | 武汉格蓝若智能技术股份有限公司 | Method and device for on-line monitoring breakdown fault of main capacitor of capacitive voltage transformer |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160335548A1 (en) * | 2015-05-12 | 2016-11-17 | Rolls-Royce Plc | Methods and apparatus for predicting fault occurrence in mechanical systems and electrical systems |
| US20180114113A1 (en) * | 2016-10-20 | 2018-04-26 | Uber Technologies, Inc. | Intelligent regularization of neural network architectures |
| US20210133559A1 (en) * | 2019-11-04 | 2021-05-06 | Cisco Technology, Inc. | Iot-based network architecture for detecting faults using vibration measurement data |
| CN113435375A (en) * | 2021-07-05 | 2021-09-24 | 西安交通大学 | Cross-equipment mobility measurement method for rolling bearing diagnosis knowledge |
| US20210307621A1 (en) * | 2017-05-29 | 2021-10-07 | Saltor Pty Ltd | Method And System For Abnormality Detection |
| KR102398046B1 (en) * | 2021-03-10 | 2022-05-12 | 서울대학교산학협력단 | Fault diagnosis device using unsupervised domain adaptation technique and fault diagnosis method using the same |
| CN114529817A (en) * | 2022-02-21 | 2022-05-24 | 东南大学 | Unmanned aerial vehicle photovoltaic fault diagnosis and positioning method based on attention neural network |
| CN114757305A (en) * | 2022-06-13 | 2022-07-15 | 华中科技大学 | Voltage transformer insulation fault identification method and system based on ensemble learning |
| US20230152792A1 (en) * | 2021-11-16 | 2023-05-18 | Clemson University | System for predictive operational analysis of a machinery component |
-
2023
- 2023-10-26 CN CN202311396755.3A patent/CN117131786B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160335548A1 (en) * | 2015-05-12 | 2016-11-17 | Rolls-Royce Plc | Methods and apparatus for predicting fault occurrence in mechanical systems and electrical systems |
| US20180114113A1 (en) * | 2016-10-20 | 2018-04-26 | Uber Technologies, Inc. | Intelligent regularization of neural network architectures |
| US20210307621A1 (en) * | 2017-05-29 | 2021-10-07 | Saltor Pty Ltd | Method And System For Abnormality Detection |
| US20210133559A1 (en) * | 2019-11-04 | 2021-05-06 | Cisco Technology, Inc. | Iot-based network architecture for detecting faults using vibration measurement data |
| KR102398046B1 (en) * | 2021-03-10 | 2022-05-12 | 서울대학교산학협력단 | Fault diagnosis device using unsupervised domain adaptation technique and fault diagnosis method using the same |
| CN113435375A (en) * | 2021-07-05 | 2021-09-24 | 西安交通大学 | Cross-equipment mobility measurement method for rolling bearing diagnosis knowledge |
| US20230152792A1 (en) * | 2021-11-16 | 2023-05-18 | Clemson University | System for predictive operational analysis of a machinery component |
| CN114529817A (en) * | 2022-02-21 | 2022-05-24 | 东南大学 | Unmanned aerial vehicle photovoltaic fault diagnosis and positioning method based on attention neural network |
| CN114757305A (en) * | 2022-06-13 | 2022-07-15 | 华中科技大学 | Voltage transformer insulation fault identification method and system based on ensemble learning |
Non-Patent Citations (3)
| Title |
|---|
| TING LEI 等 ON-LINE FAULT DETECTION TECHNIQUE FOR VOLTAGE TRANSFORMERS ON-LINE FAULT DETECTION TECHNIQUE FOR VOLTAGE TRANSFORMER: "On-line fault detection technique for voltage transformers", SCIENCEDIRECT, pages 193 - 200 * |
| YUNWU XU等: "On-line Evaluation Method of CVT Internal Insulation State Based on Support Vector Machine", 2021 IEEE 15TH ICEMI, pages 137 - 144 * |
| YUXUAN ZHANG 等: "A physics-information-enabled self-updating method to monitor steady-state error of capacitor voltage transformers", RESEARCHGATE, pages 1 - 12 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117849558A (en) * | 2024-03-06 | 2024-04-09 | 武汉格蓝若智能技术股份有限公司 | Method and device for on-line monitoring breakdown fault of main capacitor of capacitive voltage transformer |
| CN117849558B (en) * | 2024-03-06 | 2024-05-24 | 武汉格蓝若智能技术股份有限公司 | Method and device for on-line monitoring breakdown fault of main capacitor of capacitive voltage transformer |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117131786B (en) | 2024-01-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Hong et al. | Vehicle energy system active defense: a health assessment of lithium‐ion batteries | |
| CN107330286A (en) | A kind of large oil immersed power transformer reliability assessment dynamic correcting method | |
| CN109344517A (en) | A kind of high-voltage isulation method for diagnosing faults of new-energy automobile | |
| CN113702895B (en) | Online quantitative evaluation method for error state of voltage transformer | |
| CN117131786B (en) | Voltage transformer insulation fault online identification method | |
| CN103983757A (en) | Transformer insulation thermal ageing state reliability evaluation method based on Weibull distribution | |
| CN113447783B (en) | Voltage transformer insulation fault identification model construction method and device | |
| Nosratian Ahour et al. | Determination and localisation of turn‐to‐turn fault in transformer winding using frequency response analysis | |
| CN102034019A (en) | Method for comprehensively estimating insulation state of sulfur hexafluoride electrical equipment | |
| CN108037460A (en) | One kind batch production capacity of lithium ion battery real time evaluating method | |
| CN103996099B (en) | Method for conducting intelligent evaluation on student operation in training system | |
| CN114814501B (en) | On-line diagnosis method for capacitor breakdown fault of capacitor voltage transformer | |
| Luo et al. | Locating faults in the transmission network using sparse field measurements, simulation data and genetic algorithm | |
| CN111797566A (en) | Method and system for determining key characteristic quantity for representing health state of transformer | |
| CN114779029B (en) | CVT internal insulation online evaluation method fusing group redundancy association and structural parameters | |
| CN108875796A (en) | Diagnosing failure of photovoltaic array method based on linear discriminant analysis and support vector machines | |
| Davis et al. | Estimation of transmission line parameters from historical data | |
| CN108681835B (en) | Method for evaluating degradation state of transformer insulating oil | |
| CN115494349B (en) | Active power distribution network single-phase earth fault section positioning method | |
| CN109031020B (en) | Transformer inrush current identification method based on logistic regression | |
| CN113052249B (en) | Transformer winding fault type identification method based on support vector machine and current deviation coefficient | |
| CN114384319A (en) | Grid-connected inverter island detection method, system, terminal and medium | |
| CN117849558B (en) | Method and device for on-line monitoring breakdown fault of main capacitor of capacitive voltage transformer | |
| Thapa et al. | A machine learning-based approach to detection of fault locations in transmission networks | |
| Subroto et al. | Artificial intelligence for DGA interpretation methods using weighting factor |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |