CN112067052A - Oil-immersed transformer fault diagnosis method based on feature selection - Google Patents

Oil-immersed transformer fault diagnosis method based on feature selection Download PDF

Info

Publication number
CN112067052A
CN112067052A CN202010913195.4A CN202010913195A CN112067052A CN 112067052 A CN112067052 A CN 112067052A CN 202010913195 A CN202010913195 A CN 202010913195A CN 112067052 A CN112067052 A CN 112067052A
Authority
CN
China
Prior art keywords
vector
concentration
feature
elements
oil
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.)
Withdrawn
Application number
CN202010913195.4A
Other languages
Chinese (zh)
Inventor
陈泰麒
葛英辉
其他发明人请求不公开姓名
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Ningbo University
Original Assignee
Ningbo University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Ningbo University filed Critical Ningbo University
Priority to CN202010913195.4A priority Critical patent/CN112067052A/en
Publication of CN112067052A publication Critical patent/CN112067052A/en
Withdrawn legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01D—MEASURING NOT SPECIALLY ADAPTED FOR A SPECIFIC VARIABLE; ARRANGEMENTS FOR MEASURING TWO OR MORE VARIABLES NOT COVERED IN A SINGLE OTHER SUBCLASS; TARIFF METERING APPARATUS; MEASURING OR TESTING NOT OTHERWISE PROVIDED FOR
    • G01D21/00—Measuring or testing not otherwise provided for
    • G01D21/02—Measuring two or more variables by means not covered by a single other subclass

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Housings And Mounting Of Transformers (AREA)

Abstract

The invention discloses an oil-immersed transformer fault diagnosis method based on feature selection. Specifically, the method firstly performs double construction of statistical characteristics and ratio characteristics on the dissolved gas concentration data. Secondly, the method selects the most suitable characteristic variable for fault classification diagnosis by using a neighbor component analysis algorithm. And finally, establishing a probabilistic neural network model by using the selected characteristic variables to implement transformer fault diagnosis. The method has the advantages that: firstly, the method firstly carries out feature optimization through feature expansion, thereby greatly ensuring the precision of the classification model; secondly, the method of the invention is simple to operate and very easy to implement.

Description

Oil-immersed transformer fault diagnosis method based on feature selection
Technical Field
The invention relates to a transformer fault diagnosis method, in particular to an oil-immersed transformer fault diagnosis method based on feature selection.
Background
With the increasing demand for electric power, transformers have become indispensable electrical devices in electric power transmission systems. As a key link of power supply and distribution, the operation performance of the transformer directly affects the operation of the whole power system. Any transformer fault type results in wasted power and even more serious economic losses, so fault diagnosis of transformer equipment is of great research significance to avoid potential power or other economic losses. Since transformers used in power supply and distribution systems are generally oil-immersed transformers, a common idea for performing fault diagnosis of transformers is to analyze gases (hydrogen, methane, ethane, ethylene, and acetylene) dissolved in transformer oil. The method has the defects of code defect and critical value criterion defect. In recent years, the emerging transformer fault diagnosis methods use the dissolved gas concentration data to classify the faults, so as to realize the diagnosis of the transformer faults.
Discriminant analysis and neural network are the most common mode classification techniques, and can be applied to solving the problem of transformer fault classification diagnosis. However, the classification accuracy of the neural network-based transformer fault diagnosis model is directly affected by network input parameters or variables. In other words, the input variables or parameters of the neural network directly affect the accuracy of the diagnosis. In addition, discriminant analysis is a linear classification diagnosis strategy and cannot effectively adapt to the variability and nonlinear characteristics of the concentration data of the dissolved gas of the transformer. Most importantly, these methods for classification diagnosis require as much sample data as possible for training, and fail to select the most suitable features for diagnosing faults to establish the classification diagnosis fault types.
Data-driven transformer fault diagnosis is directly dependent on dissolved gas concentration data in transformer oil, and the dissolved gas concentration data can reflect different fault types. However, from the viewpoint of the requirement of reliable and accurate transformer fault diagnosis task, it is difficult to implement the task by directly relying on the concentration data of the dissolved gas alone, and it is necessary to further mine the change characteristics of the dissolved gas concentration data on the basis of the dissolved gas concentration data and perform the fault diagnosis of the transformer by using more characteristic data. However, in the existing scientific research literature and patent materials, there are few research results related to the above, and generally, the research is to directly use the dissolved gas concentration data to perform fault classification diagnosis.
In addition, as the category characteristics of the dissolved gas concentration data in the oil-immersed transformer are limited, the three-ratio method does not directly use the dissolved gas concentration data, and uses the ratio between the dissolved gas concentrations for further classification diagnosis. Therefore, further feature expansion and selection of the dissolved gas concentration data is positively effective for fault diagnosis. It can be said that the fault diagnosis of the transformer using the dissolved gas concentration data requires a fault diagnosis method technique capable of coping with the problem of a small sample and selecting from optimal characteristics.
Disclosure of Invention
The invention aims to solve the main technical problems that: and (3) implementing feature expansion and classification feature selection by using the concentration data of the dissolved gas in the oil-immersed transformer oil, and establishing a probabilistic neural network model by using the selected features, thereby implementing transformer fault diagnosis. Specifically, the method firstly performs double construction of statistical characteristics and ratio characteristics on the dissolved gas concentration data. Secondly, the method of the invention utilizes a neighbor Analysis (NCA) algorithm to select the most suitable characteristic variable for fault classification diagnosis. And finally, establishing a probabilistic neural network model by using the selected characteristic variables to implement transformer fault diagnosis.
The technical scheme adopted by the method for solving the problems is as follows: an oil-immersed transformer fault diagnosis method based on feature selection comprises the following steps:
step (1): carrying out characteristic expansion on the dissolved gas concentration data of the oil-immersed transformer in 7 different working states so as to obtain N in a healthy working state0A feature vector
Figure BSA0000218487810000021
N in partial discharge fault state1A feature vector
Figure BSA0000218487810000022
N in spark-over fault condition2A feature vector
Figure BSA0000218487810000023
N in arc discharge fault condition3A feature vector
Figure BSA0000218487810000024
N in medium temperature overheat fault state4A feature vector
Figure BSA0000218487810000025
N at low temperature over-temperature fault condition5A feature vector
Figure BSA0000218487810000026
And N in a high temperature overheat fault condition6A feature vector
Figure BSA0000218487810000027
The specific implementation process comprises the following steps (1.1) to (1.5).
Step (1.1): the analysis data of the dissolved gas in the transformer oil specifically comprises the following steps: concentration of hydrogen
Figure BSA0000218487810000028
Concentration of methane
Figure BSA0000218487810000029
Ethane concentration
Figure BSA00002184878100000210
Ethylene concentration
Figure BSA00002184878100000211
And acetylene concentration
Figure BSA00002184878100000212
The 5 concentration data can be used for constructing a concentration vector of the dissolved gas
Figure BSA00002184878100000213
Wherein k represents a sample number, and the upper label c is belonged to {0, 1, 2, 3, 4, 5, 6} to respectively indicate a healthy working state, a partial discharge fault state, a spark discharge fault state, an arc discharge fault state, a medium-temperature overheat fault state, a low-temperature overheat fault state, and a high-temperature overheat fault state.
Step (1.2): respectively calculating the mean value of each dissolved gas concentration vector according to the formula shown below
Figure BSA00002184878100000214
Standard deviation of
Figure BSA00002184878100000215
Kurtosis
Figure BSA00002184878100000216
Deflection degree
Figure BSA00002184878100000217
Root mean square
Figure BSA00002184878100000218
Crest factor
Figure BSA00002184878100000219
Form factor
Figure BSA00002184878100000220
Pulse factor
Figure BSA00002184878100000221
Edge factor
Figure BSA00002184878100000222
Maximum logarithm of
Figure BSA00002184878100000223
Figure BSA00002184878100000224
Figure BSA00002184878100000225
Figure BSA00002184878100000226
Figure BSA00002184878100000227
Figure BSA00002184878100000228
Figure BSA00002184878100000229
Figure BSA00002184878100000230
Figure BSA00002184878100000231
Figure BSA00002184878100000232
Figure BSA00002184878100000233
Wherein b is equal to {1, 2, 3, 4, 5}, and the sample number k is equal to {1, 2, …, N ∈ [ ]c}，
Figure BSA00002184878100000234
Representation calculation
Figure BSA00002184878100000235
The maximum value of the medium element.
Step (1.3): the ratio coefficient of each dissolved gas concentration vector is calculated according to the formula shown below
Figure BSA0000218487810000031
Figure BSA0000218487810000032
In the above formula, d ∈ {1, 2, …, 15 }.
Step (1.4): according to
Figure BSA0000218487810000033
Constructing a multi-feature fusion vector
Figure BSA0000218487810000034
Wherein
Figure BSA0000218487810000035
R25×1A real number vector of 25 × 1 dimensions is represented, and the upper symbol T represents a transposed symbol of a matrix or vector.
Step (1.5): and (4) repeating the steps (1.2) to (1.4) to respectively obtain the multi-feature fusion vectors of the oil-immersed transformer operating in 7 different working states.
Step (2): building a feature matrix
Figure BSA0000218487810000036
Building class label vector y ═ y0，y1，…，y6]T∈RN×1(ii) a Wherein N is N0+N1+…+N6，RN×1Representing vectors, of real numbers in dimension Nx 1
Figure BSA0000218487810000037
All elements in are equal to 0, vector
Figure BSA0000218487810000038
All elements in are equal to 1, vector
Figure BSA0000218487810000039
All elements in (1) are equal to 2, vector
Figure BSA00002184878100000310
All of the elements inAre all equal to 3, vector
Figure BSA00002184878100000311
All elements in (1) are equal to 4, vector
Figure BSA00002184878100000312
All elements in (1) are equal to 5, vector
Figure BSA00002184878100000313
All elements in (a) are equal to 6.
And (3): for the feature matrix X ∈ RN×25After the standardization processing is carried out, a characteristic weight vector w epsilon R is obtained by utilizing a neighbor component analysis algorithm1×25。
And (4): finding out the largest f elements from the characteristic weight vector w, recording the positions of the f elements as a position set phi, correspondingly selecting the column vectors at the same positions from the characteristic matrix X, and establishing a new characteristic matrix
Figure BSA00002184878100000314
Wherein R isN×fA matrix of real numbers representing dimensions N × f.
And (5): to be provided with
Figure BSA00002184878100000315
And taking each row vector as input data, taking elements of a corresponding row in the class label vector y as output data, and establishing a probabilistic neural network model.
And (6): the new measurement obtains the concentration data of the dissolved gas in the oil-immersed transformer, specifically including 5 concentration data of hydrogen concentration, methane concentration, ethane concentration, ethylene concentration and acetylene concentration
And (7): performing feature expansion on the 5 concentration data in the step (6) to obtain a feature vector xnew∈R1 ×25The specific implementation process is the same as the steps (1.1) to (1.4), and then the feature vector x is selected from the feature vector x according to the position set recorded in the step (4)newSelecting elements at corresponding positions in the image to construct a new feature vector
Figure BSA00002184878100000316
Wherein R is1×fRepresenting a real number vector of dimension 1 xf.
And (8): with new feature vectors
Figure BSA0000218487810000041
Using the probabilistic neural network model established in the step (5) as input data to calculate and obtain an output estimation value ynewAnd based on the output estimate ynewAnd determining the current working state of the oil-immersed transformer.
By carrying out the steps described above, the advantages of the method of the invention are presented below.
Firstly, the method of the invention involves substantially no complex transformations or mathematical calculations, and is simple to operate and very easy to implement. Secondly, the method firstly carries out feature optimization through feature expansion, and greatly ensures the precision of the classification model. Finally, in the specific implementation case, the reliability and the superiority of the method are fully illustrated by the comparison of case implementation results.
Drawings
FIG. 1 is a schematic flow chart of the method of the present invention.
Detailed Description
The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.
The invention discloses an oil-immersed transformer fault diagnosis method based on feature selection, and the specific implementation mode of the method is described by combining an implementation flow schematic diagram shown in figure 1.
In this embodiment, the oil immersed transformer has N in a healthy working state050 dissolved gas concentration data, N in partial discharge fault state1In the spark failure state, N is present in 21 groups of data2For 16 sets of data, there is N in the arc fault condition318 groups of data, N in medium-temperature overheat fault state423 groups of data, N in low-temperature overheat fault state523 groups of data, highN in the warm overheat fault state624 sets of data. And establishing a fault diagnosis model of the transformer by using the data and carrying out online fault diagnosis, wherein the method specifically comprises the following steps.
Step (1): carrying out characteristic expansion on the dissolved gas concentration data of the oil-immersed transformer operating under 7 different working states according to the steps (1.1) to (1.5), thereby obtaining N under the healthy working state0A feature vector
Figure BSA0000218487810000042
N in partial discharge fault state1A feature vector
Figure BSA0000218487810000043
N in spark-over fault condition2A feature vector
Figure BSA0000218487810000044
N in arc discharge fault condition3A feature vector
Figure BSA0000218487810000045
N in medium temperature overheat fault state4A feature vector
Figure BSA0000218487810000046
N at low temperature over-temperature fault condition5A feature vector
Figure BSA0000218487810000047
And N in a high temperature overheat fault condition6A feature vector
Figure BSA0000218487810000048
Step (2): building a feature matrix
Figure BSA0000218487810000049
Building class label vector y ═ y0，y1，…，y6]T∈RN×1。
Step (3): for the feature matrix X ∈ RN×25Obtaining a matrix after carrying out standardization processing
Figure BSA00002184878100000410
And obtaining a characteristic weight vector w epsilon R by utilizing a neighbor component analysis algorithm1×25The specific implementation process is as follows:
first, the initialization gradient step α is 1, and the initialization objective function value F is set to 10(w)＝-106And initializing the weight coefficient vector w ═ 1, 1, …, 1]∈R1×25。
② calculating the objective function value F (w) under the condition of the current weight coefficient vector w:
Figure BSA0000218487810000051
wherein the probability p of correct classificationiCan be calculated according to the formula shown below:
Figure BSA0000218487810000052
in the above formula, pijThe calculation formula of (a) is as follows:
Figure BSA0000218487810000053
wherein the content of the first and second substances,
Figure BSA0000218487810000054
and
Figure BSA0000218487810000055
respectively represent matrices
Figure BSA0000218487810000056
The ith and jth row vectors of (1), if and only if
Figure BSA0000218487810000057
And
Figure BSA0000218487810000058
when belonging to the same class (i.e. the ith element is equal to the jth element in the class label vector y), yijOther cases y 1ij0; distance Dw(xi，xj) The calculation of (c) is as follows:
Figure BSA0000218487810000059
where i ═ 1, 2, …, N, j ═ 1, 2, …, N, the notation | | | | | denotes the length of the calculated vector, and diag (w) denotes the transformation of the vector w into a diagonal matrix.
(iii) determining whether the convergence condition | F (w) -F is satisfied0(w)|＜10-6(ii) a If yes, outputting a weight coefficient vector w; if not, continuing to implement the fourth step.
Fourthly, F is arranged0Calculating a gradient value delta f after (w) f (w), and updating the weight coefficient vector according to a formula w + alpha delta f; the gradient value Δ f is calculated as follows:
Figure BSA00002184878100000510
fifthly, according to the above formula
Figure BSA00002184878100000511
Calculating an objective function value F (w), and judging whether a condition F (w) > F is met0(w)? If yes, updating the gradient step length alpha according to the formula alpha which is 1.01 alpha; if not, updating the gradient step length alpha according to the formula alpha being 0.4 alpha.
And sixthly, returning to the step III to continue the next iterative optimization until the convergence condition in the step III is met.
And (4): finding out the largest f elements from the characteristic weight vector w, recording the positions of the f elements in the w as a position set phi, correspondingly selecting the column vectors at the same positions from the characteristic matrix X, and establishing a new characteristic matrix
Figure BSA00002184878100000512
And (5): to be provided with
Figure BSA00002184878100000513
And taking each row vector as input data, taking elements of a corresponding row in the class label vector y as output data, and establishing a probabilistic neural network model.
And (6): the new measurement obtains the concentration data of the dissolved gas in the oil-immersed transformer, and specifically comprises 5 concentration data of hydrogen concentration, methane concentration, ethane concentration, ethylene concentration and acetylene concentration.
And (7): performing feature expansion on the 5 concentration data in the step (6) to obtain a feature vector xnew∈R1 ×25The specific implementation process is the same as the steps (1.1) to (1.4), and then the position set phi recorded in the step (4) is used for determining the position from the feature vector xnewSelecting elements at corresponding positions in the image to construct a new feature vector
Figure BSA0000218487810000061
And (8): with new feature vectors
Figure BSA0000218487810000062
Using the probabilistic neural network model established in the step (5) as input data to calculate and obtain an output estimation value ynewAnd based on the output estimate ynewAnd determining the current working state of the oil-immersed transformer.
Specifically, the molar ratio of ynewIf the voltage is equal to 0, the transformer runs in a healthy working state; if ynew1, the transformer is operated in a partial discharge fault state; if ynewWhen it is 2, it indicates a spark discharge failure state, and if ynewWhen the value is 3, the arc discharge fault state is indicated; if ynewWhen y is 4, it indicates a medium-temperature overheat fault statenew5, indicating a low-temperature overheat fault state; if ynew6 denotes high-temperature overheatingA fault condition.

Claims (1)

1. An oil-immersed transformer fault diagnosis method based on feature selection is characterized by comprising the following steps:
step (1): carrying out characteristic expansion on the dissolved gas concentration data of the oil-immersed transformer in 7 different working states so as to obtain N in a healthy working state0A feature vector
Figure FSA0000218487800000011
N in partial discharge fault state1A feature vector
Figure FSA0000218487800000012
N in spark-over fault condition2A feature vector
Figure FSA0000218487800000013
N in arc discharge fault condition3A feature vector
Figure FSA0000218487800000014
N in medium temperature overheat fault state4A feature vector
Figure FSA0000218487800000015
N at low temperature over-temperature fault condition5A feature vector
Figure FSA0000218487800000016
And N in a high temperature overheat fault condition6A feature vector
Figure FSA0000218487800000017
The specific implementation process comprises the following steps (1.1) to (1.5);
step (1.1): the dissolved gas concentration data in the transformer oil specifically comprises: concentration of hydrogen
Figure FSA0000218487800000018
Concentration of methane
Figure FSA0000218487800000019
Ethane concentration
Figure FSA00002184878000000110
Ethylene concentration
Figure FSA00002184878000000111
And acetylene concentration
Figure FSA00002184878000000112
The 5 concentration data can be used for constructing a concentration vector of the dissolved gas
Figure FSA00002184878000000113
Wherein k represents a sample number, and the upper label c is belonged to {0, 1, 2, 3, 4, 5, 6} to respectively indicate a healthy working state, a partial discharge fault state, a spark discharge fault state, an arc discharge fault state, a medium-temperature overheat fault state, a low-temperature overheat fault state, and a high-temperature overheat fault state;
step (1.2): respectively calculating the mean value of each dissolved gas concentration vector according to the formula shown below
Figure FSA00002184878000000114
Standard deviation of
Figure FSA00002184878000000115
Kurtosis
Figure FSA00002184878000000116
Deflection degree
Figure FSA00002184878000000117
Root mean square
Figure FSA00002184878000000118
Peak valueFactor(s)
Figure FSA00002184878000000119
Form factor
Figure FSA00002184878000000120
Pulse factor
Figure FSA00002184878000000121
Edge factor
Figure FSA00002184878000000122
Maximum logarithm of
Figure FSA00002184878000000123
Figure FSA00002184878000000124
Figure FSA00002184878000000125
Figure FSA00002184878000000126
Figure FSA00002184878000000127
Figure FSA00002184878000000128
Figure FSA00002184878000000129
Figure FSA00002184878000000130
Figure FSA00002184878000000131
Figure FSA00002184878000000132
Figure FSA00002184878000000133
Wherein b is ∈ {1, 2, 3, 4, 5},
Figure FSA00002184878000000134
representation calculation
Figure FSA00002184878000000135
Maximum value of medium element;
step (1.3): the ratio coefficient of each dissolved gas concentration vector is calculated according to the formula shown below
Figure FSA00002184878000000136
Figure FSA0000218487800000021
In the above formula, d is belonged to {1, 2, …, 15 };
step (1.4): according to
Figure FSA0000218487800000022
Constructing feature vectors
Figure FSA0000218487800000023
Wherein, the upper label T represents the transposition symbol of the matrix or the vector;
step (1.5): repeating the steps (1.2) to (1.4) to respectively obtain characteristic vectors of the oil-immersed transformer operating in 7 different working states;
step (2): building a feature matrix
Figure FSA0000218487800000024
Building class label vector y ═ y0，y1，…，y6]T∈RN×1(ii) a Wherein N is N0+N1+…+N6，RN×1Representing vectors, of real numbers in dimension Nx 1
Figure FSA0000218487800000025
All elements in are equal to 0, vector
Figure FSA0000218487800000026
All elements in are equal to 1, vector
Figure FSA0000218487800000027
All elements in (1) are equal to 2, vector
Figure FSA0000218487800000028
All elements in (1) are equal to 3, vector
Figure FSA0000218487800000029
All elements in (1) are equal to 4, vector
Figure FSA00002184878000000210
All elements in (1) are equal to 5, vector
Figure FSA00002184878000000211
All elements in (1) are equal to 6;
and (3): for the feature matrix X ∈ RN×25After the standardization processing is carried out, a characteristic weight vector w epsilon R is obtained by utilizing a neighbor component analysis algorithm1×25；
And (4): finding out the largest f elements from the characteristic weight vector w, recording the positions of the f elements in the w as a position set phi, correspondingly selecting the column vectors at the same positions from the characteristic matrix X, and establishing a new characteristic matrix
Figure FSA00002184878000000212
Wherein R isN×fA real number matrix representing dimensions N × f;
and (5): to be provided with
Figure FSA00002184878000000213
Taking each row vector as input data, taking elements of corresponding rows in the class label vector y as output data, and establishing a probabilistic neural network model;
and (6): newly measuring to obtain concentration data of dissolved gas in the oil-immersed transformer, wherein the concentration data specifically comprises 5 concentration data of hydrogen concentration, methane concentration, ethane concentration, ethylene concentration and acetylene concentration;
and (7): performing feature expansion on the 5 concentration data in the step (6) to obtain a feature vector xnew∈R1×25The specific implementation process is the same as the steps (1.1) to (1.4), and then the position set phi recorded in the step (4) is used for determining the position from the feature vector xnewSelecting elements at corresponding positions in the image to construct a new feature vector
Figure FSA00002184878000000214
And (8): with new feature vectors
Figure FSA00002184878000000215
Using the probabilistic neural network model established in the step (5) as input data to calculate and obtain an output estimation value ynewAnd based on the output estimate ynewAnd determining the current working state of the oil-immersed transformer.
CN202010913195.4A 2020-08-24 2020-08-24 Oil-immersed transformer fault diagnosis method based on feature selection Withdrawn CN112067052A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010913195.4A CN112067052A (en) 2020-08-24 2020-08-24 Oil-immersed transformer fault diagnosis method based on feature selection

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010913195.4A CN112067052A (en) 2020-08-24 2020-08-24 Oil-immersed transformer fault diagnosis method based on feature selection

Publications (1)

Publication Number Publication Date
CN112067052A true CN112067052A (en) 2020-12-11

Family

ID=73666463

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010913195.4A Withdrawn CN112067052A (en) 2020-08-24 2020-08-24 Oil-immersed transformer fault diagnosis method based on feature selection

Country Status (1)

Country Link
CN (1) CN112067052A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114545294A (en) * 2022-01-14 2022-05-27 国电南瑞科技股份有限公司 Transformer fault diagnosis method and system, storage medium and computing device

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105930901A (en) * 2016-07-18 2016-09-07 河海大学 RBPNN-based transformer fault diagnosis method
CN107063349A (en) * 2017-04-17 2017-08-18 云南电网有限责任公司电力科学研究院 A kind of method and device of Fault Diagnosis Method of Power Transformer
CN107169514A (en) * 2017-05-05 2017-09-15 清华大学 The method for building up of diagnosing fault of power transformer model
CN109376778A (en) * 2018-10-09 2019-02-22 宁波大学 A kind of failure modes diagnostic method based on characteristic variable weighting
CN109407649A (en) * 2018-10-09 2019-03-01 宁波大学 A kind of fault type matching process based on fault signature variables choice
CN110363277A (en) * 2019-07-15 2019-10-22 南京工业大学 Method for diagnosing fault of power transformer and system based on improved glowworm swarm algorithm optimization probabilistic neural network

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105930901A (en) * 2016-07-18 2016-09-07 河海大学 RBPNN-based transformer fault diagnosis method
CN107063349A (en) * 2017-04-17 2017-08-18 云南电网有限责任公司电力科学研究院 A kind of method and device of Fault Diagnosis Method of Power Transformer
CN107169514A (en) * 2017-05-05 2017-09-15 清华大学 The method for building up of diagnosing fault of power transformer model
CN109376778A (en) * 2018-10-09 2019-02-22 宁波大学 A kind of failure modes diagnostic method based on characteristic variable weighting
CN109407649A (en) * 2018-10-09 2019-03-01 宁波大学 A kind of fault type matching process based on fault signature variables choice
CN110363277A (en) * 2019-07-15 2019-10-22 南京工业大学 Method for diagnosing fault of power transformer and system based on improved glowworm swarm algorithm optimization probabilistic neural network

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
王莹等: "变压器故障诊断的概率神经网络研究", 《信息系统工程》 *
陈如意等: "基于最大信息系数的变压器过热故障特征选择", 《电力工程技术》 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114545294A (en) * 2022-01-14 2022-05-27 国电南瑞科技股份有限公司 Transformer fault diagnosis method and system, storage medium and computing device
CN114545294B (en) * 2022-01-14 2023-06-16 国电南瑞科技股份有限公司 Transformer fault diagnosis method, system, storage medium and computing device

Similar Documents

Publication Publication Date Title
Liu et al. Intelligent fault diagnosis of photovoltaic array based on variable predictive models and I–V curves
CN113343564A (en) Transformer top layer oil temperature prediction method based on multi-element empirical mode decomposition
CN112067052A (en) Oil-immersed transformer fault diagnosis method based on feature selection
CN113485261A (en) CAEs-ACNN-based soft measurement modeling method
CN112085084B (en) Transformer fault diagnosis method based on multi-feature fusion common vector
CN114167180B (en) Oil-filled electrical equipment fault diagnosis method based on graph attention neural network
CN114595883A (en) Oil-immersed transformer residual life personalized dynamic prediction method based on meta-learning
CN111143968A (en) Method for identifying variable forgetting factor (TFF) deduction space of proton exchange membrane fuel cell
CN114266396A (en) Transient stability discrimination method based on intelligent screening of power grid characteristics
CN112085083B (en) Transformer fault diagnosis method based on similarity analysis strategy
CN112149054B (en) Construction and application of orthogonal neighborhood preserving embedding model based on time sequence expansion
Luo et al. Prediction for dissolved gas in power transformer oil based on TCN and GCN
CN108491878B (en) Fault classification diagnosis method based on multiple error generation models
CN116432359A (en) Variable topology network tide calculation method based on meta transfer learning
CN109214112B (en) Reliability simulation analysis method and system of fuel cell system
CN116070302A (en) Cable insulation state prediction method and device
CN115935814A (en) Transformer fault prediction method based on ARIMA-SVM model
CN114707596A (en) Intelligent transformer fault diagnosis method based on long-term and short-term memory network
CN114169118A (en) Power distribution network topological structure identification method considering distributed power supply output correlation
El Marghichi A Solar PV Model Parameter Estimation Based on the Enhanced Self-Organization Maps
Jessen et al. Grey-box Modeling of Reversible Solid Oxide Cell Stack’s Electrical Dynamics Based on Electrochemical Impedance Spectroscopy
CN115616437B (en) Intelligent debugging control method and system for EPC special power supply
CN116106430B (en) Acoustic emission technology-based refractory material cracking diagnosis method for casting
CN107870282A (en) OPLC failure modes and diagnostic method
CN117828314B (en) Method, device, equipment and storage medium for testing insulation resistance of charging gun

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
WW01 Invention patent application withdrawn after publication

Application publication date: 20201211

WW01 Invention patent application withdrawn after publication