CN108510002B - Method for detecting short circuit impact resistance of wound core traction transformer winding - Google Patents

Method for detecting short circuit impact resistance of wound core traction transformer winding Download PDF

Info

Publication number
CN108510002B
CN108510002B CN201810296549.8A CN201810296549A CN108510002B CN 108510002 B CN108510002 B CN 108510002B CN 201810296549 A CN201810296549 A CN 201810296549A CN 108510002 B CN108510002 B CN 108510002B
Authority
CN
China
Prior art keywords
short
frequency
sample
circuit
frequency response
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.)
Active
Application number
CN201810296549.8A
Other languages
Chinese (zh)
Other versions
CN108510002A (en
Inventor
高仕斌
周利军
江俊飞
严静荷
郭蕾
吴振宇
李威
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.)
Southwest Jiaotong University
Original Assignee
Southwest Jiaotong 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 Southwest Jiaotong University filed Critical Southwest Jiaotong University
Priority to CN201810296549.8A priority Critical patent/CN108510002B/en
Publication of CN108510002A publication Critical patent/CN108510002A/en
Application granted granted Critical
Publication of CN108510002B publication Critical patent/CN108510002B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/11—Complex mathematical operations for solving equations, e.g. nonlinear equations, general mathematical optimization problems
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • General Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Artificial Intelligence (AREA)
  • Computational Mathematics (AREA)
  • Mathematical Analysis (AREA)
  • Mathematical Optimization (AREA)
  • Evolutionary Computation (AREA)
  • Pure & Applied Mathematics (AREA)
  • Operations Research (AREA)
  • Algebra (AREA)
  • Databases & Information Systems (AREA)
  • Software Systems (AREA)
  • Complex Calculations (AREA)
  • Investigating Strength Of Materials By Application Of Mechanical Stress (AREA)

Abstract

The invention discloses a method for detecting short-circuit impact resistance of a winding of a wound core traction transformer, which mainly comprises the following steps: testing the frequency response of each winding before and after short circuit; binarizing the curve graph and calculating image characteristic parameters in different frequency bands; fitting frequency response curves in different frequency intervals; calculating coefficients of each order under different conditions; and detecting the effect of the short-circuit resistance by taking the image parameters and the transfer function coefficients as input quantities. The detection method for the short-circuit impact resistance of the wound core traction transformer winding provided by the invention has the advantages of richer information, more accurate evaluation and more intelligent operation.

Description

Method for detecting short circuit impact resistance of wound core traction transformer winding
Technical Field
The invention belongs to the technical field of large-scale wound core transformer tests, and particularly relates to a method for detecting short circuit impact resistance of a wound core traction transformer winding.
Background
The traction transformer is one of core devices in a traction power supply system, and traction load has the characteristic of intermittence, so that no-load loss becomes one of main losses of the traction transformer due to long-term no-load operation, the system operation economy is directly influenced, the wound core technology is a mature and reliable energy-saving technology which can be applied to a large traction transformer at present, and the operation economy of the traction power supply system can be effectively improved.
The wound core has the characteristics of small magnetic resistance, low loss and the like because no air gap exists in a magnetic circuit, and the large wound core usually adopts a closed structure in order to ensure that the wound core has low no-load loss. And coil winding mode on the large-scale closed book iron core is different from the detachable iron core of conventional lamination formula, must directly carry out the coiling on the iron core post rather than carrying out holistic concatenation and installation in later stage equipment, if coil winding is unqualified or breaks down, is difficult to directly replace, and cost of maintenance is higher. The traction load has the characteristic of impact and has higher requirement on the short-circuit resistance of the transformer winding, so that the service life of the wound core traction transformer can be effectively prolonged if the short-circuit resistance of the wound core traction transformer can be effectively detected, evaluated and taken when the wound core traction transformer leaves a factory.
The existing method for judging the short-circuit impact resistance of the winding mainly comprises the following steps: 1) measuring and comparing the reactance changes of the windings before and after short circuit to judge the short circuit resistance; 2) the short-circuited front and rear suspension covers check whether there is a significant change in the winding. However, the above method has a problem of low sensitivity, and some hidden defects are difficult to be found. The frequency response method is widely applied to detection of field transformer winding faults due to high precision and interference resistance, and can be used for detection of short circuit resistance of windings in the delivery process of the wound core traction transformer, but the existing diagnosis standard is to judge through a correlation coefficient, the judgment coefficient is single, the accuracy is not high enough, and field technicians with abundant experience are required to carry out auxiliary judgment. Therefore, a method with more abundant information, more accurate evaluation and more intellectualization is urgently needed to improve the detection of the short-circuit impact resistance of the winding of the wound core traction transformer.
Disclosure of Invention
In view of the technical defects, the invention provides a method for detecting the short-circuit impact resistance of a winding of a wound core traction transformer, which aims to more reliably and effectively obtain the short-circuit impact resistance of the wound core traction transformer.
The purpose of the invention is realized by the following means:
1) before the short-circuit impact test of the wound core traction transformer, the frequency response of each winding is respectively measured to obtain the reference frequency response Y based on the frequency f0(f) The frequency test range is 1kHz to 1 MHz;
2) performing short-circuit impact test on the wound core traction transformer, and measuring when the short-circuit impact test is finishedThe frequency response of each winding is obtained to obtain a test frequency response Y based on the frequency f1(f) The frequency test range is 1kHz to 1 MHz;
3) the frequency range is divided into 4 intervals: f1: 1kHz to 100kHz, F2: 100kHz to 300kHz, F3: 300kHz to 600kHz, F4: 600kHz to 1 MHz; drawing F under the same coordinate respectively1～F4Frequency response Y in these 4 intervals0(f) And Y1(f) Wherein the abscissa is frequency f and the ordinate is frequency response amplitude Y;
4) respectively draw F1～F4Frequency response curves in the 4 intervals, wherein the abscissa is frequency f, the ordinate is frequency response amplitude Y, and the area ratio S between the curves is obtainedrAnd CM (r, theta) is deviated from the curve centroid, and the specific method is as follows:
a. binarizing the curve graph if Y0(f)≤Y(f)≤Y1(f) Or Y1(f)≤Y(f)≤Y0(f) Then the image value is 1, otherwise 0, and the image resolution size is defined as follows:
w is the binary image width and H is the binary image height, where fstart、fendRespectively the start and end frequencies of the frequency interval, Ymax、Ymin△ f and △ Y are the frequency difference and amplitude difference represented by the unit pixel;
calculating the proportion S of 1 part of the image value in each interval to the whole imager：
Figure BDA0001615670330000022
M1The sum of the number of pixels of which the pixel is 1 in the image;
b. respectively binarizing the curve graphs, if Y (f) is less than or equal to Y0(f) The image value is 1, otherwise 0, the image S1 is obtained, if Y (f) is less than or equal to Y1(f)，If the image value is 1, otherwise, the image value is 0, and an image S2 is obtained, wherein the resolution size of the image is consistent with that in the step a;
respectively calculating the centroids of the parts with the image values of 1 in S1 and S2 in each interval based on the regionprops function:
C1＝(f0,Y0),C2＝(f1,Y1)
C1and C2Calculating a curve centroid shift CM (r, θ) for the reference and test frequency response centroid coordinates, respectively:
Figure BDA0001615670330000023
Figure BDA0001615670330000024
r is the centroid offset distance, and theta is the centroid offset angle;
5) fitting Y in the interval from F1 to F40(f) And Y1(f) The method comprises the following steps of calculating coefficients of each order of rational fraction numerator and denominator of a frequency domain transfer function, and specifically comprises the following steps:
a. fitting the frequency response curve by using a rapid relaxation vector matching algorithm to obtain a transfer function H(s) under a frequency domain, wherein the expression is as follows:
Figure BDA0001615670330000031
wherein s is Laplace operator, a0、a1、…、an-1And b0、b1、…、bn-1Respectively are coefficients of each order;
b. for each order coefficient a0、a1、…、an-1And b1、…、bn-1Performing sorting deformation to obtain a parameter sequence X thereofa，XbThe following were used:
Xa＝[log|a0| log|a1| … log|an-1|]，Xb＝[log|b1| log|b2| … log|bn-1|]
6) the parameter sequence X1 obtained by fitting in the first frequency interval is ═ Xa1，Xb1，Sr1，r1，θ1]As a first type of input feature of the least squares support vector machine, a parameter sequence X2 ═ X obtained by fitting in a second frequency intervala2，Xb2，Sr2，r2，θ2]As a second input feature of the least squares support vector machine, a parameter sequence X3 ═ X obtained by fitting in a third frequency intervala3，Xb3，Sr3，r3，θ3]As a third input feature of the least squares support vector machine, a parameter sequence X4 ═ X obtained by fitting in a fourth frequency intervala4，Xb4，Sr4，r4，θ4]The fourth kind of input features are used as a least square support vector machine; the four types of input features are combined to obtain a total input matrix X5 ═ X1, X2, X3 and X4]；
7) Reducing the dimension of an input matrix X5 by using a principal component analysis method, calculating the accumulated contribution rate of characteristic values, selecting characteristic vectors corresponding to the first 90% of the characteristic values to form a dimension reduction matrix, removing redundant components, and reconstructing to obtain an input matrix X6 of a sample 1, wherein the column number n represents a characteristic dimension;
8) obtaining m groups of traction transformer frequency response data with different short-circuit resistance capabilities through field transformer field tests, obtaining m sample input matrixes through steps 3), 4) and 5), and finally obtaining a total sample input matrix Rm×nThe row number m represents the number of input samples, and the column number n represents the feature dimension;
9) in a random manner at the input matrix Rm×nThe method comprises the following steps of extracting a training sample X7 and a test sample X8 in proportion, training a least square support vector machine model by using X7 to obtain optimal parameters in the model, and testing the accuracy of the model by using X8, wherein the method comprises the following specific steps:
a. normalizing the X7 and the X8;
b. introducing a radial basis kernel function
Figure BDA0001615670330000032
Mapping the nonlinear samples into a high-dimensional linear separable space, and searching optimal values of a parameter sigma and a penalty coefficient C by combining a grid search algorithm;
c. and (3) performing derivation on the lagrange function corresponding to the least square support vector machine, and obtaining a linear equation set according to a KKT condition:
Figure BDA0001615670330000041
the classification function of the least squares support vector machine obtained by solving the linear equation set is:
Figure BDA0001615670330000042
wherein f (x) > 0 is a positive sample, and the corresponding label is 1; f (x) < 0 is a negative sample, corresponding to a label of-1;
d. in the above steps, the corresponding class labels of good, general and poor short-circuit resistance of the winding are respectively 1, 2, 3 and 4 in the design of the multi-classifier. Utilizing MOC (minimum Output coding) codes to construct a plurality of two classifiers, and specifically realizing the steps as follows:
d1, normalizing the X7 and sending the normalized X7 to each single classifier to obtain an output vector h (X) (h1(X), h2 (X));
d2 calculating M between output vector and each column vector of coding matrix by using Hamming distanceyDistance D (M)y,H(x))；
D3, taking the minimum distance D (M)yH (x)) as the predicted output, η (x) argmin { D (M) is the degree of short-circuit resistance obtainedy,H(x))}；
10) When the short-circuit resistance of the wound core traction transformer needs to be evaluated, measuring to obtain frequency responses before and after a transformer short-circuit impact test, obtaining a sample input matrix through steps 3), 4), 5), 6) and 7), and substituting the input matrix into a classifier to calculate to obtain the short-circuit resistance.
According to the invention, the image characteristics and the transfer function characteristics of the frequency response curve in different frequency intervals are extracted by measuring the frequency responses before and after the short-circuit impact test of the wound core traction transformer, the training of a least square support vector machine is carried out based on field data to obtain a classification function, and the short-circuit impact resistance of the transformer is detected. The method increases the feature dimension in the classification judgment process, has richer feature types, improves the classification accuracy, and can continuously improve the classification precision and the detection efficiency by continuously increasing the training samples.
Drawings
FIG. 1 is a flow chart of a method for detecting short-circuit impact resistance provided by the present invention
FIG. 2 is a graph of actual classification and predicted classification of a test set
Detailed Description
The following describes the process of the present invention in detail with reference to the accompanying drawings.
As shown in fig. 1, a method for detecting short-circuit impact resistance of a winding of a wound core traction transformer includes the following steps:
1) before the short-circuit impact test of the wound core traction transformer, the frequency response of each winding is respectively measured to obtain the reference frequency response Y based on the frequency f0(f) The frequency test range is 1kHz to 1 MHz;
2) carrying out short-circuit impact test on the wound core traction transformer, and measuring the frequency response of each winding when the short-circuit impact test is finished to obtain a test frequency response Y based on the frequency f1(f) The frequency test range is 1kHz to 1 MHz;
3) the frequency range is divided into 4 intervals: f1: 1kHz to 100kHz, F2: 100kHz to 300kHz, F3: 300kHz to 600kHz, F4: 600kHz to 1 MHz; drawing F under the same coordinate respectively1～F4Frequency response Y in these 4 intervals0(f) And Y1(f) Wherein the abscissa is frequency f and the ordinate is frequency response amplitude Y;
4) respectively draw F1～F4Frequency response curves in the 4 intervals, wherein the abscissa is frequency f, the ordinate is frequency response amplitude Y, and the area ratio S between the curves is obtainedrAnd CM (r, theta) is deviated from the curve centroid, and the specific method is as follows:
a. binarizing the curve graph if Y0(f)≤Y(f)≤Y1(f) Or Y1(f)≤Y(f)≤Y0(f) Then the image value is 1, otherwise 0, and the image resolution size is defined as follows:
Figure BDA0001615670330000051
w is the binary image width and H is the binary image height, where fstart、fendRespectively the start and end frequencies of the frequency interval, Ymax、Ymin△ f and △ Y are the frequency difference and amplitude difference represented by the unit pixel;
calculating the proportion S of 1 part of the image value in each interval to the whole imager：
Figure BDA0001615670330000052
M1The sum of the number of pixels of which the pixel is 1 in the image;
b. respectively binarizing the curve graphs, if Y (f) is less than or equal to Y0(f) The image value is 1, otherwise 0, the image S1 is obtained, if Y (f) is less than or equal to Y1(f) If yes, the image value is 1, otherwise, the image value is 0, and an image S2 is obtained, wherein the resolution size of the image is consistent with that in the step a;
respectively calculating the centroids of the parts with the image values of 1 in S1 and S2 in each interval based on the regionprops function:
C1＝(f0,Y0),C2＝(f1,Y1)
C1and C2Calculating a curve centroid shift CM (r, θ) for the reference and test frequency response centroid coordinates, respectively:
Figure BDA0001615670330000053
Figure BDA0001615670330000054
r is the centroid offset distance, and theta is the centroid offset angle;
5) fitting Y in the interval from F1 to F40(f) And Y1(f) The method comprises the following steps of calculating coefficients of each order of rational fraction numerator and denominator of a frequency domain transfer function, and specifically comprises the following steps:
a. fitting the frequency response curve by using a rapid relaxation vector matching algorithm to obtain a transfer function H(s) under a frequency domain, wherein the expression is as follows:
Figure BDA0001615670330000061
wherein s is Laplace operator, a0、a1、…、an-1And b0、b1、…、bn-1Respectively are coefficients of each order;
b. for each order coefficient a0、a1、…、an-1And b1、…、bn-1Performing sorting deformation to obtain a parameter sequence X thereofa，XbThe following were used:
Xa＝[log|a0| log|a1| … log|an-1|]，Xb＝[log|b1| log|b2| … log|bn-1|]
6) the parameter sequence X1 obtained by fitting in the first frequency interval is ═ Xa1，Xb1，Sr1，r1，θ1]As a first type of input feature of the least squares support vector machine, a parameter sequence X2 ═ X obtained by fitting in a second frequency intervala2，Xb2，Sr2，r2，θ2]As a second input feature of the least squares support vector machine, a parameter sequence X3 ═ X obtained by fitting in a third frequency intervala3，Xb3，Sr3，r3，θ3]As a third input feature of the least squares support vector machine, a parameter sequence X4 ═ X obtained by fitting in a fourth frequency intervala4，Xb4，Sr4，r4，θ4]The fourth kind of input features are used as a least square support vector machine; the four types of input features are combined to obtain a total input matrix X5 ═ X1, X2, X3 and X4]；
7) Reducing the dimension of an input matrix X5 by using a principal component analysis method, calculating the accumulated contribution rate of characteristic values, selecting characteristic vectors corresponding to the first 90% of the characteristic values to form a dimension reduction matrix, removing redundant components, and reconstructing to obtain an input matrix X6 of a sample 1, wherein the column number n represents a characteristic dimension;
8) obtaining m groups of traction transformer frequency response data with different short-circuit resistance capabilities through field transformer field tests, obtaining m sample input matrixes through steps 3), 4) and 5), and finally obtaining a total sample input matrix Rm×nThe row number m represents the number of input samples, and the column number n represents the feature dimension;
9) in a random manner at the input matrix Rm×nThe method comprises the following steps of extracting a training sample X7 and a test sample X8 in proportion, training a least square support vector machine model by using X7 to obtain optimal parameters in the model, and testing the accuracy of the model by using X8, wherein the method comprises the following specific steps:
a. normalizing the X7 and the X8;
b. introducing a radial basis kernel function
Figure BDA0001615670330000062
Mapping the nonlinear samples into a high-dimensional linear separable space, and searching optimal values of a parameter sigma and a penalty coefficient C by combining a grid search algorithm;
c. and (3) performing derivation on the lagrange function corresponding to the least square support vector machine, and obtaining a linear equation set according to a KKT condition:
Figure BDA0001615670330000063
the classification function of the least squares support vector machine obtained by solving the linear equation set is:
Figure BDA0001615670330000071
wherein f (x) > 0 is a positive sample, and the corresponding label is 1; f (x) < 0 is a negative sample, corresponding to a label of-1;
d. in the above steps, the corresponding class labels of good, general and poor short-circuit resistance of the winding are respectively 1, 2, 3 and 4 in the design of the multi-classifier. Utilizing MOC (minimum Output coding) codes to construct a plurality of two classifiers, and specifically realizing the steps as follows:
d1, normalizing the X7 and sending the normalized X7 to each single classifier to obtain an output vector h (X) (h1(X), h2 (X));
d2 calculating M between output vector and each column vector of coding matrix by using Hamming distanceyDistance D (M)y,H(x))；
D3, taking the minimum distance D (M)yH (x)) as the predicted output, η (x) argmin { D (M) is the degree of short-circuit resistance obtainedy,H(x))}；
10) When the short-circuit resistance of the wound core traction transformer needs to be evaluated, measuring to obtain frequency responses before and after a transformer short-circuit impact test, obtaining a sample input matrix through steps 3), 4), 5), 6) and 7), and substituting the input matrix into a classifier to calculate to obtain the short-circuit resistance.
Fig. 2 shows classification results obtained based on a finite element simulation test platform, wherein the influence caused by short circuit impact is simulated by artificially displacing and deforming a transformer winding in a finite element simulation model, the transformer winding is divided into four degrees, the four degrees correspond to four short circuit resistance capabilities respectively, and simulation, calculation and classification are performed on 28 simulation conditions, wherein for the four defined conditions, the accuracy of the classification results reaches 90%, and the accuracy is high.
The method for detecting the short-circuit impact resistance of the winding of the wound core traction transformer can accurately identify some hidden defects possibly existing after a short-circuit impact test of the coil, effectively evaluate the short-circuit resistance of the coil, take measures in a targeted manner, and avoid potential safety hazards existing when the transformer is assembled and leaves a factory.

Claims (2)

1. The method for detecting the short-circuit impact resistance of the winding of the wound core traction transformer is characterized by comprising the following steps of:
1) before the short-circuit impact test of the wound core traction transformer, the frequency response of each winding is respectively measured to obtain the reference frequency response Y based on the frequency f0(f) The frequency test range is 1kHz to 1 MHz;
2) carrying out short-circuit impact test on the wound core traction transformer, and measuring the frequency response of each winding when the short-circuit impact test is finished to obtain a test frequency response Y based on the frequency f1(f) The frequency test range is 1kHz to 1 MHz;
3) the frequency range is divided into 4 intervals: f1：1kHz≤f≤100kHz，F2：100kHz＜f≤300kHz，F3：300kHz＜f≤600kHz，F4: f is more than 600kHz and less than or equal to 1 MHz; drawing F under the same coordinate respectively1～F4Frequency response Y in these 4 intervals0(f) And Y1(f) Wherein the abscissa is frequency f and the ordinate is frequency response amplitude Y;
4) extracting the frequency response Y in the interval from F1 to F40(f) And Y1(f) Respectively obtaining characteristic parameter sequences X1-X4 in each interval;
5) taking X1-X4 as input features of a least square support vector machine, and combining the four input features to obtain a total input matrix X5 ═ X1, X2, X3 and X4; performing dimensionality reduction on an input matrix X5 by using a principal component analysis method, and obtaining an input matrix X6 of a sample 1 through reconstruction, wherein the number of columns represents a characteristic dimension;
6) obtaining m groups of traction transformer frequency response data with different short-circuit resistance capabilities through field transformer field tests, obtaining m sample input matrixes through steps 3), 4) and 5), and finally obtaining a total sample input matrix Rm×nThe row number m represents the number of input samples, and the column number n represents the feature dimension;
7) in a random manner at the input matrix Rm×nThe method comprises the steps of extracting a training sample X7 and a test sample X8 according to a proportion, training a least square support vector machine model by using X7 to obtain an optimal parameter in the model, testing the accuracy of the model by using X8, and constructing a plurality of two classifiers by using MOC (Minimum Output Coding), wherein in the design of the two classifiers, η (X) is taken as a predicted Output value, if the Output of η (X) is 1, the short-circuit resistance of a winding is good, if the Output of η (X) is 2, the short-circuit resistance is good, if the Output of η (X) is 3, the short-circuit resistance is general, and if the Output of η (X) is 4, the short-circuit resistance is poor;
8) when the short-circuit resistance of the wound core traction transformer is evaluated, measuring to obtain frequency responses before and after a transformer short-circuit impact test, obtaining a sample input matrix through steps 3), 4) and 5), and substituting the input matrix into a classifier for calculation to obtain the short-circuit resistance;
the characteristic parameters include: (1) area ratio between curves SrThe curve centroid shift CM (r, θ); (2) the frequency domain transfer function has rational numerator and denominator order coefficients.
2. The method of claim 1, wherein each new sample is added to a sample input matrix Rm×nAnd input the matrix R at the new sample(m+1)×nAnd randomly extracting a training sample X7 and a testing sample X8, training the training samples after normalization processing, and improving the precision of a classification function by combining the real situation of the transformer.
CN201810296549.8A 2018-04-02 2018-04-02 Method for detecting short circuit impact resistance of wound core traction transformer winding Active CN108510002B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810296549.8A CN108510002B (en) 2018-04-02 2018-04-02 Method for detecting short circuit impact resistance of wound core traction transformer winding

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810296549.8A CN108510002B (en) 2018-04-02 2018-04-02 Method for detecting short circuit impact resistance of wound core traction transformer winding

Publications (2)

Publication Number Publication Date
CN108510002A CN108510002A (en) 2018-09-07
CN108510002B true CN108510002B (en) 2020-02-21

Family

ID=63380345

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810296549.8A Active CN108510002B (en) 2018-04-02 2018-04-02 Method for detecting short circuit impact resistance of wound core traction transformer winding

Country Status (1)

Country Link
CN (1) CN108510002B (en)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110514924B (en) * 2019-08-12 2021-04-27 武汉大学 Power transformer winding fault positioning method based on deep convolutional neural network fusion visual identification
CN114675118B (en) * 2022-05-30 2022-08-23 广东电网有限责任公司佛山供电局 Transformer winding abnormality detection method, device, equipment and storage medium
CN116127240B (en) * 2022-11-22 2023-12-05 西南交通大学 Evaluation method for overload capacity of wound core of traction transformer

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104596767B (en) * 2015-01-13 2017-04-26 北京工业大学 Method for diagnosing and predicating rolling bearing based on grey support vector machine
CN106646096B (en) * 2016-11-15 2019-07-02 国网四川省电力公司广安供电公司 Transformer fault classification and recognition methods based on vibration analysis method
CN106646098B (en) * 2016-12-31 2018-04-13 西南交通大学 Winding frequency response test platform and method between lamination under short trouble
CN106771816B (en) * 2016-12-31 2018-07-31 西南交通大学 Core loss test platform and test method under rewinding material multipoint earth faults
CN107808761A (en) * 2017-12-04 2018-03-16 河南省森电电力设备股份有限公司 A kind of anti-sudden short circuit three dimensional wound core transformer

Also Published As

Publication number Publication date
CN108510002A (en) 2018-09-07

Similar Documents

Publication Publication Date Title
US11474163B2 (en) Power transformer winding fault positioning method based on deep convolutional neural network integrated with visual identification
CN106646096B (en) Transformer fault classification and recognition methods based on vibration analysis method
CN108510002B (en) Method for detecting short circuit impact resistance of wound core traction transformer winding
US20200200813A1 (en) Online diagnosis method for deformation position on trasnformation winding
US20190243735A1 (en) Deep belief network feature extraction-based analogue circuit fault diagnosis method
CN103076547B (en) Method for identifying GIS (Gas Insulated Switchgear) local discharge fault type mode based on support vector machines
Zhao et al. Diagnosing transformer winding deformation faults based on the analysis of binary image obtained from FRA signature
CN109188211B (en) High-voltage equipment insulation fault diagnosis method and system
CN108896161B (en) A kind of assessment method of Wound iron-core transformer winding failure analog platform
CN110118928B (en) Breaker fault diagnosis method based on error inverse propagation algorithm
CN112149569B (en) Voiceprint fault diagnosis method of transformer based on fuzzy C-means clustering algorithm
CN108872803B (en) Method for detecting insulation state of transformer based on insulation medium return voltage
CN110543730A (en) Transformer winding deformation fault determination method
CN116087647A (en) Building electrical fault diagnosis method for optimizing random forest based on PCA and sparrow algorithm
CN114814501A (en) On-line diagnosis method for capacitor breakdown fault of capacitor voltage transformer
CN111079647A (en) Circuit breaker defect identification method
CN112132226A (en) Method for diagnosing defects of +/-1100 kV composite insulator
CN106250937B (en) A kind of failure modes diagnostic method based on non-index of similarity
CN117406011A (en) Charging cable abnormality detection device, method and equipment based on charging vehicle data
CN112287953A (en) Method and system for GIS insulation defect category identification
CN109978250A (en) A kind of qualitative forecasting method of battery pole piece coating production surface density
CN110991240B (en) Single-phase fault identification and classification method for low-voltage neutral point ungrounded system
CN103018658B (en) Circuit board health condition monitoring method based on volt-ampere characteristic curves
CN110399911A (en) A kind of transformer mechanical failure diagnostic method based on sound-filed simulation cloud atlas
Qi et al. Research on Transformer Fault Detection Based on CEEMDAN-DBN

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