CN108919041A - A kind of transformer winding state on-line monitoring method based on clustering - Google Patents
A kind of transformer winding state on-line monitoring method based on clustering Download PDFInfo
- Publication number
- CN108919041A CN108919041A CN201810463284.6A CN201810463284A CN108919041A CN 108919041 A CN108919041 A CN 108919041A CN 201810463284 A CN201810463284 A CN 201810463284A CN 108919041 A CN108919041 A CN 108919041A
- Authority
- CN
- China
- Prior art keywords
- class
- transformer
- vibration signal
- vibration
- curve
- 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
Classifications
-
- 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
- G01R31/50—Testing of electric apparatus, lines, cables or components for short-circuits, continuity, leakage current or incorrect line connections
- G01R31/72—Testing of electric windings
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M7/00—Vibration-testing of structures; Shock-testing of structures
Abstract
The invention discloses a kind of transformer winding state on-line monitoring method based on clustering, it includes：The vibration signal for acquiring transformer box wall, is standardized collected vibration signal；Vibration signal after standardization is intercepted, m group vibration signal is obtained；Clustering is carried out to the m group vibration signal of transformer, obtains transformer vibration performance curve；Calculating class cluster is Ct (k)In all kinds of vibration signals relative to transformer vibration performance curve ZtCurvilinear motion rate YciRoot-mean-square value, seek root-mean-square value average value be Ycm；Calculate transformer oscillating curve y ' to be detectedcRelative to transformer vibration performance curve ZtCurvilinear motion rate Y 'cMean-square value Y 'cm；By Y 'cmWith YcmIt is compared, Y 'cm≤YcmThen judge that transformer winding operating status is normal；Y′cm> YcmThen transformer winding is loosened or is deformed；This method can monitor out loosening and the deformation condition of transformer winding on-line effectively, in high sensitivity.
Description
Technical field
The present invention relates to a kind of transformer signal monitoring method more particularly to a kind of transformer windings based on clustering
State on_line monitoring method.
Background technique
Transformer is one of most important equipment in electric system, security implication of the stability run to electric system
It is great.With the increasingly increase of China's electrical network capacity, capacity of short circuit also constantly increases therewith, and what cutting-out of voltage changer was formed rushes
The huge electromagnetic force for hitting electric current generation constitutes serious threat to the mechanical strength of winding.Transformer by after sudden short circuit,
Loosening or slight deformation may occur first for its winding, and deformation of transformer winding has cumulative effect.Correspondingly, if for
Loosening or deformation cannot find and repair in time, then the loosening or deformation accumulation in transformer can make it anti-afterwards to a certain extent
Short-circuit capacity declines to a great extent, and can cause big accident even if under by lesser dash current and occur.
Therefore, conventional inspection after transformer experienced external short circuit accident or after operation a period of time in the process of running
In repairing, transformer winding how is effectively detected out with the presence or absence of loosening and deforming, to judge whether transformer needs to overhaul
Processing seems particularly significant, is an important means for ensureing transformer safety operation.If regarding transformer winding as a machine
Tool structural body can obtain then when any variation occurs for winding construction or stress from the variation of its mechanical vibration performance
Reflection.The vibration of winding is transmitted to transformer-cabinet by inside transformer structural connection, so transformer-cabinet surface is examined
The basket vibration characteristic of the vibration signal and transformer that measure has a close relationship, therefore the vibration signal on transformer-cabinet surface
Analysis can be used as an approach of transformer winding fault diagnosis.This method is passed by the vibration being adsorbed on transformer box wall
Sensor obtains the vibration signal of transformer, judges the situation of change of winding state accordingly.As long as the mechanical property of winding is (as tied
Structure deformation, pretightning force loosening etc.) it changes, it can be reflected from the variation of its mechanical vibration performance, thus significantly
Improve the sensitivity of detection.In addition, vibrating sensor is placed in vibration detection on tank wall and entire strong power system is not straight
The connection connect, any influence no for the normal operation of entire electrical system are a kind of more accurate, convenient, safe online
Monitoring method.But the shadow by many factors such as transformer mechanical structure, the dispersibility of technique and substation operation environment
Ring, how from transformer system of condition monitoring obtain magnanimity vibration signal obtain accurate winding state monitoring side
Method is always Research Challenges.
Summary of the invention
The technical problem to be solved by the present invention is to：Improve a kind of transformer winding state on-line monitoring based on clustering
Method diagnoses presence as transformer winding fault using the analysis of vibration signal on transformer-cabinet surface to solve the prior art
Due to by transformer mechanical structure, technique dispersibility and many factors such as substation operation environment influenced, Bu Nenggao
Effect accurately obtains accurate winding state monitoring from the magnanimity vibration signal that transformer system of condition monitoring obtains
Etc. technical problems.
Technical solution of the present invention is：
A kind of transformer winding state on-line monitoring method based on clustering, it includes：
Step 1, the vibration signal x (t) for acquiring transformer box wall, are standardized place to collected vibration signal x (t)
Reason；
Step 2 intercepts the vibration signal after standardization, and Δ t, intercepted length N are divided between interception1, will
Starting point at the minimum point of vibration signal as each group vibration signal obtains m group vibration signal；
Step 3 carries out clustering to the m group vibration signal of transformer, obtains transformer vibration performance curve；
Step 4, calculating class cluster are Ct (k)In all kinds of vibration signals relative to transformer vibration performance curve ZtCurve become
Rate YciRoot-mean-square value Ycmi(i=1 ..., h) seeks root-mean-square value YcmiAverage value be Ycm；
Step 5 calculates transformer oscillating curve y' to be detectedcRelative to transformer vibration performance curve ZtCurve become
Rate Y 'cMean-square value Y 'cm；
Step 6, by Y 'cmWith YcmIt is compared, if meeting Y 'cm≤Ycm, then judge that transformer winding operating status is normal；
If meeting Y 'cm> Ycm, then judge that transformer winding is loosened or deformed.
The calculation formula of vibration signal standardization described in step 1 is
In formula,For the average value of vibration signal x (t)；N is the length of vibration signal x (t)；Y (t) is at standardization
Vibration signal after reason.
The calculation method of transformer vibration performance curve described in step 3 includes：
Step 3.1, note m group data are C1 (0)、C2 (0)、……、Cm (0), and every group of data are considered as independent one kind, it calculates
Euclidean distance between every two class obtains m rank initial distance matrix D(0)；
Step 3.2 finds initial distance matrix D(0)The minimum value of middle all elements, is denoted as d0ij, in formula：I=1 ..., m,
J=1 ..., m, and meet i ≠ j, it will be with d0ijCorresponding two classes Ci (0)And Cj (0)It merges, obtains new class cluster Cij (1)={ Ci (0),Cj (0)}；
Step 3.3, by new class cluster Cij (1)Be considered as new m-1 class data without combined class；It calculates every in this m-1 class
The distance between two classes obtain new m-1 rank Distance matrix D(1), wherein the distance between every two class without merging uses Europe
Family name's distance calculates, class cluster Cij (1)It is calculated, is counted using intermediate distance formula with without the distance between combined m-2 class
Calculating formula is
In formula, h is class cluster Cij (1)In class number；dxpFor class cluster Cij (1)In p-th of class to x-th without merging
The Euclidean distance of class；dxpFor class cluster Cij (1)In q-th of class to x-th of Euclidean distance without combined class；dpqFor class cluster Cij (1)
In p-th of class and q-th of class Euclidean distance；
Step 3.4 finds Distance matrix D(1)The minimum value of middle all elements, is denoted as d1ij, wherein i=1 ..., m-1, j
=1 ..., m-1, and meet i ≠ j, it will be with d1ijCorresponding two classes Ci (1)And Cj (1)It merges, obtains new class cluster Cij (2)=
{Ci (1),Cj (1), repeat step step 3.3, the Distance matrix D obtained after k operation(k)The minimum value of middle all elements
dkijWhen less than given threshold or distance matrix dimension being 0, stop calculating, by Distance matrix D(k)In each group of data it is corresponding each
A class cluster is denoted as C1 (k)、C2 (k)、……、Cb (k)；
The largest number of class clusters including class are denoted as C by step 3.5t (k), 1≤t≤b, calculating class cluster Ct (k)Mass center Zt
As the vibration performance curve of transformer, mass center ZtCalculation formula be
In formula, h is class cluster Ct (k)In include class number；yjkFor class cluster CtIn j-th strip vibration signal curve.
Calculating class cluster described in step 4 is Ct (k)In all kinds of vibration signals relative to transformer vibration performance curve ZtSong
Linear change rate YciCalculation formula be
In formula, yiFor class cluster CtIn i-th vibration signal curve；For the mean value of i-th vibration signal curve；For
The mean value of transformer vibration performance curve.
The transformer oscillating curve y' to be detected of calculating described in step 5cRelative to transformer vibration performance curve ZtSong
Linear change rate Yc' calculation formula is
Beneficial effects of the present invention：
Transformer box wall vibration signal in certain time period is carried out clustering by the present invention, and it is special to obtain transformer vibration
All kinds of vibration signals in class cluster where sign curve and transformer vibration performance curve are relative to transformer vibration performance curve
Curvilinear motion rate root-mean-square value average value (i.e. Ycm), then by transformer oscillating curve to be detected relative to transformer
Root-mean-square value (the i.e. Y ' of vibration performance curvilinear motion ratecm) and above-mentioned average value (i.e. Ycm) be compared, according to the variation of the two
It may determine that the working condition of transformer winding out.
Present invention employs above-mentioned technical proposal, make it possible to through the real-time prison to transformer box wall vibration signal
Control directly to judge the working condition of transformer winding, and the judgment method is efficient, accurate and easy to implement, is convenient for operator
The exception of discovery transformer winding greatly reduces change to be overhauled in time according to abnormal conditions to transformer in time
The failure spoilage of depressor；The prior art is solved using the analysis of vibration signal on transformer-cabinet surface as transformer winding
The many factors such as the existing dispersibility and substation operation environment due to by transformer mechanical structure, technique of fault diagnosis
Influence, be unable to efficiently and accurately slave transformer system of condition monitoring obtain magnanimity vibration signal obtain accurately around
The technical problems such as group status monitoring.
Detailed description of the invention
Fig. 1 is the flow chart of the vibration signal clustering method used in embodiment of the present invention；
Fig. 2 is cluster analysis result schematic diagram obtained in the present embodiment；
Fig. 3 is using transformer vibration performance curve synoptic diagram obtained in the embodiment of the present invention.
Specific embodiment
The present invention provides a kind of transformer winding state on-line monitoring method based on clustering, it includes
(1) the vibration signal x (t) for acquiring transformer box wall, is standardized vibration signal x (t).The vibration
Dynamic signal normalization handles calculation formula
In formula,For the average value of vibration signal x (t)；N is the length of vibration signal x (t)；Y (t) is at standardization
Vibration signal after reason.
(2) vibration signal after standardization is intercepted, wherein Δ t, intercepted length N are divided between interception1,
By the starting point at the minimum point of vibration signal as each group vibration signal, m group vibration signal is obtained.
(3) clustering is carried out to the m group vibration signal of transformer, obtains transformer vibration performance curve.Specific steps
It is as follows：
3a. remembers that m group data are C1 (0)、C2 (0)、……、Cm (0), and every group of data are considered as independent one kind, calculate every two
Euclidean distance between class obtains m rank initial distance matrix D(0).The calculation method of Euclidean distance is in the art in this step
Common mathematical method, therefore inventor is no longer described in detail herein；
3b. finding initial distance matrix D(0)The minimum value of middle all elements, is denoted as d0ij, wherein i=1 ..., m, j=
1 ..., m, and meet i ≠ j.It will be with d0ijCorresponding two classes Ci (0)And Cj (0)It merges, obtains new class cluster Cij (1)={ Ci (0),
Cj (0)}；
3c. is by new class cluster Cij (1)Be considered as new m-1 class data without combined class, wherein refer to without combined class
Be m group data in step 3a do not include class Ci (0)And Cj (0)Data, total m-2 class.Remember that new m-1 class is C1 (0)、C2 (0)、……、Ci-1 (0)、Cij (1)、Cj-1 (0)、……、Cm-1 (0).The distance between every two class in this m-1 class is calculated, new m-1 is obtained
Rank Distance matrix D(1), wherein the distance between every two class without merging is calculated using Euclidean distance, class cluster Cij (1)With
It is calculated without the distance between combined m-2 class using intermediate distance formula, calculation formula is
In formula, h is class cluster Cij (1)In class number；dxpFor class cluster Cij (1)In p-th of class to x-th without merging
The Euclidean distance of class；dxpFor class cluster Cij (1)In q-th of class to x-th of Euclidean distance without combined class；dpqFor class cluster Cij (1)
In p-th of class and q-th of class Euclidean distance；
3d. finds Distance matrix D(1)The minimum value of middle all elements, is denoted as d1ij, wherein i=1 ..., m-1, j=
1 ..., m-1, and meet i ≠ j.It will be with d1ijCorresponding two classes Ci (1)And Cj (1)It merges, obtains new class cluster Cij (2)={ Ci (1),Cj (1)}.Repeat step 3c, the Distance matrix D obtained after k operation(k)The minimum value d of middle all elementskijLess than setting
When to determine threshold value or distance matrix dimension be 0, stop calculating, by Distance matrix D(k)In the corresponding each class cluster note of each group of data
For C1 (k)、C2 (k)、……、Cb (k)；
The largest number of class clusters including class are denoted as C by 3e.t (k), wherein 1≤t≤b calculates class cluster Ct (k)Mass center Zt
Vibration performance curve as transformer, wherein mass center ZtCalculation formula be
In formula, h is class cluster Ct (k)In include class number；yjkFor class cluster CtIn j-th strip vibration signal curve.
(4) calculating class cluster is Ct (k)In all kinds of vibration signals relative to transformer vibration performance curve ZtCurvilinear motion
Rate YciRoot-mean-square value Ycmi(i=1 ..., h) seeks root-mean-square value YcmiAverage value be Ycm.The curvilinear motion rate Yci
Calculation formula be
In formula, yiFor class cluster CtIn i-th vibration signal curve；For the mean value of i-th vibration signal curve；For
The mean value of transformer vibration performance curve.
The calculation method of root mean square is common mathematical method in the art in this step, thus inventor herein no longer into
The detailed description of row.
(5) transformer oscillating curve y' to be detected is calculatedcRelative to transformer vibration performance curve ZtCurvilinear motion rate
Y′cMean-square value Y 'cm, the curvilinear motion rate Y 'cCalculation formula be
(6) by Y 'cmWith YcmIt is compared, if meeting Y 'cm≤Ycm, then judge that transformer winding operating status is normal；If full
Sufficient Y 'cm> Ycm, then judge that transformer winding is loosened or deformed.
That is, the technical program is that the transformer box wall vibration signal in certain time period is carried out clustering,
All kinds of vibration signals in the class cluster where transformer vibration performance curve and transformer vibration performance curve are obtained relative to change
Average value (the i.e. Y of the root-mean-square value of the curvilinear motion rate of depressor vibration performance curvecm), then transformer to be detected is vibrated
Root-mean-square value (i.e. Y ' of the curve relative to transformer vibration performance curvilinear motion ratecm) and above-mentioned average value (i.e. Ycm) compared
Compared with according to the variation of the two it may determine that the working condition of transformer winding out.
It is monitored on-line using the 500kV transformer of certain substation of Utilities Electric Co. as subjects, according to the following steps
Judge the working condition of the transformer winding：
(1) the vibration signal x (t) for acquiring transformer box wall, is standardized vibration signal x (t), the vibration
Dynamic signal normalization handles calculation formula
In formula,For the average value of vibration signal x (t)；N is the length of vibration signal x (t)；Y (t) is at standardization
Vibration signal after reason.
(2) vibration signal after standardization is intercepted, wherein Δ t, intercepted length N are divided between interception1,
By the starting point at the minimum point of vibration signal as each group vibration signal, m group vibration signal is obtained.Herein, Δ t=5 minutes；
N1=10000；M=3000.
(3) clustering is carried out to the m group vibration signal of transformer, obtains transformer vibration performance curve, calculating process
Flow chart is as shown in Figure 1.Specific step is as follows：
3a. remembers that m group data are C1 (0)、C2 (0)、……、Cm (0), and every group of data are considered as independent one kind, calculate every two
Euclidean distance between class obtains m rank initial distance matrix D(0).The calculation method of Euclidean distance is in the art in this step
Common mathematical method, therefore inventor is no longer described in detail herein；
3b. finds initial distance matrix D(0)The minimum value of middle all elements, is denoted as d0ij, wherein i=1 ..., m, j=
1 ..., m, and meet i ≠ j.It will be with d0ijCorresponding two classes Ci (0)And Cj (0)It merges, obtains new class cluster Cij (1)={ Ci (0),
Cj (0)}；
3c. is by new class cluster Cij (1)Be considered as new m-1 class data without combined class, wherein refer to without combined class
Be m group data in step 3a do not include class Ci (0)And Cj (0)Data, total m-2 class.Remember that new m-1 class is C1 (0)、C2 (0)、……、Ci-1 (0)、Cij (1)、Cj-1 (0)、……、Cm-1 (0).The distance between every two class in this m-1 class is calculated, new m-1 is obtained
Rank Distance matrix D(1), wherein the distance between every two class without merging is calculated using Euclidean distance, class cluster Cij (1)With
It is calculated without the distance between combined m-2 class using intermediate distance formula, calculation formula is
In formula, h is class cluster Cij (1)In class number；dxpFor class cluster Cij (1)In p-th of class to x-th without merging
The Euclidean distance of class；dxpFor class cluster Cij (1)In q-th of class to x-th of Euclidean distance without combined class；dpqFor class cluster Cij (1)
In p-th of class and q-th of class Euclidean distance；
3d. finds Distance matrix D(1)The minimum value of middle all elements, is denoted as d1ij, wherein i=1 ..., m-1, j=
1 ..., m-1, and meet i ≠ j.It will be with d1ijCorresponding two classes Ci (1)And Cj (1)It merges, obtains new class cluster Cij (2)={ Ci (1),Cj (1)}.Repeat step 3c, the Distance matrix D obtained after k operation(k)The minimum value d of middle all elementskijLess than setting
When to determine threshold value or distance matrix dimension be 0, stop calculating, by Distance matrix D(k)In the corresponding each class cluster note of each group of data
For C1 (k)、C2 (k)、……、Cb (k).Herein, threshold value 1.0；K=3.As shown in Figure 2；
The largest number of class clusters including class are denoted as C by 3e.t (k), wherein 1≤t≤b calculates class cluster Ct (k)Mass center Zt
As the vibration performance curve of transformer, as shown in Figure 3.Wherein, mass center ZtCalculation formula be
In formula, h is class cluster Ct (k)In include class number；yjkFor class cluster CtIn j-th strip vibration signal curve.This
Place, t=2.
(4) calculating class cluster is Ct (k)In all kinds of vibration signals relative to transformer vibration performance curve ZtCurvilinear motion
Rate YciRoot-mean-square value Ycmi(i=1 ..., h) seeks root-mean-square value YcmiAverage value be Ycm.The curvilinear motion rate Yci
Calculation formula be
In formula, yiFor class cluster CtIn i-th vibration signal curve；For the mean value of i-th vibration signal curve；For
The mean value of transformer vibration performance curve.
The calculation method of root mean square is common mathematical method in the art in this step, thus inventor herein no longer into
The detailed description of row.
(5) transformer oscillating curve y' to be detected is calculatedcRelative to transformer vibration performance curve ZtCurvilinear motion rate
Y′cmMean-square value Y 'cm, the curvilinear motion rate Y 'cCalculation formula be
(6) by Y 'cmWith YcmIt is compared, if meeting Y 'cm≤Ycm, then judge that transformer winding operating status is normal；If full
Sufficient Y 'cm> Ycm, then judge that transformer winding is loosened or deformed.
Table 1 is shown in the present embodiment according to 5 groups of vibration signals of transformer in a period of time obtained by the above method
Analyze result.From table 1 it follows that the change rate of this 5 groups of vibration signals is all satisfied Y 'cm≤Ycm, illustrate at this time transformer around
Group is in normal condition.
Table 1
Claims (5)
1. a kind of transformer winding state on-line monitoring method based on clustering, it includes：
Step 1, the vibration signal x (t) for acquiring transformer box wall, are standardized collected vibration signal x (t)；
Step 2 intercepts the vibration signal after standardization, and Δ t, intercepted length N are divided between interception1, vibration is believed
Number minimum point at starting point as each group vibration signal, obtain m group vibration signal；
Step 3 carries out clustering to the m group vibration signal of transformer, obtains transformer vibration performance curve；
Step 4, calculating class cluster are Ct (k)In all kinds of vibration signals relative to transformer vibration performance curve ZtCurvilinear motion rate
YciRoot-mean-square value Ycmi(i=1 ..., h) seeks root-mean-square value YcmiAverage value be Ycm；
Step 5 calculates transformer oscillating curve y' to be detectedcRelative to transformer vibration performance curve ZtCurvilinear motion rate
Y'cMean-square value Y'cm；
Step 6, by Y'cmWith YcmIt is compared, if meeting Y'cm≤Ycm, then judge that transformer winding operating status is normal；If full
Sufficient Y'cm> Ycm, then judge that transformer winding is loosened or deformed.
2. a kind of transformer winding state on-line monitoring method based on clustering according to claim 1, feature
It is, the calculation formula of vibration signal standardization described in step 1 is
In formula,For the average value of vibration signal x (t)；N is the length of vibration signal x (t)；Y (t) is after standardization
Vibration signal.
3. a kind of transformer winding state on-line monitoring method based on clustering according to claim 1, feature
It is, the calculation method of transformer vibration performance curve described in step 3 includes：
Step 3.1, note m group data are C1 (0)、C2 (0)、……、Cm (0), and every group of data are considered as independent one kind, calculate every two
Euclidean distance between class obtains m rank initial distance matrix D(0)；
Step 3.2 finds initial distance matrix D(0)The minimum value of middle all elements, is denoted as d0ij, in formula：I=1 ..., m, j=
1 ..., m, and meet i ≠ j, it will be with d0ijCorresponding two classes Ci (0)And Cj (0)It merges, obtains new class cluster Cij (1)={ Ci (0),
Cj (0)}；
Step 3.3, by new class cluster Cij (1)Be considered as new m-1 class data without combined class；Calculate every two class in this m-1 class
The distance between, obtain new m-1 rank Distance matrix D(1), wherein the distance between every two class without merging using Euclidean away from
From being calculated, class cluster Cij (1)It is calculated, is calculated public using intermediate distance formula with without the distance between combined m-2 class
Formula is
In formula, h is class cluster Cij (1)In class number；dxpFor class cluster Cij (1)In p-th of class to x-th without combined class
Euclidean distance；dxpFor class cluster Cij (1)In q-th of class to x-th of Euclidean distance without combined class；dpqFor class cluster Cij (1)In
The Euclidean distance of p class and q-th of class；
Step 3.4 finds Distance matrix D(1)The minimum value of middle all elements, is denoted as d1ij, wherein i=1 ..., m-1, j=
1 ..., m-1, and meet i ≠ j, it will be with d1ijCorresponding two classes Ci (1)And Cj (1)It merges, obtains new class cluster Cij (2)={ Ci (1),Cj (1), repeat step step 3.3, the Distance matrix D obtained after k operation(k)The minimum value d of middle all elementskij
When less than given threshold or distance matrix dimension being 0, stop calculating, by Distance matrix D(k)In each group of data it is corresponding each
Class cluster is denoted as C1 (k)、C2 (k)、……、Cb (k)；
The largest number of class clusters including class are denoted as C by step 3.5t (k), 1≤t≤b, calculating class cluster Ct (k)Mass center ZtAs
The vibration performance curve of transformer, mass center ZtCalculation formula be
In formula, h is class cluster Ct (k)In include class number；yjkFor class cluster CtIn j-th strip vibration signal curve.
4. a kind of transformer winding state on-line monitoring method based on clustering according to claim 1, feature
It is, calculating class cluster described in step 4 is Ct (k)In all kinds of vibration signals relative to transformer vibration performance curve ZtCurve become
Rate YciCalculation formula be
In formula, yiFor class cluster CtIn i-th vibration signal curve；For the mean value of i-th vibration signal curve；For transformation
The mean value of device vibration performance curve.
5. a kind of transformer winding state on-line monitoring method based on clustering according to claim 1, feature
It is, the transformer oscillating curve y' to be detected of calculating described in step 5cRelative to transformer vibration performance curve ZtCurve
Change rate Yc' calculation formula is
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810463284.6A CN108919041B (en) | 2018-05-15 | 2018-05-15 | Transformer winding state online monitoring method based on cluster analysis |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810463284.6A CN108919041B (en) | 2018-05-15 | 2018-05-15 | Transformer winding state online monitoring method based on cluster analysis |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108919041A true CN108919041A (en) | 2018-11-30 |
| CN108919041B CN108919041B (en) | 2021-08-20 |
Family
ID=64403655
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810463284.6A Active CN108919041B (en) | 2018-05-15 | 2018-05-15 | Transformer winding state online monitoring method based on cluster analysis |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108919041B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111983524A (en) * | 2020-08-26 | 2020-11-24 | 西南交通大学 | Transformer winding fault assessment method based on oscillatory wave time-frequency transformation |
| CN112985676A (en) * | 2021-01-28 | 2021-06-18 | 国网江苏省电力有限公司南京供电分公司 | On-line monitoring method for fastener looseness based on transformer vibration characteristics |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101782426A (en) * | 2010-01-29 | 2010-07-21 | 浙江大学 | Detection method of looseness fault vibration of power transformer winding |
| JP2014194727A (en) * | 2013-03-28 | 2014-10-09 | Toenec Corp | Rotary machine quality diagnostic system |
| CN105547465A (en) * | 2015-12-08 | 2016-05-04 | 华北电力大学（保定） | Transformer vibration signal winding state feature extraction method |
| CN106932184A (en) * | 2017-03-15 | 2017-07-07 | 国网四川省电力公司广安供电公司 | A kind of Diagnosis Method of Transformer Faults based on improvement hierarchical clustering |
| CN107290041A (en) * | 2017-07-28 | 2017-10-24 | 河海大学 | It is a kind of that state monitoring method is loosened based on phase space reconfiguration and the KPCM Transformer Winding clustered |
| CN107730084A (en) * | 2017-09-18 | 2018-02-23 | 杭州安脉盛智能技术有限公司 | Repair of Transformer decision-making technique based on gray prediction and risk assessment |
-
2018
- 2018-05-15 CN CN201810463284.6A patent/CN108919041B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101782426A (en) * | 2010-01-29 | 2010-07-21 | 浙江大学 | Detection method of looseness fault vibration of power transformer winding |
| JP2014194727A (en) * | 2013-03-28 | 2014-10-09 | Toenec Corp | Rotary machine quality diagnostic system |
| CN105547465A (en) * | 2015-12-08 | 2016-05-04 | 华北电力大学（保定） | Transformer vibration signal winding state feature extraction method |
| CN106932184A (en) * | 2017-03-15 | 2017-07-07 | 国网四川省电力公司广安供电公司 | A kind of Diagnosis Method of Transformer Faults based on improvement hierarchical clustering |
| CN107290041A (en) * | 2017-07-28 | 2017-10-24 | 河海大学 | It is a kind of that state monitoring method is loosened based on phase space reconfiguration and the KPCM Transformer Winding clustered |
| CN107730084A (en) * | 2017-09-18 | 2018-02-23 | 杭州安脉盛智能技术有限公司 | Repair of Transformer decision-making technique based on gray prediction and risk assessment |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111983524A (en) * | 2020-08-26 | 2020-11-24 | 西南交通大学 | Transformer winding fault assessment method based on oscillatory wave time-frequency transformation |
| CN111983524B (en) * | 2020-08-26 | 2021-06-08 | 西南交通大学 | Transformer winding fault assessment method based on oscillatory wave time-frequency transformation |
| CN112985676A (en) * | 2021-01-28 | 2021-06-18 | 国网江苏省电力有限公司南京供电分公司 | On-line monitoring method for fastener looseness based on transformer vibration characteristics |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108919041B (en) | 2021-08-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| TWI422460B (en) | Tool nose detection method for cutting machine tool | |
| CN206114822U (en) | Many information detection means of power transformer winding deformation state | |
| CN106443316A (en) | Power transformer winding deformation state multi-information detection method and device | |
| CN105738722B (en) | A kind of Aviation ElctroMechanical actuator method for diagnosing faults | |
| CN103983881B (en) | The method for diagnosing faults of nuclear detector and device | |
| GB2476246A (en) | Diagnosing an operation mode of a machine | |
| CN108536945A (en) | A kind of fault diagnosis method and system for large-scale phase modifier | |
| CN110954354B (en) | Equipment fault diagnosis method and device | |
| CN108872852A (en) | A kind of wind-driven generator fault diagnosis system and method | |
| CN110222765A (en) | A kind of permanent magnet synchronous motor health status monitoring method and system | |
| CN114255784A (en) | Substation equipment fault diagnosis method based on voiceprint recognition and related device | |
| CN108919041A (en) | A kind of transformer winding state on-line monitoring method based on clustering | |
| CN103821750B (en) | A kind of ventilator stall based on electric current and surge monitoring and diagnostic method | |
| CN109581268A (en) | A kind of optical fiber current mutual inductor method for diagnosing faults and device | |
| CN206193186U (en) | Isolator mechanical breakdown diagnostic device | |
| CN116315173A (en) | Battery temperature sampling system based on new energy automobile | |
| CN109613372B (en) | Power grid fault diagnosis method based on multi-element power grid database | |
| CN104569814B (en) | A kind of DC traction motor health status real-time analysis method based on approximate entropy | |
| CN101968379B (en) | Method for extracting characteristic information of operating condition vibration signal of aircraft engine rotor system | |
| CN115128345B (en) | Power grid safety early warning method and system based on harmonic monitoring | |
| CN105954620A (en) | Power fault diagnosis system and method | |
| CN109489719A (en) | A kind of inspection detection artificial intelligence automatic checkout system | |
| CN109034028A (en) | A kind of mechanical equipment fault feature extracting method based on symbolism and TF-IDF | |
| Ma et al. | GIS mechanical fault diagnosis and test analysis based on S transform-SVM-AlexNet model | |
| CN104267337A (en) | All-in-one detecting platform and state recognition system based on circuit breaker operating mechanism |
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 |