CN106291275B - A kind of local discharge superhigh frequency single waveform frequency domain character extracts and recognition methods - Google Patents
A kind of local discharge superhigh frequency single waveform frequency domain character extracts and recognition methods Download PDFInfo
- Publication number
- CN106291275B CN106291275B CN201610601321.6A CN201610601321A CN106291275B CN 106291275 B CN106291275 B CN 106291275B CN 201610601321 A CN201610601321 A CN 201610601321A CN 106291275 B CN106291275 B CN 106291275B
- Authority
- CN
- China
- Prior art keywords
- electric discharge
- single waveform
- discharge type
- obtains
- identification
- 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
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
- G01R31/12—Testing dielectric strength or breakdown voltage ; Testing or monitoring effectiveness or level of insulation, e.g. of a cable or of an apparatus, for example using partial discharge measurements; Electrostatic testing
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Testing Relating To Insulation (AREA)
Abstract
The invention discloses a kind of extraction of local discharge superhigh frequency single waveform frequency domain character and recognition methods, belong to power equipment Partial Discharge Detecting Technology field, in this method, the method combined using fast Fourier and singular value decomposition, it is extracted the frequency domain character parameter of local discharge superhigh frequency single waveform, this feature vector represents discharge characteristic of the different electric discharge types in different frequency point, the feature vector extracted according to the present invention, shelf depreciation type can be effectively identified using the recognition methods for having supervision, by taking support vector machines as an example, accuracy of identification, which reaches, is greater than 96%.
Description
Technical field:
The invention belongs to power equipment Partial Discharge Detecting Technology fields, and in particular to a kind of local discharge superhigh frequency single
Waveform frequency domain character extracts and recognition methods, the identification for power equipment shelf depreciation type.
Background technique:
The insulating materials of power equipment is the significant components for guaranteeing power equipment and operating normally, but since insulating materials exists
Aging or insulating materials manufacturing deficiency under forceful electric power field action will appear part inside insulating materials in power equipment operation and put
Electricity, the development of shelf depreciation can accelerate the aging of insulating materials, so as to cause the power equipment lost of life, so must send out as early as possible
Now with the type of identification shelf depreciation, slow down the aging of power equipment using measure.
The advantages that local discharge superhigh frequency detection relies on high sensitivity, and live anti-jamming effectiveness is good, in power equipment part
It is had obtained relatively broad application in discharge examination.The bandwidth of local discharge superhigh frequency signal is contained from 0MHz~3000MHz
Discharge information abundant, analysis and feature extraction to hyperfrequency single signal, facilitates the identification of shelf depreciation electric discharge type.
Summary of the invention:
The problem of the purpose of the present invention is to solve shelf depreciation type identifications, provides a kind of local discharge superhigh frequency
Single waveform frequency domain character extracts and recognition methods, and demonstrates validity of the characteristic parameter in shelf depreciation type identification.
In order to achieve the above objectives, the present invention adopts the following technical scheme that realize:
A kind of local discharge superhigh frequency single waveform frequency domain character extracts and recognition methods, comprising the following steps:
1) the multiple groups local discharge superhigh frequency impulse discharge signal of different shelf depreciation types is acquired；
2) Fast Fourier Transform (FFT) is carried out to the hyperfrequency impulse discharge signal of the same electric discharge type of multiple groups, obtains multiple groups one
Power spectrum array FreArray is tieed up, multiple groups One-dimensional power spectrum array FreArray constitutes the spectral power matrix of same electric discharge type
FreMatrix；
3) singular value decomposition is carried out to the spectral power matrix FreMatrix of same electric discharge type, takes singular value maximum value pair
The one-dimensional vector answered obtains the feature vector of the spectral power matrix FreMatrix of same electric discharge type, i.e. this feature vector is corresponding
The feature vector of same electric discharge type hyperfrequency single waveform；
4) step 2) and operation 3) are repeated to different electric discharge types, obtains different electric discharge type hyperfrequency single waveforms
Feature vector；
5) feature vector extracted using step 4), using the identification for the local electric discharge type of learning method training for having supervision
Model, and electric discharge type is identified using identification model.
A further improvement of the present invention lies in that the concrete methods of realizing of step 2) is as follows:
201) Fast Fourier Transform (FFT) is used to hyperfrequency single waveform, obtains the One-dimensional power spectrum array of single waveform
FreArray；
202) 201) step is repeated to multiple hyperfrequency single waveforms of the same electric discharge type, by the multiple of FFT transform
FreArray obtains the spectral power matrix FreMatrix of same electric discharge type by row combination, wherein row indicates a single waveform
Power spectrum, column indicate sample number.
A further improvement of the present invention lies in that the concrete methods of realizing of step 5) is as follows:
501) FreMatrix for obtaining different electric discharge types, in the range of FreMatrix is normalized to [- 1,1], it
After be classified as test set and training set；
502) method for using supervised learning, obtains identification model using training set；
503) test set is taken in identification model, examines the effect of identification model, and according to accuracy of identification, adjust model
Parameter；
502) and 503) 504) step is repeated, until reaching the accuracy of identification that meets the requirements to get to corresponding shelf depreciation
Hyperfrequency single waveform identification model.
A further improvement of the present invention lies in that in step 501), in the range of FreMatrix is normalized to [- 1,1],
Formula is as follows:
Wherein, 1 ymax, ymin are that -1, x is characterized parameter, and xmin is characterized the minimum value of parameter, and xmax is characterized ginseng
The maximum value of amount.
A further improvement of the present invention lies in that carrying out identification mould according to training set using support vector machines in step 502)
The training of type obtains the identification model of SVM, and the kernel function of support vector machines is as follows:
K (x, y)=exp (- gamma × | x-y |2) (2)
Wherein, x is characterized parameter, and y is the corresponding value of electric discharge type, and gmmma is default value 1/ (characteristic value number).
A further improvement of the present invention lies in that the value of gmmma is 1/2000；Punishment parameter is default value 1.
The present invention, which compares prior art, has following innovative point:
1. being extracted hyperfrequency single waveform for the first time in such a way that Fast Fourier Transform (FFT) and singular value decomposition combine
Feature vector, can intuitively reflect the frequency domain character of different electric discharge types；
2, the frequecy characteristic extracted using the present invention can be reached in the case where not carrying out parameter optimization to identification model
To relatively high accuracy of identification.
The present invention, which compares prior art, has following remarkable advantage:
1, the frequecy characteristic parameter that extracts of the present invention, can include the frequency domain character of ultra-high frequency signal as far as possible, and energy
Enough characteristic quantity sizes intuitively shown on each Frequency point；
2, it can reach and compare in the case where not Statistical error model parameter using the characteristic parameter that the present invention extracts
High accuracy of identification, by taking support vector machines as an example, discrimination reaches 96% or more.Model Kernel Function is radial basis function, ginseng
Number c and g is all made of default value 1 and (1/ characteristic value number).
In conclusion the invention proposes a kind of new local discharge superhigh frequency single waveform frequency domain character extracting method,
The frequency domain character that different electric discharge types can be intuitively indicated using the characteristic parameter extracted in the present invention, using being mentioned in the present invention
Characteristic parameter out can effectively carry out shelf depreciation type identification.
Detailed description of the invention:
Fig. 1 is four kinds of shelf depreciation type ultra-high frequency signal single waveform frequency domain character vectors that the method for the present invention is extracted；
Wherein, Fig. 1 (a) is the ultra-high frequency signal single waveform frequency domain character vector of electric discharge type P1, and Fig. 1 (b) is electric discharge type P2's
Ultra-high frequency signal single waveform frequency domain character vector, Fig. 1 (c) are the ultra-high frequency signal single waveform frequency domain character of electric discharge type P3
Vector, Fig. 1 (d) are the ultra-high frequency signal single waveform frequency domain character vector of electric discharge type P4.
Specific embodiment:
The present invention is made further instructions below in conjunction with attached drawing.
The basic idea of the invention is that extracting superelevation based on the mode that Fast Fourier Transform (FFT) and singular value decomposition combine
The frequency domain character parameter of frequency single waveform carries out shelf depreciation type identification using the characteristic parameter of extraction, and detailed process is as follows:
1) four kinds of multiple offices of electric discharge type (being electric discharge type 1,2,3,4, hereinafter referred P1, P2, P3 and P4 respectively) are acquired
Discharge hyperfrequency single waveform in portion, its bandwidth of the oscillograph used is 100MHz~3GHz, sample rate 5GS/s；
2) Fast Fourier Transform (FFT) (FFT) is carried out to the hyperfrequency single waveform of local electric discharge type P1, obtains power spectrum
FreArray；
3) operation that step 2) is repeated to the multiple groups hyperfrequency single waveform of local electric discharge type P1, obtains multiple groups hyperfrequency
The FreArray of single waveform, as shown in Fig. 1 (a)；
4) is repeated by step 2) and 3), is obtained respectively for the multiple groups hyperfrequency single waveform of four kinds of electric discharge types P2, P3 and P4
Spectral power matrix FreMatrix, wherein row indicate a single waveform power spectrum, column indicate sample number, extremely such as Fig. 1 (b)
(d) shown in；
5) singular value decomposition is carried out to the FreMatrix of four kinds of electric discharge types, it is one group corresponding extracts singular value maximum value
Vector obtains the frequency domain character parameter of four kinds of electric discharge type hyperfrequency single waveforms, as shown in Figure 1；
6) shelf depreciation identification is carried out using obtained feature vector, the specific steps are as follows:
A) normalization characteristic parameter, in the range of characteristic quantity is normalized to [- 1,1], formula is as follows:
Wherein, 1 ymax, ymin are that -1, x is characterized parameter, and xmin is characterized the minimum value of parameter, and xmax is characterized ginseng
The maximum value of amount；
B) sample of four kinds of electric discharge types is randomly divided into training set and test set；
C) training for carrying out identification model according to training set using support vector machines (SVM), obtains the identification model of SVM,
The kernel function of support vector machines is as follows:
K (x, y)=exp (- gamma × | x-y |2) (2)
Wherein, x is characterized parameter, and y is that the corresponding value of electric discharge type (defines: corresponding 1, the P2 corresponding 2, P3 of P1 in the present invention
Corresponding 3, P4 is corresponding 4), and gmmma is default value 1/ (characteristic value number), is 1/2000 in the present invention；Punishment parameter is default value
1；
D) recognition effect inspection is carried out to test set, obtained accuracy of identification is 96% or more.
Claims (2)
1. a kind of local discharge superhigh frequency single waveform frequency domain character extracts and recognition methods, which is characterized in that including following step
It is rapid:
1) the multiple groups local discharge superhigh frequency impulse discharge signal of different shelf depreciation types is acquired；
2) Fast Fourier Transform (FFT) is carried out to the hyperfrequency impulse discharge signal of the same electric discharge type of multiple groups, obtains the one-dimensional function of multiple groups
Rate composes array FreArray, and multiple groups One-dimensional power spectrum array FreArray constitutes the spectral power matrix of same electric discharge type
FreMatrix；Concrete methods of realizing is as follows:
201) Fast Fourier Transform (FFT) is used to hyperfrequency single waveform, obtains the One-dimensional power spectrum array of single waveform
FreArray；
202) 201) step is repeated to multiple hyperfrequency single waveforms of the same electric discharge type, by the multiple of FFT transform
FreArray obtains the spectral power matrix FreMatrix of same electric discharge type by row combination, wherein row indicates a single waveform
Power spectrum, column indicate sample number；
3) singular value decomposition is carried out to the spectral power matrix FreMatrix of same electric discharge type, takes singular value maximum value corresponding
One-dimensional vector obtains the feature vector of the spectral power matrix FreMatrix of same electric discharge type, i.e. this feature vector corresponds to same
The feature vector of electric discharge type hyperfrequency single waveform；
4) step 2) and operation 3) are repeated to different electric discharge types, obtains the feature of different electric discharge type hyperfrequency single waveforms
Vector；
5) feature vector extracted using step 4), using the identification mould for the local electric discharge type of learning method training for having supervision
Type, and electric discharge type is identified using identification model, concrete methods of realizing is as follows:
501) FreMatrix for obtaining different electric discharge types in the range of FreMatrix is normalized to [- 1,1], later will
It is divided into test set and training set；Formula is as follows:
Wherein, 1 ymax, ymin are that -1, x is characterized parameter, and xmin is characterized the minimum value of parameter, and xmax is characterized parameter
Maximum value；
502) method for using supervised learning, obtains identification model using training set；The specific method is as follows:
The training for carrying out identification model according to training set using support vector machines, obtains the identification model of SVM, support vector machines
Kernel function is as follows:
K (x, y)=exp (- gamma × | x-y |2) (2)
Wherein, x is characterized parameter, and y is the corresponding value of electric discharge type, and gmmma is default value 1/ (characteristic value number)；
503) test set is taken in identification model, examines the effect of identification model, and according to accuracy of identification, adjust model ginseng
Number；
502) and 503) 504) step is repeated, until reaching the accuracy of identification that meets the requirements to get to corresponding shelf depreciation superelevation
Frequency single waveform identification model.
2. local discharge superhigh frequency single waveform frequency domain character according to claim 1 extracts and recognition methods, feature
It is, the value of gmmma is 1/2000；Punishment parameter is default value 1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610601321.6A CN106291275B (en) | 2016-07-27 | 2016-07-27 | A kind of local discharge superhigh frequency single waveform frequency domain character extracts and recognition methods |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610601321.6A CN106291275B (en) | 2016-07-27 | 2016-07-27 | A kind of local discharge superhigh frequency single waveform frequency domain character extracts and recognition methods |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106291275A CN106291275A (en) | 2017-01-04 |
| CN106291275B true CN106291275B (en) | 2019-03-22 |
Family
ID=57662562
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610601321.6A Active CN106291275B (en) | 2016-07-27 | 2016-07-27 | A kind of local discharge superhigh frequency single waveform frequency domain character extracts and recognition methods |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106291275B (en) |
Families Citing this family (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107144769A (en) * | 2017-04-17 | 2017-09-08 | 西安热工研究院有限公司 | The three-dimensional clustering recognition method of shelf depreciation for amplitude sum of being discharged based on different frequency range |
| CN107238782A (en) * | 2017-05-10 | 2017-10-10 | 西安热工研究院有限公司 | A kind of a variety of shelf depreciation mixed signal separation methods of feature based phase |
| JP6602512B2 (en) * | 2017-06-14 | 2019-11-06 | 三菱電機株式会社 | Aging deterioration diagnosis apparatus and aging deterioration diagnosis method |
| CN107991590A (en) * | 2017-11-28 | 2018-05-04 | 广东电网有限责任公司珠海供电局 | A kind of cable local discharge signal characteristic vector extracting method based on frequency domain power spectrum |
| CN109596955B (en) * | 2018-12-30 | 2021-06-22 | 国网北京市电力公司 | Partial discharge state determination method and device |
| CN110531234B (en) * | 2019-09-26 | 2021-06-04 | 武汉三相电力科技有限公司 | Method for identifying and extracting discharge pulse of power transmission line |
| EP3961227A1 (en) * | 2020-08-31 | 2022-03-02 | General Electric Company | Online and offline partial discharge detection for electrical drive systems |
| CN115542101A (en) * | 2022-11-30 | 2022-12-30 | 杭州兆华电子股份有限公司 | Voiceprint preprocessing method of transformer voiceprint detection system |
| CN115602191A (en) * | 2022-12-12 | 2023-01-13 | 杭州兆华电子股份有限公司（Cn） | Noise elimination method of transformer voiceprint detection system |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102645620A (en) * | 2012-05-17 | 2012-08-22 | 广东电网公司电力科学研究院 | Multisource partial discharge detection method and device of transformer substation based on time-frequency characteristic parameters |
| CN103077402A (en) * | 2012-12-28 | 2013-05-01 | 湖北省电力公司电力科学研究院 | Transformer partial-discharging mode recognition method based on singular value decomposition algorithm |
| CN104020398A (en) * | 2014-06-03 | 2014-09-03 | 华北电力大学 | Method for extracting partial discharge waveform features of converter transformer |
| CN105425076A (en) * | 2015-12-11 | 2016-03-23 | 厦门理工学院 | Method of carrying out transformer fault identification based on BP neural network algorithm |
| CN105606966A (en) * | 2015-12-21 | 2016-05-25 | 安徽理工大学 | Partial discharge pattern recognition method based on mixed neural network algorithm |
-
2016
- 2016-07-27 CN CN201610601321.6A patent/CN106291275B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102645620A (en) * | 2012-05-17 | 2012-08-22 | 广东电网公司电力科学研究院 | Multisource partial discharge detection method and device of transformer substation based on time-frequency characteristic parameters |
| CN103077402A (en) * | 2012-12-28 | 2013-05-01 | 湖北省电力公司电力科学研究院 | Transformer partial-discharging mode recognition method based on singular value decomposition algorithm |
| CN104020398A (en) * | 2014-06-03 | 2014-09-03 | 华北电力大学 | Method for extracting partial discharge waveform features of converter transformer |
| CN105425076A (en) * | 2015-12-11 | 2016-03-23 | 厦门理工学院 | Method of carrying out transformer fault identification based on BP neural network algorithm |
| CN105606966A (en) * | 2015-12-21 | 2016-05-25 | 安徽理工大学 | Partial discharge pattern recognition method based on mixed neural network algorithm |
Non-Patent Citations (3)
| Title |
|---|
| GIS典型缺陷的局部放电超高频检测及模式识别;韩磊等;《内蒙古电力技术》;20151231;第33卷(第1期);第11页 |
| 用谐波小波包变换法提取 GIS 局部放电信号;唐炬等;《电工技术学报》;20150228;第30卷(第3期);全文 |
| 电力变压器局部放电信号的特征提取与模式;尚海昆;《华北电力大学博士学位论文》;20141231;第4章 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106291275A (en) | 2017-01-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106291275B (en) | A kind of local discharge superhigh frequency single waveform frequency domain character extracts and recognition methods | |
| Song et al. | GIS partial discharge pattern recognition via deep convolutional neural network under complex data source | |
| CN103197218B (en) | A kind of high-voltage cable insulation defect partial discharge electrification detection diagnostic method | |
| CN104849633A (en) | Switchgear partial discharge mode recognition method | |
| CN104198898B (en) | Local discharge development process diagnosis method based on pulse-train analysis | |
| CN103645425B (en) | High-voltage cable insulation defect partial discharge on-line monitoring diagnosis method | |
| CN103076547B (en) | Method for identifying GIS (Gas Insulated Switchgear) local discharge fault type mode based on support vector machines | |
| CN109116203A (en) | Power equipment partial discharges fault diagnostic method based on convolutional neural networks | |
| CN105548832A (en) | High-voltage power cable fault recognition method | |
| CN102841297B (en) | Self-adaptive partial-discharge monitoring method and device for sampling frequency | |
| CN105938177A (en) | Feature extraction and identification method based on partial discharge statistical amount | |
| CN104330711A (en) | Saturated reactor insulation accelerated aging test device and test method | |
| CN110703078A (en) | GIS fault diagnosis method based on spectral energy analysis and self-organizing competition algorithm | |
| Purkait et al. | Impulse fault classification in transformers by fractal analysis | |
| CN104125050A (en) | Ultrahigh-frequency RFID (radio frequency identification) reader protocol conformance testing method | |
| CN112881879A (en) | High-voltage cable terminal partial discharge mode identification method, device and equipment | |
| CN107578016B (en) | Sparse representation-based automatic residual current waveform identification method | |
| CN101930049B (en) | On-line local discharging monitoring method of ultrahigh voltage cable | |
| CN104569756A (en) | Switch cabinet partial discharge on-line monitoring system based on electronic nose sensor | |
| EP4001936A1 (en) | Automatic partial discharge and noise signals separation using arithmetic coding in time domain and magnitude distributions in frequency domain | |
| Guan et al. | Partial discharge pattern recognition of transformer based on deep forest algorithm | |
| CN109917245B (en) | Ultrasonic detection partial discharge signal mode identification method considering phase difference | |
| CN110703080B (en) | GIS spike discharge diagnosis method, discharge degree identification method and device | |
| Duan et al. | Research on Identification of Magnetizing Inrush Current Based on PSO-SVM | |
| Yaoheng et al. | Severity diagnosis and assessment on defects in GIS based on partial discharge detection |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |