CN109655711A - Power distribution network internal overvoltage kind identification method - Google Patents
Power distribution network internal overvoltage kind identification method Download PDFInfo
- Publication number
- CN109655711A CN109655711A CN201910022876.9A CN201910022876A CN109655711A CN 109655711 A CN109655711 A CN 109655711A CN 201910022876 A CN201910022876 A CN 201910022876A CN 109655711 A CN109655711 A CN 109655711A
- Authority
- CN
- China
- Prior art keywords
- phase
- overvoltage
- power distribution
- distribution network
- singular value
- 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.)
- Pending
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
- G01R31/08—Locating faults in cables, transmission lines, or networks
- G01R31/081—Locating faults in cables, transmission lines, or networks according to type of conductors
- G01R31/086—Locating faults in cables, transmission lines, or networks according to type of conductors in power transmission or distribution networks, i.e. with interconnected conductors
-
- 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
- G01R31/08—Locating faults in cables, transmission lines, or networks
- G01R31/088—Aspects of digital computing
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Theoretical Computer Science (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention discloses a kind of power distribution network internal overvoltage kind identification methods, include the following steps: step S1: when detecting that residual voltage is mutated, acquisition is mutated A, B, C three-phase voltage signal of totally six cycles of five cycles after previous cycle and mutation；Step S2: dual-tree complex wavelet transform is carried out to step S1 A, B, C three-phase voltage sampled data obtained respectively and obtains time-frequency matrix；Step S3: dimension-reduction treatment is carried out to the corresponding time-frequency matrix of A, B, C three-phase voltage respectively using singular value decomposition method, obtain the singular value features amount of every phase, and the singular value features amount of three-phase signal is end to end by the sequence of A phase, B phase, C phase, obtain one-dimensional characteristic vector；Step S4: in the deepness belief network that the resulting one-dimensional characteristic vector input of step S3 was trained, recognition result is obtained.It has the following advantages: overvoltage type identification accuracy with higher, adaptability are stronger.
Description
Technical field
The present invention relates to a kind of power distribution network internal overvoltage kind identification methods.
Background technique
Low and medium voltage distribution network is often referred to the electric power networks of 35kV and following voltage class, has a very wide distribution, and structure is complicated,
It is closely connected with user.According to statistics, about 70% overvoltage betides power distribution network in electric system, the overvoltage thus caused
It is to influence the principal element of power distribution network safe and stable operation with overcurrent.Because insulation damages caused by overvoltage often cause it is all kinds of
Short trouble causes to seriously affect to the normal operation of electric system.Therefore, how research extracts internal overvoltage signal characteristic
Identification classification is measured and carried out, there is practical application value to the safe operation for guaranteeing power grid.
Summary of the invention
The present invention provides a kind of power distribution network internal overvoltage kind identification methods, and which overcome described in background technology
The deficiencies in the prior art.
The technical solution adopted by the present invention to solve the technical problems is:
Power distribution network internal overvoltage kind identification method, it includes the following steps:
Step S1: when detecting that residual voltage is mutated, acquisition is mutated five cycles after previous cycle and mutation
A, B, C three-phase voltage signal of totally six cycles；
Step S2: dual-tree complex wavelet transform acquisition is carried out respectively to step S1 A, B, C three-phase voltage sampled data obtained
Time-frequency matrix；
Step S3: carrying out dimension-reduction treatment to the corresponding time-frequency matrix of A, B, C three-phase voltage respectively using singular value decomposition method,
The singular value features amount of every phase is obtained, and the singular value features amount of three-phase signal is pressed to the sequence head and the tail phase of A phase, B phase, C phase
It connects, obtains one-dimensional characteristic vector；
Step S4: it in the deepness belief network that the resulting one-dimensional characteristic vector input of step S3 was trained, is identified
As a result.
Among one embodiment: the residual voltage abrupt climatic change process in the step S1 is as follows:
Db4 wavelet decomposition is carried out to residual voltage signal, obtained d3 detail coefficients will be decomposed and carry out single branch reconstruct, work as list
The modulus maximum of certain sampled point is equal to greatly setting value in branch reconstruction signal, is determined as that voltage signal is mutated.
Among one embodiment: the process for obtaining time-frequency matrix using dual-tree complex wavelet transform in the step S2 is as follows:
Every phase voltage signal is subjected to dual-tree complex wavelet decomposition, detail coefficients and similarity factor are obtained, respectively to each system
Number carries out single branch reconstruct, and forms time-frequency matrix.
Among one embodiment: carrying out the process of dimensionality reduction such as to time-frequency matrix using singular value decomposition method in the step S3
Under:
The corresponding three time-frequency matrixes of A, B, C tri- resulting to step S2 carry out singular value decomposition respectively, obtain matrix
Expression formula is A=U Λ VT, A is the matrix of m × n, U in formulam×m、Vn×nFor two orthogonal matrixes, Λm×nFor diagonal matrix, by Λm×n
Element on diagonal line switchs to one-dimensional row vector；Three time-frequency matrixes resulting to S2 carry out singular value decomposition and obtain three respectively
One-dimensional singular value features amount, and singular value features amount is end to end by the sequence of A phase, B phase, C phase, obtain one-dimensional characteristic
Vector, as characteristic quantity to be identified.
Among one embodiment: carrying out the process packet of Classification and Identification in the step S4 to overvoltage using deepness belief network
Include following steps:
Step S41: with the output desired value of binary number setting sample；
Step S42: using maximin method as method for normalizing, all data are converted between [- 1,1]
Number, expression formula areIn formula, xiThe amount of being characterized, xmin、xmaxRespectively to the maximum in normalization data
Value and minimum value；
Step S43: the optimal parameter of deepness belief network is determined using single argument optimization method: deepness belief network is implicit
The number of plies is 3 layers, neuron number 60, and being limited Boltzmann machine learning rate is 0.04, hidden layer and output layer biasing initialization
It is 0, the biasing log (p of visual elementi/(1-pi)) indicate, piIndicate that ith feature is active institute in training sample
The ratio accounted for；
Step S44: the maximum frequency of training that classifier is arranged is 2000 times, and minimal gradient is set as 10-7, will be special obtained by S3
It levies vector to input in trained deepness belief network, obtains classification results.
Among one embodiment: the overvoltage type includes Subharmonic Resonance, single phase metal ground connection, fundamental resonance, excision
Capacitor group cuts off nonloaded line, intermittent arc grounding, high-frequency resonant totally seven class overvoltage type.
Among one embodiment: the output desired value of the sample is set to: Subharmonic Resonance: 0000001；Single phase metal
Ground connection: 0000010；Fundamental resonance: 0000100；Excision capacitor group: 0001000；Excision nonloaded line: 0010000；Intermittently
Property arc grounding: 0100000；High-frequency resonant: 1000000.
Among one embodiment: the setting value is 0.1.
The technical program compared with the background art, it has the following advantages:
1, the present invention has preferable anti-frequency mixed using the time-frequency matrix of dual-tree complex wavelet transform construction overvoltage signal
Folded characteristic, can imperfectly describe time-frequency feature of the overvoltage signal in each frequency band, contain characterization signal substantive characteristics
Time-Frequency Information.
2, the present invention is subtracted using singular value decomposition to the resulting further dimension-reduction treatment of time-frequency matrix of dual-tree complex wavelet transform
Few characteristic quantity dimension, effectively reduces operand, improves the speed of recognizer.
3, the deepness belief network used in the present invention has stronger learning ability, and more stable structure can be accurately
Identification invention can accurately identify single phase metal ground connection, high-frequency resonant, fundamental resonance, Subharmonic Resonance, intermittent arc light
Ground connection, excision nonloaded line and excision seven class power distribution network internal overvoltage type of capacitor group.
4, power distribution network internal overvoltage kind identification method of the invention is still with higher under the operating condition of noise jamming
Overvoltage type identification accuracy, adaptability are stronger.
Detailed description of the invention
Present invention will be further explained below with reference to the attached drawings and examples.
Fig. 1 is power distribution network internal overvoltage type identification flow chart described in the present embodiment.
Fig. 2 is the 10kV electricity distribution network model figure built using ATP/EMTP simulation software in a preferred application example.
Specific embodiment
The present embodiment provides a kind of power distribution network internal overvoltage kind identification methods, as shown in Figure 1, including the following steps:
Step S1: when detecting that residual voltage is mutated, acquisition is mutated five cycles after previous cycle and mutation
A, B, C three-phase voltage signal of totally six cycles；This step specifically comprises the following steps:
Db4 wavelet decomposition is carried out to residual voltage signal, obtained d3 detail coefficients will be decomposed and carry out single branch reconstruct, work as list
The modulus maximum of certain sampled point is equal to greatly setting value in branch reconstruction signal, and the setting value is set as 0.1 in the present embodiment, is determined as
Voltage signal is mutated, system acquire at once the previous cycle of catastrophe point and mutation after five cycles A, B of totally six cycles,
C three-phase voltage signal.
Step S2: dual-tree complex wavelet transform acquisition is carried out respectively to step S1 A, B, C three-phase voltage sampled data obtained
Time-frequency matrix.This step specifically comprises the following steps:
By taking A phase voltage as an example, A phase voltage signal is subjected to dual-tree complex wavelet decomposition, obtains detail coefficients and similarity factor,
Single branch reconstruct is carried out to each coefficient respectively, and forms time-frequency matrix.Above-mentioned operation is carried out to A, B, C three-phase voltage respectively, is obtained
To three time-frequency matrixes.
Step S3: carrying out dimension-reduction treatment to the corresponding time-frequency matrix of A, B, C three-phase voltage respectively using singular value decomposition method,
The singular value features amount of every phase is obtained, and the singular value features amount of three-phase signal is pressed to the sequence head and the tail phase of A phase, B phase, C phase
It connects, obtains one-dimensional characteristic vector.This step specifically comprises the following steps:
Three time-frequency matrixes resulting to S2 carry out singular value decomposition respectively, and obtaining matrix expression is A=U Λ VT, formula
Middle A is the matrix of m × n, Um×m、Vn×nFor two orthogonal matrixes, Λm×nFor diagonal matrix, by Λm×nElement on diagonal line switchs to
One-dimensional row vector.Three time-frequency matrixes resulting to S2 carry out singular value decomposition and obtain three one-dimensional singular value features respectively
Amount, and singular value features amount is end to end by the sequence of A phase, B phase, C phase, one-dimensional characteristic vector is obtained, as to be identified
Characteristic quantity.
Step S4: in the deepness belief network that the resulting one-dimensional characteristic vector input of S3 was trained, identification is directly obtained
As a result.This step specifically comprises the following steps:
Step S41: the difference in order to calculate reality output and desired output it is expected with the output of binary system setting sample:
Subharmonic Resonance: 0000001；Single phase metal ground connection: 0000010；Fundamental resonance: 0000100；Cut off capacitor group:
0001000；Excision nonloaded line: 0010000；Intermittent arc grounding: 0100000；High-frequency resonant: 1000000.
Step S42: in order to cancel order of magnitude difference between data, using maximin method as method for normalizing, institute
There are data to be converted into the number between [- 1,1], expression formula isIn formula, xiThe amount of being characterized, xmin、xmax
Respectively to the maximum value and minimum value in normalization data.
Step S43: the optimal parameter of deepness belief network is determined using single argument optimization method: deepness belief network is implicit
The number of plies is 3 layers, neuron number 60, and being limited Boltzmann machine learning rate is 0.04, hidden layer and output layer biasing initialization
It is 0, the biasing log (p of visual elementi/(1-pi)) indicate, piIndicate that ith feature is active institute in training sample
The ratio accounted for.
Step S44: the maximum frequency of training that classifier is arranged is 2000 times, and minimal gradient is set as 10-7, will be special obtained by S3
It levies vector to input in trained deepness belief network, immediately arrives at classification results.
In one preferred application example, it is used for as shown in Fig. 2, building 10kV electricity distribution network model using ATP/EMTP simulation software
Overvoltage data are obtained, test result shows the distribution that different fault points, initial phase angle, ground resistance occur for this method
Net temporary overvoltage can be identified quick and precisely, and the well adapting to property under noise jamming, carry out seven kinds on this basis
The simulated experiment of overvoltage type, and acquire A, B, C three-phase voltage waveform.In simulation model, 110kV high-tension line three-phase electricity
Source replaces, and emulation element specifically includes that voltage transformer, system power supply, transformer, route etc..110kV/10kV transformer connects
Group is connect as Ynd11, the resistance per unit value of primary side and secondary side is 0.0019, and inductance per unit value is 0.75, field core electricity
Hindering per unit value is 1615.12, and field core inductance per unit value is 833.23；It is Dyn11 that 10kV/0.4kV transformer, which connects group,
The resistance per unit value of primary side and secondary side is 0.00501, and inductance per unit value is 0.0223, and field core resistance per unit value is
869.27, field core inductance per unit value is 142.35；The excitation parameter of Electromagnetic PT are as follows: voltage per unit value be 1,1.328,
1.501,1.79,1.963, corresponding electric current per unit value is 1,1.733,3.067,7.33,11.93；10kV line module is selected
Three-phase π type equivalent circuit module, 0.17 Ω of positive sequence resistance/km of overhead transmission line, 0.0097 μ F/km of positive sequence capacitor, positive sequence inductance
1.21mH/km, 0.23 Ω of zero sequence resistance/km, zero sequence capacitor 0.008 μ F/km, zero sequence inductance 5.478mH/km；Cable run
0.27 Ω of positive sequence resistance/km, 0.339 μ F/km of positive sequence capacitor, positive sequence inductance 0.255mH/km, 2.7 Ω of zero sequence resistance/km, zero sequence
Capacitor 0.28 μ F/km, zero sequence inductance 1.019mH/km.
The adaptability of recognition methods is proposed come inspection institute by following test result:
When table 1 is noiseless, sample data carries out the effect of fault waveform identification by this case the method.In emulation sample
The white Gaussian noise of addition -10dB in notebook data, then carry out overvoltage classification identification, shadow of the verifying noise to this paper recognition methods
It rings, recognition result is as shown in table 2, and recognition accuracy is up to 97.33%, illustrates that mentioned method has good noiseproof feature.
Fault waveform recognition effect when 1 noiseless of table
Fault waveform recognition effect under 2 noise jamming of table
The above is only the preferred embodiment of the present invention, the range implemented of the present invention that therefore, it cannot be limited according to, i.e., according to
Equivalent changes and modifications made by the invention patent range and description, should still be within the scope of the present invention.
Claims (8)
1. power distribution network internal overvoltage kind identification method, characterized by the following steps:
Step S1: when detecting that residual voltage is mutated, acquisition is mutated after previous cycle and mutation five cycles totally six
A, B, C three-phase voltage signal of cycle；
Step S2: dual-tree complex wavelet transform is carried out to step S1 A, B, C three-phase voltage sampled data obtained respectively and obtains time-frequency
Matrix；
Step S3: dimension-reduction treatment is carried out to the corresponding time-frequency matrix of A, B, C three-phase voltage respectively using singular value decomposition method, is obtained
The singular value features amount of every phase, and the singular value features amount of three-phase signal is end to end by the sequence of A phase, B phase, C phase, it obtains
One-dimensional characteristic vector；
Step S4: in the deepness belief network that the resulting one-dimensional characteristic vector input of step S3 was trained, recognition result is obtained.
2. power distribution network internal overvoltage kind identification method according to claim 1, it is characterised in that: in the step S1
Residual voltage abrupt climatic change process it is as follows:
Db4 wavelet decomposition is carried out to residual voltage signal, obtained d3 detail coefficients will be decomposed and carry out single branch reconstruct, work as Dan Zhichong
The modulus maximum of certain sampled point is equal to greatly setting value in structure signal, is determined as that voltage signal is mutated.
3. power distribution network internal overvoltage kind identification method according to claim 1, it is characterised in that: in the step S2
The process for obtaining time-frequency matrix using dual-tree complex wavelet transform it is as follows:
By every phase voltage signal carry out dual-tree complex wavelet decomposition, obtain detail coefficients and similarity factor, respectively to each coefficient into
Row list branch reconstruct, and form time-frequency matrix.
4. power distribution network internal overvoltage kind identification method according to claim 1, it is characterised in that: in the step S3
The process for carrying out dimensionality reduction to time-frequency matrix using singular value decomposition method is as follows:
The corresponding three time-frequency matrixes of A, B, C tri- resulting to step S2 carry out singular value decomposition respectively, obtain expression matrix
Formula is A=U Λ VT, A is the matrix of m × n, U in formulam×m、Vn×nFor two orthogonal matrixes, Λm×nFor diagonal matrix, by Λm×nDiagonally
Element on line switchs to one-dimensional row vector；Respectively resulting to S2 three time-frequency matrixes carry out singular value decomposition obtain three it is one-dimensional
Singular value features amount, and by singular value features amount by A phase, B phase, C phase sequence it is end to end, obtain one-dimensional characteristic vector,
As characteristic quantity to be identified.
5. power distribution network internal overvoltage kind identification method according to claim 1, it is characterised in that: in the step S4
Included the following steps: using the process that deepness belief network carries out Classification and Identification to overvoltage
Step S41: with the output desired value of binary number setting sample；
Step S42: using maximin method as method for normalizing, all data are converted into the number between [- 1,1], table
It is up to formulaIn formula, xiThe amount of being characterized, xmin、xmaxRespectively to the maximum value in normalization data and most
Small value；
Step S43: the optimal parameter of deepness belief network is determined using single argument optimization method: deepness belief network implies the number of plies
It is 3 layers, neuron number 60, being limited Boltzmann machine learning rate is 0.04, and hidden layer and output layer biasing are initialized as 0,
The biasing of visual element log (pi/(1-pi)) indicate, piIt is shared to indicate that ith feature in training sample is active
Ratio；
Step S44: the maximum frequency of training that classifier is arranged is 2000 times, and minimal gradient is set as 10-7, by feature obtained by S3 to
Amount inputs in trained deepness belief network, obtains classification results.
6. power distribution network internal overvoltage kind identification method according to claim 5, it is characterised in that: the overvoltage class
Type includes that Subharmonic Resonance, single phase metal ground connection, fundamental resonance, excision capacitor group, excision nonloaded line, intermittent arc light connect
Ground, high-frequency resonant totally seven class overvoltage type.
7. power distribution network internal overvoltage kind identification method according to claim 6, it is characterised in that: the sample it is defeated
Desired value is set to Subharmonic Resonance out: 0000001；Single phase metal ground connection: 0000010；Fundamental resonance: 0000100；Excision
Capacitor group: 0001000；Excision nonloaded line: 0010000；Intermittent arc grounding: 0100000；High-frequency resonant:
1000000。
8. power distribution network internal overvoltage kind identification method according to claim 2, it is characterised in that: the setting value is
0.1。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910022876.9A CN109655711A (en) | 2019-01-10 | 2019-01-10 | Power distribution network internal overvoltage kind identification method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910022876.9A CN109655711A (en) | 2019-01-10 | 2019-01-10 | Power distribution network internal overvoltage kind identification method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109655711A true CN109655711A (en) | 2019-04-19 |
Family
ID=66119328
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910022876.9A Pending CN109655711A (en) | 2019-01-10 | 2019-01-10 | Power distribution network internal overvoltage kind identification method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109655711A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111123036A (en) * | 2019-12-27 | 2020-05-08 | 深圳供电局有限公司 | System and method for identifying overvoltage faults |
| CN113125992A (en) * | 2021-04-23 | 2021-07-16 | 合肥工业大学 | NPC three-level inverter fault diagnosis method and system based on DBN |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103197124A (en) * | 2013-03-14 | 2013-07-10 | 重庆市电力公司电力科学研究院 | Overvoltage identification method based on time-frequency matrix singular value |
| CN106680576A (en) * | 2017-01-11 | 2017-05-17 | 福州大学 | Power distribution network internal overvoltage identification method based on partitioned time-frequency spectrum and deep learning algorithm |
| CN106709820A (en) * | 2017-01-11 | 2017-05-24 | 中国南方电网有限责任公司电网技术研究中心 | Power system load prediction method and device based on deep belief network |
| CN108089099A (en) * | 2017-12-18 | 2018-05-29 | 广东电网有限责任公司佛山供电局 | The diagnostic method of distribution network failure based on depth confidence network |
| CN109142851A (en) * | 2018-07-26 | 2019-01-04 | 福州大学 | A kind of novel power distribution network internal overvoltage recognition methods |
-
2019
- 2019-01-10 CN CN201910022876.9A patent/CN109655711A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103197124A (en) * | 2013-03-14 | 2013-07-10 | 重庆市电力公司电力科学研究院 | Overvoltage identification method based on time-frequency matrix singular value |
| CN106680576A (en) * | 2017-01-11 | 2017-05-17 | 福州大学 | Power distribution network internal overvoltage identification method based on partitioned time-frequency spectrum and deep learning algorithm |
| CN106709820A (en) * | 2017-01-11 | 2017-05-24 | 中国南方电网有限责任公司电网技术研究中心 | Power system load prediction method and device based on deep belief network |
| CN108089099A (en) * | 2017-12-18 | 2018-05-29 | 广东电网有限责任公司佛山供电局 | The diagnostic method of distribution network failure based on depth confidence network |
| CN109142851A (en) * | 2018-07-26 | 2019-01-04 | 福州大学 | A kind of novel power distribution network internal overvoltage recognition methods |
Non-Patent Citations (3)
| Title |
|---|
| 李欣等: "110kV变电站过电压智能识别系统应用研究", 《电磁避雷器》 * |
| 杨勇等: "采用时频矩阵奇异值分解和多级支持向量机的雷电及操作过电压识别", 《电网技术》 * |
| 高伟 等: "基于奇异值分解和多级支持向量机的配电网故障类型识别", 《电子测量与仪器学报》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111123036A (en) * | 2019-12-27 | 2020-05-08 | 深圳供电局有限公司 | System and method for identifying overvoltage faults |
| CN113125992A (en) * | 2021-04-23 | 2021-07-16 | 合肥工业大学 | NPC three-level inverter fault diagnosis method and system based on DBN |
| CN113125992B (en) * | 2021-04-23 | 2022-07-19 | 合肥工业大学 | NPC three-level inverter fault diagnosis method and system based on DBN |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Duan et al. | Identification of partial discharge defects based on deep learning method | |
| CN109307824B (en) | Clustering-based power distribution network single-phase earth fault section positioning method | |
| Sedighi et al. | Simulation of high impedance ground fault in electrical power distribution systems | |
| CN109142851B (en) | Novel power distribution network internal overvoltage identification method | |
| Atthapol et al. | Internal Fault Classification in Transformer Windings using Combination of Discrete Wavelet-Transforms and Back-propagation Neural Networks | |
| CN102570392A (en) | Method for identifying exciting inrush current of transformer based on improved probability neural network | |
| CN110543921A (en) | cable early fault identification method based on waveform learning | |
| CN109655711A (en) | Power distribution network internal overvoltage kind identification method | |
| Gowrishankar et al. | Transmission line fault detection and classification using discrete wavelet transform and artificial neural network | |
| Li et al. | Ferromagnetic resonance over-voltage identification method based on Gram angle field | |
| Hong et al. | Compensation of distorted secondary current caused by saturation and remanence in a current transformer | |
| Dey et al. | Rough-granular approach for impulse fault classification of transformers using cross-wavelet transform | |
| Rajamani et al. | Identification of simultaneously occurring dynamic disc-to-disc insulation failures in transformer winding under impulse excitation | |
| CN116466182A (en) | Line fault distance measurement method for power system | |
| CN107578016A (en) | A kind of residual current waveform automatic identifying method based on rarefaction representation | |
| Sirojan et al. | High Impedance Fault Detection by Convolutional Deep Neural Network | |
| CN114755529A (en) | Multi-feature fusion single-phase earth fault type identification method based on deep learning | |
| bin Sulaiman et al. | Using probabilistic neural network for classification high impedance faults on power distribution feeders | |
| Zhu et al. | Extraction of partial discharge signal feature based on dual-tree complex wavelet transform and singular-value decomposition | |
| Jiao et al. | A new approach to identify inrush current based on generalized S-transform | |
| Yang et al. | Detection of downed conductor in distribution system | |
| Zhang et al. | A structure for automatically extracting and identifying internal overvoltage measured in distribution networks based on FSWT-SSAE | |
| Mahmood et al. | Application of wavelet multi-resolution analysis & perceptron neural networks for classification of transients on transmission line | |
| Yang et al. | Detection of high impedance faults in distribution system | |
| CN113985733B (en) | Power distribution network fault identification method based on self-adaptive probability learning |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190419 |