CN110991240B - Single-phase fault identification and classification method for low-voltage neutral point ungrounded system - Google Patents
Single-phase fault identification and classification method for low-voltage neutral point ungrounded system Download PDFInfo
- Publication number
- CN110991240B CN110991240B CN201911052914.1A CN201911052914A CN110991240B CN 110991240 B CN110991240 B CN 110991240B CN 201911052914 A CN201911052914 A CN 201911052914A CN 110991240 B CN110991240 B CN 110991240B
- Authority
- CN
- China
- Prior art keywords
- fault
- phase
- frequency component
- low
- neutral point
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
- Y04S10/52—Outage or fault management, e.g. fault detection or location
Abstract
The invention discloses a single-phase fault identification and classification method of a low-voltage neutral point ungrounded system, which comprises the following steps: classifying single-phase faults of the neutral point ungrounded system; coding the fault type; recording phase current and phase voltage data of a certain phase during faults; performing wavelet decomposition on the data to obtain two vectors; carrying out wavelet reconstruction on the two vectors to obtain a high-frequency component and a low-frequency component; the high-frequency component and the low-frequency component are used as recognition features of electric shock signals, and data feature quantity is regulated; randomly classifying the obtained feature quantity in combination with the corresponding label, and generating a classification model to obtain a fault identification and classification model; and obtaining a fault discrimination result based on the fault identification and classification model. The invention performs feature extraction and classification model establishment by means of the obtained phase voltage and phase current data, and finally realizes single-phase fault identification and classification in a neutral point ungrounded system, thereby reducing additional hardware measuring devices.
Description
Technical Field
The invention relates to a single-phase fault identification and classification method for a low-voltage neutral point ungrounded system.
Background
At present, with the rapid development of society and the increasing popularization of household appliances, electric energy is the most widely used energy source in the contemporary society, and great convenience is brought to people's life. With the wide application of electric energy, the opportunities for people to contact electrical equipment are increased, and the probability of electric shock accidents is also increased. The electric shock accident caused by improper use of electrical equipment or line leakage occurs, and the life safety of people is greatly threatened.
Currently, methods for solving such accidents are widely used: residual current protection devices are used in distribution areas and are divided into two main types, namely voltage operation type and current operation type according to different working principles. The voltage-operated residual current protector has the defects of complex structure, poor stability of the operation characteristics under the external interference, high manufacturing cost and the like, and is gradually replaced by the current-operated residual current protector.
With the rapid development of power electronics and computer technology, several types of residual current protection devices such as pulse action type, amplitude and phase discrimination action type, current separation action type and the like are developed at home and abroad. The current action type residual current protection device is usually referred to as a zero sequence current transformer residual current protection device, and the device has a large action dead zone. In practical application, phenomena such as refusal operation and misoperation are easy to occur. The current pulse action type and current amplitude and phase discrimination action type residual current device has high circuit integration level, high required cost and false operation or refusal operation of the protection device. The current separation action type residual current device can completely eliminate the problems of the rejection and the misoperation of the general residual current protection device in theory, but the algorithm is still immature, the hardware requirement is higher, and the current residual current protection device is not put into practical application at present. In addition, the residual current protection device cannot determine the type of fault.
The accurate classification and identification of faults have certain guiding significance on whether the residual current protection device acts correctly. Because of the difficulty in directly identifying and classifying the obtained current or voltage original signals, the current or voltage signal classification method in the power system mainly comprises two key steps of feature quantity extraction and classifier design. The method is characterized in that the method comprises the steps of extracting statistical characteristic parameters from total leakage current signals according to a novel method [ Han Xiaohui, du Songhuai, su Juan, liu Guangeng ] for electric shock signal transient characteristic extraction and fault type identification of a support vector machine [ J ]. A power grid technology, 2016,40 (11): 3591-3596 ], and then inputting characteristic quantities into the support vector machine for analysis through dimension reduction processing. This method is based on the accuracy of the total leakage current measured by the residual current protection device.
The technical scheme is based on the residual current protection device, the device is required to be additionally arranged in the low-voltage distribution network, the measurement accuracy of the device seriously influences whether the subsequent electric shock fault is identified or not, and whether the subsequent electric shock fault can be accurately classified or not is judged. The method has the advantages that the material of the high-sensitivity sensor is researched and replaced, the measurement accuracy of the device is improved, and the action value and action principle of the fault detection device are not changed.
Disclosure of Invention
In order to solve the technical problems, the invention provides a single-phase fault identification and classification method for a low-voltage neutral point ungrounded system, which is simple in algorithm and high in identification precision.
The technical scheme for solving the problems is as follows: a single-phase fault identification and classification method for a low-voltage neutral point ungrounded system comprises the following steps:
1) Classifying single-phase faults of a neutral point ungrounded system to obtain 3 fault types;
2) Carrying out one-hot coding for various fault types, namely labeling each fault type;
3) 3 faults are implemented at the tail end of a certain phase line at random, the faults occur at a certain moment in 9 cycles at random, and the data of 9 cycles of the phase current and the phase voltage of the phase are recorded;
4) Carrying out wavelet decomposition on phase current and phase voltage data in layers 2, 3, 4, 5 and 6, wherein each layer of wavelet decomposition obtains a wavelet decomposition vector and a length vector;
5) Carrying out wavelet reconstruction on the two vectors obtained after each layer of wavelet decomposition to obtain high-frequency components of layers 2, 3, 4 and 5 and high-frequency and low-frequency components of layer 6;
6) Taking the high-frequency component and the low-frequency component as the identification characteristic of the electric shock signal, and carrying out the normalization of each data characteristic quantity;
7) Randomly classifying the obtained characteristic quantity in combination with the corresponding label, wherein 60% of the characteristic quantity is used as a training set, 40% of the characteristic quantity is used as a test set, the data of the training set and the corresponding label are used as the input of an XGBoost algorithm, and a model is obtained through training of the algorithm, so that a fault recognition and classification model is obtained;
8) For new fault judgment, collecting phase voltage and phase current data of 9 cycles, extracting characteristic quantity after wavelet decomposition and reconstruction, and then inputting the characteristic quantity into a fault identification and classification model to obtain a fault judgment result: one of 3 fault types, no fault, ground fault, zero fault.
In the above method for identifying and classifying single-phase faults of a low-voltage neutral point ungrounded system, in the step 1), for a neutral point ungrounded system, there are only 2 single-phase faults: ground fault and zero fault, regarding the fault-free situation as a fault type, 3 fault types are obtained: no fault exists; zero fault connection; ground fault.
In the above single-phase fault identification and classification method for the low-voltage neutral point ungrounded system, in the step 2), the result of performing one-hot coding on the fault type is that there is no fault: [1 0 0] Zero fault [ 01 0], ground fault [ 01 ].
In the above single-phase fault identification and classification method for the low-voltage neutral point ungrounded system, in the step 3), 200 points are collected for each cycle, and one time is 1800 points.
In the above single-phase fault identification and classification method for the low-voltage neutral point ungrounded system, in step 5), the data length of each high-frequency component and each low-frequency component is 1800.
In the above method for identifying and classifying single-phase faults of a system with a low-voltage neutral point not grounded, in the step 6), the current data has 6 feature values, the voltage data has 6 feature values, one piece of current data+voltage data is 3600×1, the final obtained feature gauge is 1800×12 in matrix form, and the corresponding column sequences are respectively: a 6 th layer low frequency component, a 2 nd layer high frequency component, a 3 rd layer high frequency component, a 4 th layer high frequency component, a 5 th layer high frequency component, a 6 th layer low frequency component, a 2 nd layer high frequency component, a 3 rd layer high frequency component, a 4 th layer high frequency component, a 5 th layer high frequency component, a 6 th layer high frequency component of the current.
The invention has the beneficial effects that:
1. the method comprises the steps of firstly collecting phase current phase voltage data of a certain phase under various fault conditions, then adopting a wavelet decomposition and wavelet reconstruction signal processing method to obtain high-frequency components of layers 2, 3, 4 and 5, and using the high-frequency components and the low-frequency components of layer 6 as characteristic quantity combined fault type labels, and simultaneously inputting an xgboost algorithm to perform model training, wherein the obtained model is a fault identification and classification model; and for the new fault judgment, collecting phase voltage and phase current data, extracting the characteristic quantity after wavelet decomposition and reconstruction, and inputting the characteristic quantity into a fault identification and classification model to obtain a fault judgment result.
2. The invention can reduce additional hardware measuring devices, and only performs feature extraction and establishment of a classification model by means of easily available phase voltage and phase current data, thereby finally realizing single-phase fault identification and classification in a neutral point ungrounded system.
3. The algorithm adopted by the invention can be converted into c language, can be deployed in the embedded terminal equipment, and has a certain practical application value.
Drawings
FIG. 1 is a flow chart of the present invention.
Fig. 2 is a schematic diagram of a simulink model built in matlab in an embodiment of the present invention.
Fig. 3 is a schematic diagram of voltage characteristic quantities collected in an embodiment of the present invention.
Fig. 4 is a schematic diagram of current characteristic quantities collected in an embodiment of the present invention.
Detailed Description
The invention is further described below with reference to the drawings and examples.
Taking a single-phase fault of the A phase as an example for specific step description, as shown in fig. 1, a single-phase fault identification and classification method for a low-voltage neutral point ungrounded system comprises the following steps:
1) Classifying single-phase faults of the neutral point ungrounded system: for neutral ungrounded systems, there are only 2 single-phase faults: ground fault and zero fault, regarding the fault-free situation as a fault type, 3 fault types are obtained: no fault exists; zero fault connection; ground fault.
2) Carrying out one-hot coding for various fault types, namely labeling each fault type; the result of the coding is that there is no fault: [1 0 0] Zero fault [ 01 0], ground fault [ 01 ].
3) 3 faults are randomly implemented at the tail end of the phase A, the faults randomly occur at a certain moment in 9 cycles (0.18 s), the data of 9 cycles of phase current and phase voltage of the phase A are recorded, 200 points are collected for each cycle, and 1800 points are collected at one time.
4) And carrying out layer 2, layer 3, layer 4, layer 5 and layer 6 wavelet decomposition on the phase current and phase voltage data, wherein each layer of wavelet decomposition obtains a wavelet decomposition vector and a length vector.
5) And carrying out wavelet reconstruction on the two vectors obtained after each layer of wavelet decomposition to obtain high-frequency components of layers 2, 3, 4 and 5, and obtaining high-frequency and low-frequency components of layer 6, wherein the data length of each high-frequency component and each low-frequency component is 1800.
6) And taking the high-frequency component and the low-frequency component as the identification characteristic of the electric shock signal, and carrying out the normalization of each data characteristic quantity.
The current data has 6 characteristic quantities, the voltage data has 6 characteristic quantities, one piece of current data+voltage data is 3600 x 1, the final obtained characteristic gauge is arranged into a matrix form 1800 x 12, and the corresponding column sequences are respectively: a 6 th layer low frequency component, a 2 nd layer high frequency component, a 3 rd layer high frequency component, a 4 th layer high frequency component, a 5 th layer high frequency component, a 6 th layer low frequency component, a 2 nd layer high frequency component, a 3 rd layer high frequency component, a 4 th layer high frequency component, a 5 th layer high frequency component, a 6 th layer high frequency component of the current.
7) The obtained feature quantity is combined with the corresponding label to carry out random classification, 60% part is used as a training set, 40% part is used as a test set, the test set is input into an XGBoost algorithm in python to carry out generation of a classification model, the model can be understood as a function from a simple aspect, parameters of the function are obtained through training, and therefore a fault identification and classification model is obtained and can be stored in a program in a pkl storage mode.
8) For new fault judgment, collecting phase voltage and phase current data of 9 cycles, extracting characteristic quantity after wavelet decomposition and reconstruction, and then inputting the characteristic quantity into a fault identification and classification model to obtain a fault judgment result: one of 3 fault types, no fault, ground fault, zero fault.
The invention uses XGBoost algorithm to train and generate model, and compared with classical algorithm in other machine learning, the invention has higher accuracy on the same data set, see the table below. The effectiveness of the method employed is illustrated.
Definition of accuracy: the number of samples/the number of all samples of the predictive label is the same as the normal label.
Claims (6)
1. A single-phase fault identification and classification method for a low-voltage neutral point ungrounded system comprises the following steps:
1) Classifying single-phase faults of a neutral point ungrounded system to obtain 3 fault types;
2) Carrying out one-hot coding for various fault types, namely labeling each fault type;
3) 3 faults are implemented at the tail end of a certain phase line at random, the faults occur at a certain moment in 9 cycles at random, and the data of 9 cycles of the phase current and the phase voltage of the phase are recorded;
4) Carrying out wavelet decomposition on phase current and phase voltage data in layers 2, 3, 4, 5 and 6, wherein each layer of wavelet decomposition obtains a wavelet decomposition vector and a length vector;
5) Carrying out wavelet reconstruction on the two vectors obtained after each layer of wavelet decomposition to obtain high-frequency components of layers 2, 3, 4 and 5 and high-frequency and low-frequency components of layer 6;
6) Taking the high-frequency component and the low-frequency component as the identification characteristic of the electric shock signal, and carrying out the normalization of each data characteristic quantity;
7) Randomly classifying the obtained characteristic quantity in combination with the corresponding label, wherein 60% of the characteristic quantity is used as a training set, 40% of the characteristic quantity is used as a test set, the data of the training set and the corresponding label are used as the input of an XGBoost algorithm, and a model is obtained through training of the algorithm, so that a fault recognition and classification model is obtained;
8) For new fault judgment, collecting phase voltage and phase current data of 9 cycles, extracting characteristic quantity after wavelet decomposition and reconstruction, and then inputting the characteristic quantity into a fault identification and classification model to obtain a fault judgment result: one of 3 fault types, no fault, ground fault, zero fault.
2. The method for identifying and classifying single-phase faults of a low-voltage neutral point ungrounded system according to claim 1, wherein in the step 1), for a neutral point ungrounded system, there are only 2 single-phase faults: ground fault and zero fault, regarding the fault-free situation as a fault type, 3 fault types are obtained: no fault exists; zero fault connection; ground fault.
3. The method for identifying and classifying single-phase faults of a low-voltage neutral point ungrounded system according to claim 1, wherein in the step 2), the result of performing one-hot coding on the fault type is that no fault exists: [1 0 0] Zero fault [ 01 0], ground fault [ 01 ].
4. The method for identifying and classifying single-phase faults of a low-voltage neutral point ungrounded system according to claim 1, wherein in the step 3), 200 points are collected for each cycle, and 1800 points are collected at a time.
5. The method for single-phase fault identification and classification of a low-voltage neutral point ungrounded system according to claim 1, wherein in step 5), the data length of each of the high-frequency component and the low-frequency component is 1800.
6. The method for identifying and classifying single-phase faults of a low-voltage neutral point ungrounded system according to claim 1, wherein in the step 6), the current data has 6 feature values, the voltage data has 6 feature values, one piece of current data+voltage data is 3600×1, the final obtained feature gauge is 1800×12 in matrix form, and the corresponding column sequences are respectively: a 6 th layer low frequency component, a 2 nd layer high frequency component, a 3 rd layer high frequency component, a 4 th layer high frequency component, a 5 th layer high frequency component, a 6 th layer low frequency component, a 2 nd layer high frequency component, a 3 rd layer high frequency component, a 4 th layer high frequency component, a 5 th layer high frequency component, a 6 th layer high frequency component of the current.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911052914.1A CN110991240B (en) | 2019-10-31 | 2019-10-31 | Single-phase fault identification and classification method for low-voltage neutral point ungrounded system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911052914.1A CN110991240B (en) | 2019-10-31 | 2019-10-31 | Single-phase fault identification and classification method for low-voltage neutral point ungrounded system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110991240A CN110991240A (en) | 2020-04-10 |
| CN110991240B true CN110991240B (en) | 2023-06-09 |
Family
ID=70082712
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911052914.1A Active CN110991240B (en) | 2019-10-31 | 2019-10-31 | Single-phase fault identification and classification method for low-voltage neutral point ungrounded system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110991240B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112396109A (en) * | 2020-11-19 | 2021-02-23 | 天津大学 | Motor bearing fault diagnosis method based on recursion graph and multilayer convolution neural network |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106249101A (en) * | 2016-06-30 | 2016-12-21 | 湖南大学 | A kind of intelligent distribution network fault identification method |
| CN109325526A (en) * | 2018-09-05 | 2019-02-12 | 福州大学 | A kind of distribution network failure classification method using convolution depth confidence network |
| CN109754110A (en) * | 2017-11-03 | 2019-05-14 | 株洲中车时代电气股份有限公司 | A kind of method for early warning and system of traction converter failure |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100169030A1 (en) * | 2007-05-24 | 2010-07-01 | Alexander George Parlos | Machine condition assessment through power distribution networks |
| CN107894564B (en) * | 2017-11-09 | 2020-02-18 | 合肥工业大学 | Analog circuit fault diagnosis method based on cross wavelet characteristics |
-
2019
- 2019-10-31 CN CN201911052914.1A patent/CN110991240B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106249101A (en) * | 2016-06-30 | 2016-12-21 | 湖南大学 | A kind of intelligent distribution network fault identification method |
| CN109754110A (en) * | 2017-11-03 | 2019-05-14 | 株洲中车时代电气股份有限公司 | A kind of method for early warning and system of traction converter failure |
| CN109325526A (en) * | 2018-09-05 | 2019-02-12 | 福州大学 | A kind of distribution network failure classification method using convolution depth confidence network |
Non-Patent Citations (2)
| Title |
|---|
| 张君琦 ; 杨帆 ; 郭谋发 ; .配电网高阻接地故障时频特征SVM分类识别方法.电气技术.2018,(第03期),全文. * |
| 王艳松 ; 谭志勇 ; 刘学民 ; .基于小波奇异熵和支持向量机的配电网故障类型识别.电力系统保护与控制.2011,(第23期),全文. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110991240A (en) | 2020-04-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106526400B (en) | The earth fault diagnostic method and device of DC600V power supply system of train | |
| Wang et al. | ArcNet: Series AC arc fault detection based on raw current and convolutional neural network | |
| CN107037313B (en) | The method for establishing deformation of transformer winding failure Yu frequency sweep impedance characteristic corresponding relationship | |
| CN109633368A (en) | The method of duration power quality disturbances containing distributed power distribution network based on VMD and DFA | |
| CN103116090A (en) | Three-phrase pulse-width modulation (PWM) rectifier fault diagnosis method based on wavelet packet analysis and support vector machine | |
| CN103018629A (en) | Method for analyzing power system fault recording data based on Marla algorithm | |
| CN108169583B (en) | Autotransformer direct-current magnetic bias discrimination method and system with neutral point grounded through capacitor | |
| CN110261746A (en) | Electric cable stoppage detection method based on oscillating wave voltage periodic attenuation characteristic | |
| CN112069930A (en) | Vibration signal processing method and device for improving GIS equipment fault diagnosis accuracy | |
| CN105137277A (en) | Diagnosis method of winding machinery state in case of sudden short circuit of transformer | |
| CN110703078A (en) | GIS fault diagnosis method based on spectral energy analysis and self-organizing competition algorithm | |
| CN109142963B (en) | Interturn short circuit fault positioning system and method | |
| CN103558532A (en) | Partial discharge on-line detection system of high-voltage crosslinked polyethylene power cable | |
| CN102616156A (en) | Device and method for detecting and elevating contact network state based on electrical capacity analysis | |
| CN110991240B (en) | Single-phase fault identification and classification method for low-voltage neutral point ungrounded system | |
| CN106291164A (en) | Noise measuring is utilized to judge the method and system of transformer DC magnetic bias state | |
| CN106980051B (en) | A kind of intermittence tandem type fault electric arc recognition methods | |
| CN108510002B (en) | Method for detecting short circuit impact resistance of wound core traction transformer winding | |
| Wang et al. | Identification of series arc fault occurred in the three-phase motor with frequency converter load circuit via VMD and entropy-based features | |
| CN104062555B (en) | The discrimination method of distribution line high resistance earthing fault characteristic harmonics | |
| CN112132226A (en) | Method for diagnosing defects of +/-1100 kV composite insulator | |
| CN116910470A (en) | GIS combined electrical apparatus partial discharge fault mode identification method | |
| CN116842425A (en) | Method for detecting fault arc in low-voltage alternating current system | |
| CN110929673A (en) | Transformer winding vibration signal identification method based on ITD (inverse discrete cosine transformation) permutation entropy and CGWO-SVM (Carrier-support vector machine) | |
| Hosseini | Short circuit fault classification and location in transmission lines using a combination of wavelet transform and support vector machines |
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 |