CN110988597A - Resonance type detection method based on neural network - Google Patents
Resonance type detection method based on neural network Download PDFInfo
- Publication number
- CN110988597A CN110988597A CN201911288046.7A CN201911288046A CN110988597A CN 110988597 A CN110988597 A CN 110988597A CN 201911288046 A CN201911288046 A CN 201911288046A CN 110988597 A CN110988597 A CN 110988597A
- Authority
- CN
- China
- Prior art keywords
- resonance
- neural network
- module
- picture
- microcomputer
- 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
Images
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
- G01R31/088—Aspects of digital computing
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R23/00—Arrangements for measuring frequencies; Arrangements for analysing frequency spectra
- G01R23/16—Spectrum analysis; Fourier analysis
Landscapes
- Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Testing Electric Properties And Detecting Electric Faults (AREA)
Abstract
The invention discloses a resonance type detection method based on a neural network, which adopts a microcomputer resonance elimination control device to collect voltage information and current information, presets a threshold value to trigger transient recording, stores every five cycles as picture information, identifies the picture information based on a neural network operation module, further judges whether the waveform belongs to resonance or not, further judges the resonance type of the resonance, and finally controls the resonance elimination equipment to effectively act through the microcomputer resonance elimination control device; the method can effectively improve the accuracy of the resonance type, solves the problem of difficult discrimination of fundamental frequency resonance and single-phase short circuit, can effectively avoid the protection device from not moving, and can improve the action reliability of the resonance elimination equipment.
Description
Technical Field
The invention belongs to the technical field of power system detection, and particularly relates to a resonance type detection method based on a neural network.
Background
In an electric power system, there are many capacitive elements and inductive elements, which constitute a series of vibration circuits, and a ferroresonance phenomenon is likely to occur due to operations such as circuit breaker switching. In a system with ungrounded neutral points, a primary side of a voltage transformer installed on a substation bus side is a unique neutral point contact point, and when ferromagnetic resonance occurs, the current on the primary side of the voltage transformer is increased sharply to fuse a high-voltage fuse; if overcurrent occurs for a long time, the voltage transformer is burnt.
The existing resonance judging method judges based on the steady state change of each phase voltage and opening voltage, but is difficult to effectively distinguish when the fundamental frequency resonance is similar to a single-phase fault due to other appearances.
Disclosure of Invention
The invention provides a resonance type detection method based on a neural network, which aims at solving the problems in the prior art, and the method comprises the steps of utilizing a microcomputer resonance elimination control device to collect voltage information and current information, triggering transient state wave recording by presetting a threshold value, storing every five cycles as picture information, inputting the picture information into a convolution neural network operation module, identifying the picture information based on the neural network operation module, further judging whether a waveform belongs to resonance or not, further judging the resonance type of the resonance, and finally controlling effective action of resonance elimination equipment through the microcomputer resonance elimination control device.
The invention relates to a microcomputer resonance elimination control device, which comprises:
the transient wave recording module is used for recording waves based on a preset fixed value, 1.2Un is used as a wave recording condition, 0.9Un is used as a temporary drop wave recording condition, Un is a rated voltage of a line, and the number of sampling points is 512 points/cycle wave during wave recording;
the waveform conversion module is used for converting every five cycles into a picture, and the picture comprises the relation information between time and a voltage value and between the time and a current value;
the convolutional neural network module is used for processing pictures based on a convolutional neural network and matching through voltage waveforms and current waveforms, and input data comprise Ua, Ia, Ub, Ib, Uc and Ic data;
the complete connection layer module is used for judging the generation type of the ferromagnetic resonance of the processed picture, and the judged generation type comprises frequency division resonance, fundamental frequency resonance, high frequency resonance and the like;
the comprehensive judgment module is used for determining and determining the resonance generation type judged by the complete connection layer module when more than 3 of 6 output results in the generation types are the same; otherwise, waiting for the next group of picture judgment results.
The microcomputer resonance elimination control device is accessed with voltage and current signals, and sets triggering wave recording conditions based on the voltage, wherein 1.2Un is usually selected and Un is the rated voltage of a line; the convolutional neural network module cuts every five cycles into a picture, matching is carried out by utilizing a convolutional neural network algorithm, whether the picture belongs to a resonance or a resonance type is judged, and whether harmonic elimination equipment acts or not is finally determined according to voltage and current matching conditions.
The method comprises the following specific operations:
1. a transient wave recording module is added in the microcomputer harmonic elimination control device, the module records waves based on a preset value, 1.2Un is used as a wave recording condition, 0.9% Un is used as a temporary drop wave recording condition, Un is a rated voltage of a line, and the number of sampling points is 512 points/cycle wave during wave recording;
2. converting every five cycles into a picture by using a waveform conversion module in the microcomputer harmonic elimination control device, wherein the picture comprises the relation information between time and a voltage value as well as between the time and a current value;
3. inputting picture information into a convolutional neural network module, wherein the construction and training of the convolutional neural network are conventional methods, after the picture information is input into the convolutional neural network module, voltage waveforms and current waveforms are matched, and input data comprise 6 types of data such as Ua, Ia, Ub, Ib, Uc and Ic;
4. judging the generation types of ferromagnetic resonance of the processed picture by using a complete connection layer module in the microcomputer resonance elimination control device, wherein the resonance generation types comprise frequency division resonance, fundamental frequency resonance, high frequency resonance and the like;
5. in the invention, 6 output results are utilized to carry out comprehensive judgment, and when more than 3 of the 6 output results are the same, the occurrence type of the result is determined to be judged by a complete connection layer module; otherwise, waiting for the next image judgment result.
The discrimination method provided by the invention has the following advantages: 1. the transient waveform is used as a judgment basis, and other response speeds are higher; 2. ferromagnetic resonance types can be effectively distinguished based on transient waveforms, and particularly fundamental frequency resonance and single-phase short circuit can be effectively distinguished, so that the resonance device is more reliable in action. 3. Meanwhile, the reliability can be improved by using voltage and current waveforms as criteria for resonance generation.
The invention utilizes the convolution neural network to analyze the voltage and the current of the transmission resonance or the single-phase grounding, can effectively distinguish the resonance from the single-phase grounding, can avoid misleading of protective equipment, and ensures the effective work of harmonic elimination equipment.
Drawings
FIG. 1 is a flow chart of ferroresonance discrimination;
fig. 2 is a diagram of a neural network classification picture implementation process.
Detailed Description
The present invention is further illustrated by the following examples, but the scope of the invention is not limited to the above-described examples.
Example 1: as shown in fig. 1 and 2, the resonance type detection method based on the neural network is completed by using a microcomputer resonance elimination control device, and the microcomputer resonance elimination control device comprises: the device comprises a transient state wave recording module, a waveform conversion module, a convolutional neural network module, a complete connection layer module and a comprehensive judgment module;
step 1: adding a transient wave recording function in the microcomputer resonance elimination control device, namely adding a transient wave recording module, wherein the module carries out wave recording based on a preset fixed value, 1.2Un is used as a wave recording condition, 0.9Un is used as a temporary drop wave recording condition, Un is a rated voltage of a circuit, and the number of sampling points is 512 points/cycle wave during wave recording;
step 2: conversion of waveforms into pictures
In order to realize the analysis and the judgment of the resonance type based on the convolutional neural network, a wave recording waveform needs to be converted into a picture, a waveform conversion module is additionally arranged in the microcomputer harmonic elimination control device, every five cycles are converted into one picture, and the picture stores information between time and a voltage value as well as information between time and a current value; to simplify the analysis, the invention provides an embodiment that only grey scale images are considered, each pixel value in the matrix ranging between 0 and 255, 0 for black and 255 for white; converting the picture into a gray matrix of 0-255;
and step 3: a convolutional neural network module is added in the microcomputer harmonic elimination control device, the convolutional neural network is responsible for picture processing, matching is carried out on the basis of voltage waveforms and current waveforms respectively, and 6 types of data including Ua, Ia, Ub, Ib, Uc, Ic and the like are input;
the convolution is realized, a characteristic detector is used for carrying out convolution operation, a Gaussian function is used as the characteristic detector in the embodiment, and nonlinear processing is required during the convolution, so that the embodiment of the Relu curve is provided; after convolution calculation, pooling calculation is carried out until the last layer, and 35 layers are trained (figure 2);
and 4, step 4: a complete connection layer module is added in the microcomputer resonance elimination control device, and the generation type of ferromagnetic resonance is judged after the image processing of 6 types of data is carried out, and the result is frequency division resonance, fundamental frequency resonance, high frequency resonance or other results;
the fully connected layer is a traditional multilayer perceptron, which uses an activation function at the output layer, the present embodiment uses a softmax function, each neuron in the previous layer of the fully connected layer is connected to each neuron in the next layer; the output of the convolutional and pooling layers represents the high-level features of the input image; the purpose of the complete connection layer is to divide the input images into different classes by using the features obtained based on the training data set, and the complete connection layer is 2 layers in the embodiment;
and 5: a comprehensive judgment module; in order to improve the reliability of the discrimination, the embodiment utilizes 6 output results to perform comprehensive discrimination, and when more than 3 of the 6 subconstructions are the same, the occurrence type of the complete connection layer module discrimination is judged; otherwise, waiting for the next image judgment result.
Claims (2)
1. A resonance type detection method based on a neural network is characterized in that: the method comprises the steps of collecting voltage information and current information by a microcomputer resonance elimination control device, triggering transient state recording by a preset threshold value, storing every five cycles as picture information, identifying the picture information based on a neural network operation module, further judging whether a waveform belongs to resonance or not, further judging the type of resonance if the waveform belongs to resonance, and finally controlling effective action of resonance elimination equipment by the microcomputer resonance elimination control device.
2. The neural network-based resonance class detection method according to claim 1, wherein the microcomputer resonance elimination control means includes:
the transient wave recording module is used for recording waves based on a preset fixed value, 1.2Un is used as a wave recording condition, 90% Un is used as a temporary drop wave recording condition, Un is a rated voltage of a line, and the number of sampling points is 512 points/cycle wave during wave recording;
the waveform conversion module is used for converting every five cycles into a picture, and the picture comprises the relation information between time and a voltage value and between the time and a current value;
the convolutional neural network module is used for processing pictures based on a convolutional neural network, matching voltage waveforms and current waveforms, and inputting data comprising Ua, Ia, Ub, Ib, Uc and Ic data;
the complete connection layer module is used for judging the generation type of the ferromagnetic resonance of the processed picture, wherein the generation type comprises frequency division resonance, fundamental frequency resonance, high frequency resonance and the like;
and the comprehensive judgment module determines that the generated type is the resonance generation type judged by the complete connection layer module when more than 3 of the 6 output results in the generation types are the same.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911288046.7A CN110988597A (en) | 2019-12-15 | 2019-12-15 | Resonance type detection method based on neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911288046.7A CN110988597A (en) | 2019-12-15 | 2019-12-15 | Resonance type detection method based on neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110988597A true CN110988597A (en) | 2020-04-10 |
Family
ID=70093666
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911288046.7A Pending CN110988597A (en) | 2019-12-15 | 2019-12-15 | Resonance type detection method based on neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110988597A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112798894A (en) * | 2021-02-02 | 2021-05-14 | 中国南方电网有限责任公司超高压输电公司柳州局 | Transient event triggering wave recording test method for ultrahigh voltage transformer substation |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103399257A (en) * | 2013-07-31 | 2013-11-20 | 武汉大学 | Ferromagnetic resonance failure detection method of neutral point ungrounded system |
| CN104375025A (en) * | 2014-10-31 | 2015-02-25 | 国家电网公司 | Diagnostic method for ferromagnetic resonance in neutral non-grounding 10kV system |
| CN106018991A (en) * | 2016-04-15 | 2016-10-12 | 国网江苏省电力公司盐城供电公司 | Power distribution network ferroresonance and single phase earth fault calculating analysis method and device |
| CN106229946A (en) * | 2016-08-17 | 2016-12-14 | 云南电网有限责任公司电力科学研究院 | A kind of ferromagnetic resonance harmonic elimination apparatus and method |
| CN106646043A (en) * | 2016-12-13 | 2017-05-10 | 国网江苏省电力公司淮安供电公司 | Ferromagnetic resonance online monitoring system and ferromagnetic resonance classification recognition method for power distribution network |
| 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 |
| CN106709511A (en) * | 2016-12-08 | 2017-05-24 | 华中师范大学 | Urban rail transit panoramic monitoring video fault detection method based on depth learning |
| CN106771520A (en) * | 2016-12-15 | 2017-05-31 | 福州大学 | A kind of power distribution network temporary overvoltage classifying identification method and device |
| CN107332220A (en) * | 2017-08-29 | 2017-11-07 | 安徽三和电力技术有限公司 | For microcomputer harmonic elimination device control system |
| CN206696364U (en) * | 2016-04-15 | 2017-12-01 | 国网江苏省电力公司盐城供电公司 | A kind of power distribution network ferromagnetic resonance calculates analysis special purpose device with singlephase earth fault |
| CN107450016A (en) * | 2017-07-24 | 2017-12-08 | 西安工程大学 | Fault Diagnosis for HV Circuit Breakers method based on RST CNN |
| CN109612708A (en) * | 2018-12-28 | 2019-04-12 | 东北大学 | Based on the power transformer on-line detecting system and method for improving convolutional neural networks |
-
2019
- 2019-12-15 CN CN201911288046.7A patent/CN110988597A/en active Pending
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103399257A (en) * | 2013-07-31 | 2013-11-20 | 武汉大学 | Ferromagnetic resonance failure detection method of neutral point ungrounded system |
| CN104375025A (en) * | 2014-10-31 | 2015-02-25 | 国家电网公司 | Diagnostic method for ferromagnetic resonance in neutral non-grounding 10kV system |
| CN106018991A (en) * | 2016-04-15 | 2016-10-12 | 国网江苏省电力公司盐城供电公司 | Power distribution network ferroresonance and single phase earth fault calculating analysis method and device |
| CN206696364U (en) * | 2016-04-15 | 2017-12-01 | 国网江苏省电力公司盐城供电公司 | A kind of power distribution network ferromagnetic resonance calculates analysis special purpose device with singlephase earth fault |
| CN106229946A (en) * | 2016-08-17 | 2016-12-14 | 云南电网有限责任公司电力科学研究院 | A kind of ferromagnetic resonance harmonic elimination apparatus and method |
| CN106709511A (en) * | 2016-12-08 | 2017-05-24 | 华中师范大学 | Urban rail transit panoramic monitoring video fault detection method based on depth learning |
| CN106646043A (en) * | 2016-12-13 | 2017-05-10 | 国网江苏省电力公司淮安供电公司 | Ferromagnetic resonance online monitoring system and ferromagnetic resonance classification recognition method for power distribution network |
| CN106771520A (en) * | 2016-12-15 | 2017-05-31 | 福州大学 | A kind of power distribution network temporary overvoltage classifying identification method and device |
| 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 |
| CN107450016A (en) * | 2017-07-24 | 2017-12-08 | 西安工程大学 | Fault Diagnosis for HV Circuit Breakers method based on RST CNN |
| CN107332220A (en) * | 2017-08-29 | 2017-11-07 | 安徽三和电力技术有限公司 | For microcomputer harmonic elimination device control system |
| CN109612708A (en) * | 2018-12-28 | 2019-04-12 | 东北大学 | Based on the power transformer on-line detecting system and method for improving convolutional neural networks |
Non-Patent Citations (1)
| Title |
|---|
| 钱永亮 等: "基于频域谐振阻抗的中高压电缆老化分析", 《云南电力技术》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112798894A (en) * | 2021-02-02 | 2021-05-14 | 中国南方电网有限责任公司超高压输电公司柳州局 | Transient event triggering wave recording test method for ultrahigh voltage transformer substation |
| CN112798894B (en) * | 2021-02-02 | 2023-10-03 | 中国南方电网有限责任公司超高压输电公司柳州局 | Transient event triggering wave recording test method for ultra-high voltage transformer substation |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11489490B2 (en) | Arc fault detection method for photovoltaic system based on adaptive kernel function and instantaneous frequency estimation | |
| Mohamed et al. | A neural network-based scheme for fault diagnosis of power transformers | |
| Wilson et al. | Automated identification of electrical disturbance waveforms within an operational smart power grid | |
| CN111812451B (en) | Phase current transient fault component-based distributed line selection method for power distribution network | |
| CN114414942B (en) | Power transmission line fault identification classifier, identification method and system based on transient waveform image identification | |
| Manohar et al. | Enhancing resilience of PV-fed microgrid by improved relaying and differentiating between inverter faults and distribution line faults | |
| Ezzat et al. | Microgrids islanding detection using Fourier transform and machine learning algorithm | |
| Lu et al. | A morphological scheme for inrush identification in transformer protection | |
| CN111398733A (en) | Ocean nuclear power platform power grid ground fault line selection protection method and system | |
| CN102567783A (en) | Expert fault analytical and diagnostic method of parallel mixed type power quality regulator | |
| CN112986754B (en) | Small current grounding system fault identification method and device based on data driving | |
| Kapoor | Six phase transmission line boundary protection using wavelet transform | |
| Kezunovic | Translational knowledge: From collecting data to making decisions in a smart grid | |
| CN106569093A (en) | Fault judgment method for multiple power transmission lines on same tower | |
| CN1805234A (en) | Pattern matching based security protection method for relay protection information of electric system in network environment | |
| Yuan et al. | Study on fault line detection methods based on multi‐feature fusion in distribution systems | |
| CN116338449A (en) | Online testing method and system for switching characteristics of circuit breaker | |
| CN110988597A (en) | Resonance type detection method based on neural network | |
| Menezes et al. | Islanding detection based on artificial neural network and S-transform for distributed generators | |
| Gnanamalar et al. | CNN–SVM Based Fault Detection, Classification and Location of Multi-terminal VSC–HVDC System | |
| CN117791512A (en) | Protection device for electric power system | |
| Chu et al. | TDV-LSTM: A new methodology for series arc fault detection in low power AC systems | |
| CN108493909A (en) | The detection method of Distribution Network Failure based on Voltage Drop | |
| CN110929673A (en) | Transformer winding vibration signal identification method based on ITD (inverse discrete cosine transformation) permutation entropy and CGWO-SVM (Carrier-support vector machine) | |
| Chen et al. | Power quality disturbance detection based on morphology singular entropy |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20200410 |
|
| WD01 | Invention patent application deemed withdrawn after publication |