CN112270273A - Wind driven generator fault part identification method based on GCN and WPT-MD - Google Patents
Wind driven generator fault part identification method based on GCN and WPT-MD Download PDFInfo
- Publication number
- CN112270273A CN112270273A CN202011199637.XA CN202011199637A CN112270273A CN 112270273 A CN112270273 A CN 112270273A CN 202011199637 A CN202011199637 A CN 202011199637A CN 112270273 A CN112270273 A CN 112270273A
- Authority
- CN
- China
- Prior art keywords
- gcn
- wpt
- fault
- identification method
- driven generator
- 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
- 238000000034 method Methods 0.000 title claims abstract description 25
- 238000003745 diagnosis Methods 0.000 claims abstract description 7
- 239000013598 vector Substances 0.000 claims description 18
- 238000013528 artificial neural network Methods 0.000 claims description 10
- 230000009466 transformation Effects 0.000 claims description 10
- 238000000354 decomposition reaction Methods 0.000 claims description 6
- 238000000605 extraction Methods 0.000 claims description 3
- 238000011161 development Methods 0.000 claims description 2
- 238000004364 calculation method Methods 0.000 claims 1
- 238000004088 simulation Methods 0.000 abstract 1
- 239000011159 matrix material Substances 0.000 description 12
- 238000010248 power generation Methods 0.000 description 3
- 238000007781 pre-processing Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- 238000005452 bending Methods 0.000 description 1
- 230000016507 interphase Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
- 238000004804 winding Methods 0.000 description 1
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
- G06F2218/02—Preprocessing
- G06F2218/04—Denoising
- G06F2218/06—Denoising by applying a scale-space analysis, e.g. using wavelet analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- 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
Abstract
A wind driven generator fault position identification method based on GCN and WPT-MD is provided. Firstly, WTP, MD and GCN algorithms are simply introduced, then GCN and WPT-MD are combined, and a method based on GCN and WPT-MD is provided. The cage-type asynchronous generator is used as an application object to carry out simulation experiments, and the experimental result shows that the algorithm has higher diagnosis speed and minimum loss rate.
Description
Technical Field
The invention relates to a novel method for identifying a fault position of a wind driven generator, in particular to a method for identifying a fault position of a wind driven generator based on GCN and WPT-MD.
Background
The failure rate of the wind driven generator is gradually reduced along with the development of the wind turbine technology, but compared with the traditional power generation system, such as a steam turbine, a gas turbine, a water turbine and the like, the failure rate of the wind driven generator is still very high, and the operation reliability of the wind driven generator still needs to be further enhanced and improved. The cage-type asynchronous generator plays an important role in a wind power generation system, the generator is a key component in a wind turbine generator, and the generator is broken down to cause great influence on the power generation of a fan, so that the diagnosis of the faults of the cage-type asynchronous generator has certain value. Through various checking and testing methods, whether the equipment has faults or not is found, and the position of the faults or the type of the faults is further determined to be called fault diagnosis, and the fault diagnosis can also be regarded as a fault identification problem.
Disclosure of Invention
The invention aims to solve the problems of multiple faults, low diagnosis speed and the like of an asynchronous generator and provides a wind driven generator fault part identification method based on wavelet packet transformation, Mahalanobis distance and graph convolution neural network. The main study is carried out on the faults of the cage type asynchronous generator which is most commonly used in the wind power generator. The method comprises the steps of firstly collecting normal data and other five fault data of the asynchronous generator, then preprocessing the collected sample data by utilizing wavelet packet transformation and Mahalanobis distance, constructing a graph structure, then realizing fault diagnosis of the asynchronous generator by utilizing a graph convolution neural network, and effectively separating the normal generator data from the five fault data.
In order to achieve the purpose, the invention adopts the following technical scheme:
a wind driven generator fault position identification method based on GCN and WPT-MD is disclosed. The method comprises a data preprocessing module, wavelet packet transformation, Mahalanobis distance and graph convolution neural network. The data preprocessing is to normalize the acquired original current signal data and eliminate the influence of magnitude order. Wavelet packet transformation is used for extracting data features, and extracted feature vectors calculate the similarity between the vectors by using the Mahalanobis distance. The main role of the graph convolution neural network is to identify the eventual distinct faults.
The invention mainly utilizes the following technologies:
1. wavelet packet transformation
Wavelet Packet Transform (WPT) is an improvement on Wavelet Transform, and the main algorithm idea is as follows: on the basis of wavelet transformation, when each level of signal decomposition is carried out, the low-frequency sub-band is further decomposed, the high-frequency sub-band is also further decomposed, finally, an optimal signal decomposition path is calculated by minimizing a cost function, and the original signal is decomposed by the decomposition path. The formula for wavelet packet transformation is:
in the formula (1), the first and second groups,respectively, wavelet packet decomposition coefficient, hk-21,gk-21Low-pass and high-pass filter coefficients, respectively, of wavelet packet decomposition, WPT coefficient dij(k) The sum is equal to the original time domain vibration signal, resulting in a normalized energy feature vector En.
2. Mahalanobis distance
Mahalanobis Distance (MD) is a Distance measure, which is an efficient way to calculate the similarity between two unknown sample sets. Unlike the euclidean distance, it allows for a link between various characteristics. The mahalanobis distance between data points x, y can be calculated using equation (2):
wherein Σ is a covariance matrix of multidimensional random variables, and if the covariance matrix is a unit vector, that is, each dimension is independently and identically distributed, the mahalanobis distance becomes the euclidean distance.
3. Graph convolution neural network
The GCN is a neural network layer, and assuming that a group of data has N nodes, each node has its own characteristics, and the characteristics of the nodes form an N × D matrix X, and then the relationship between the nodes also forms an N × N matrix a, also called an adjacency matrix. X and A are the inputs to our model. The propagation method between layers of the GCN is as follows:
Drawings
FIG. 1 is a flow chart of a wind driven generator fault position identification method based on GCN and WPT-MD.
Detailed Description
The present invention will be described in detail below with reference to the accompanying drawings and using various techniques.
The implementation process of the whole wind driven generator fault part identification method is described in fig. 1, and the process is as follows.
The first step is data acquisition. The SCADA system monitors and collects data. The asynchronous generator in the generator set is respectively arranged in six different states of a normal state, a stator winding interphase short circuit fault, air gap eccentricity, bearing fault, rotating shaft bending and rotor broken bar, and current signal data in the six different states are acquired. Each type of state selects 5 groups, each group has 10000 sampling points, and comprises 10 periods, and 6 periods are intercepted to be used as a total sample of the experiment. 24000 as training set, 6000 as test sample.
And a second step of feature extraction. In order to avoid a frequency mixing phenomenon in a characteristic extraction process, wavelet packet transformation is adopted to extract characteristics of collected generator fault data and normal data.
The third step is to build the graph structure. The similarity between feature vectors may represent the relationship between them. Since mahalanobis distance can take into account the link between various characteristics, we have chosen mahalanobis distance to compute the similarity of each unknown sample set. Different distances are obtained by calculating the similarity among the characteristic vectors, the smaller the distance is, the greater the similarity is, and conversely, the greater the distance is, the smaller the similarity is. Through multiple experiments, when the threshold value of the mahalanobis distance is 0.3, the relationship among most feature vectors can exist as much as possible, namely: when the distance between any two feature vectors is less than 0.3, the two feature vectors may be defined as having a certain relationship. And constructing a topological graph structure by using the relation between the feature vectors. And taking each feature vector as a node, and constructing a topological structure for the edges according to the relation among the feature vectors.
And fourthly, identifying the fault part of the wind driven generator. And inputting the graph structure constructed in the third step into a graph convolution neural network. The GCN inputs two matrices, N × N adjacency matrix and N × D feature matrix. N is the number of nodes, and in this document, we use the feature vector as the number of nodes, and the feature vector of the six types of data extracted by wavelet packet transformation is 2705, so an adjacency matrix of 2705 x 2705 is input. And D is the degree matrix of the adjacency matrix. Judging whether a convergence condition is met, if so, ending the process, and realizing the identification of the fault part of the wind driven generator; if not, continuing to classify.
Claims (4)
1. A wind driven generator fault part identification method based on GCN and WPT-MD is characterized in that a graph convolution neural network (GCN), Wavelet Packet Transform (WPT) and Mahalanobis Distance (MD) are fused, and a wind driven generator is used as an application object to identify a fault part;
the WPT is used for extracting different features by performing signal decomposition on each level of signals to construct feature vectors;
the MD is used for calculating the distance between different feature vectors, and the similarity between different feature vectors is obtained by setting a threshold, so that the method is an effective method for calculating the similarity of a sample set;
the GCN is used for processing data such as high-dimensional and nonlinear big data, the collected and preprocessed data are input into the GCN, and fault classification is realized by using the GCN, so that a fault part identification method is realized.
2. The GCN and WPT-MD based wind turbine generator failure site identification method according to claim 1, wherein the wind turbine generator failure site can be identified rapidly and efficiently.
3. The GCN and WPT-MD based wind turbine generator fault location identification method according to claim 2 can quickly and effectively identify fault locations of wind turbines. The method comprises three steps: wavelet packet transformation realizes signal feature extraction, Mahalanobis distance calculation feature vector similarity, and graph convolution neural network realizes fault part identification of the wind driven generator.
4. The GCN and WPT-MD based wind turbine generator fault location identification method as claimed in claim 1, wherein the graph convolution neural network is a novel intelligent algorithm, and the algorithm is fast in diagnosis speed, high in accuracy and good in development prospect.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011199637.XA CN112270273A (en) | 2020-10-30 | 2020-10-30 | Wind driven generator fault part identification method based on GCN and WPT-MD |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011199637.XA CN112270273A (en) | 2020-10-30 | 2020-10-30 | Wind driven generator fault part identification method based on GCN and WPT-MD |
Publications (1)
Publication Number | Publication Date |
---|---|
CN112270273A true CN112270273A (en) | 2021-01-26 |
Family
ID=74344760
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202011199637.XA Pending CN112270273A (en) | 2020-10-30 | 2020-10-30 | Wind driven generator fault part identification method based on GCN and WPT-MD |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112270273A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113419172A (en) * | 2021-06-21 | 2021-09-21 | 湘潭大学 | New energy automobile three-phase asynchronous motor fault identification based on GCN and VMD-ED |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20090237021A1 (en) * | 2008-03-21 | 2009-09-24 | Denso Corporation | Apparatus for carrying out improved control of rotary machine |
CN102944435A (en) * | 2012-10-25 | 2013-02-27 | 北京航空航天大学 | Health assessment and fault diagnosis method for rotating machinery based on fisher discriminant analysis and mahalanobis distance |
CN105300692A (en) * | 2015-08-07 | 2016-02-03 | 浙江工业大学 | Bearing fault diagnosis and prediction method based on extended Kalman filtering algorithm |
CN108196161A (en) * | 2018-01-10 | 2018-06-22 | 哈尔滨理工大学 | A kind of asynchronization super-pressure Synchronous Generator after Loss-of-Excitation method for diagnosing faults |
CN111461392A (en) * | 2020-01-23 | 2020-07-28 | 华中科技大学 | Power failure prediction method and system based on graph neural network |
CN111652329A (en) * | 2020-08-05 | 2020-09-11 | 腾讯科技(深圳)有限公司 | Image classification method and device, storage medium and electronic equipment |
CN111812507A (en) * | 2020-05-27 | 2020-10-23 | 浙江工业大学 | Motor fault diagnosis method based on graph convolution |
-
2020
- 2020-10-30 CN CN202011199637.XA patent/CN112270273A/en active Pending
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20090237021A1 (en) * | 2008-03-21 | 2009-09-24 | Denso Corporation | Apparatus for carrying out improved control of rotary machine |
CN102944435A (en) * | 2012-10-25 | 2013-02-27 | 北京航空航天大学 | Health assessment and fault diagnosis method for rotating machinery based on fisher discriminant analysis and mahalanobis distance |
CN105300692A (en) * | 2015-08-07 | 2016-02-03 | 浙江工业大学 | Bearing fault diagnosis and prediction method based on extended Kalman filtering algorithm |
CN108196161A (en) * | 2018-01-10 | 2018-06-22 | 哈尔滨理工大学 | A kind of asynchronization super-pressure Synchronous Generator after Loss-of-Excitation method for diagnosing faults |
CN111461392A (en) * | 2020-01-23 | 2020-07-28 | 华中科技大学 | Power failure prediction method and system based on graph neural network |
CN111812507A (en) * | 2020-05-27 | 2020-10-23 | 浙江工业大学 | Motor fault diagnosis method based on graph convolution |
CN111652329A (en) * | 2020-08-05 | 2020-09-11 | 腾讯科技(深圳)有限公司 | Image classification method and device, storage medium and electronic equipment |
Non-Patent Citations (2)
Title |
---|
DINGCHENG ZHANG等: "Intelligent Acoustic-based Fault Diagnosis of Roller Bearings Using a Deep Graph Convolutional Network", 《MEASUREMENT》 * |
宋才华等: "基于人工智能方法的电力系统动态等值的应用研究", 《微型电脑应用》 * |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113419172A (en) * | 2021-06-21 | 2021-09-21 | 湘潭大学 | New energy automobile three-phase asynchronous motor fault identification based on GCN and VMD-ED |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110609477B (en) | Electric power system transient stability discrimination system and method based on deep learning | |
CN113094862B (en) | Data-driven platform zone multilayer topological structure identification method | |
CN112818491A (en) | Wind power plant aggregation equivalent modeling method based on principal component analysis and clustering algorithm | |
CN107590604B (en) | Coherent unit grouping method and system combining S transformation and 2DPCA | |
CN113255458A (en) | Bearing fault diagnosis method based on multi-view associated feature learning | |
Mishra et al. | Loss of main detection in distribution generation system based on hybrid signal processing and machine learning technique | |
Sima et al. | Edge–cloud collaboration detection approach for small-sample imbalanced faults in power lines | |
CN115983374A (en) | Cable partial discharge database sample expansion method based on optimized SA-CACGAN | |
CN112270273A (en) | Wind driven generator fault part identification method based on GCN and WPT-MD | |
Hassani et al. | Design of a cost-effective deep convolutional neural network–based scheme for diagnosing faults in smart grids | |
Fu et al. | Fault classification in wind turbines using principal component analysis technique | |
Aljohani et al. | Smart fault detection and classification for distribution grid hybridizing ST and MLP-NN | |
CN112653136B (en) | Power electronic multi-feed-in power system key line identification method and system | |
Zhao et al. | A fault diagnosis approach for rolling element bearing based on S-transform and artificial neural network | |
CN110441081B (en) | Intelligent diagnosis method and intelligent diagnosis system for faults of rotating machinery | |
CN115221914A (en) | Rotating machinery fault diagnosis method based on enhanced octyl geometric modal decomposition | |
CN109031020B (en) | Transformer inrush current identification method based on logistic regression | |
Zhu et al. | Bearing fault diagnosis based on sparse wavelet decomposition and sparse graph connection using GraphSAGE | |
Senanayaka et al. | Autoencoders and data fusion based hybrid health indicator for detecting bearing and stator winding faults in electric motors | |
Rababaah | Wind turbine signal fault diagnosis using deep neural networks-inspired model | |
CN117171544B (en) | Motor vibration fault diagnosis method based on multichannel fusion convolutional neural network | |
Parija et al. | Faults classification In A Microgrid Using Decision Tree Technique And Support Vector Machine | |
CN114325236B (en) | Power distribution network fault identification method and system based on frequency spectrum entropy and random forest | |
Bhuyan et al. | Fault Classification in a DG Connected Power System using Artificial Neural Network | |
Haroun et al. | Self-Organizing Map and feature selection for of IM broken rotor bars faults detection and diagnosis |
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: 20210126 |