CN110045227B - power distribution network fault diagnosis method based on random matrix and deep learning - Google Patents
power distribution network fault diagnosis method based on random matrix and deep learning Download PDFInfo
- Publication number
- CN110045227B CN110045227B CN201910224664.9A CN201910224664A CN110045227B CN 110045227 B CN110045227 B CN 110045227B CN 201910224664 A CN201910224664 A CN 201910224664A CN 110045227 B CN110045227 B CN 110045227B
- Authority
- CN
- China
- Prior art keywords
- fault
- fault diagnosis
- power distribution
- distribution network
- random matrix
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/08—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
-
- 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
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Test And Diagnosis Of Digital Computers (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention discloses a power distribution network fault diagnosis method based on a random matrix and deep learning, which relates to the technical field of power distribution network fault diagnosis, wherein a fault high-dimensional data set is processed by introducing two basic tools, namely a random matrix theory and a deep learning technology, the random matrix theory has strict and flexible mathematical analysis capability in a high-dimensional space, the deep learning has excellent high-dimensional data modeling capability, high-dimensional characteristics of a fault are extracted through the random matrix theory and the deep learning technology, and a fault criterion is formed by adopting a multi-characteristic fusion technology according to the extracted fault high-dimensional characteristics; and constructing a fault diagnosis model according to the process, obtaining effective fault diagnosis information from the real-time data of the power distribution network through the fault diagnosis model, and diagnosing the fault in real time according to the effective fault diagnosis information, so that the accuracy and the intelligent degree of fault diagnosis of the power distribution network are improved.
Description
Technical Field
the invention belongs to the technical field of power distribution network fault diagnosis, and particularly relates to a power distribution network fault diagnosis method based on a random matrix and deep learning.
background
The power distribution network fault diagnosis aims at accurately and quickly detecting, identifying and positioning faults in the power distribution network, and is a key technology and a breakthrough point for improving the power supply reliability of the power distribution network. The core difficulty of the power distribution network fault diagnosis is that the fault criterion is difficult to design. For example, in a low-current grounding system, the fault current characteristics of single-phase grounding are not obvious, and the current mainstream grounding line selection and fault location method in the actual engineering still tries to select grounding lines one by one and then finds out fault points through manual line patrol.
disclosure of Invention
The invention aims to provide a power distribution network fault diagnosis method based on a random matrix and deep learning, so that the defect that the existing power distribution network fault diagnosis fault criterion is not obvious is overcome.
In order to achieve the purpose, the invention provides a power distribution network fault diagnosis method based on a random matrix and deep learning, which comprises the following steps:
S1, acquiring original electrical measurement data and a fault report which mainly comprise fault recording;
s2, cleaning, preprocessing, labeling and structuring the data obtained in the S1;
S3, establishing a standardized database with labels according to the data obtained in the S2;
S4, selecting a data set from the standardized database to establish a random matrix model, analyzing the random matrix model based on a random matrix theory analysis tool, and extracting high-dimensional statistical characteristics of the power distribution network faults from the random matrix model; analyzing the statistical properties of the high-dimensional statistical characteristics, and observing the high-dimensional statistical characteristics by combining historical fault data to preliminarily ensure that the statistical properties of the high-dimensional statistical characteristics meet the requirements for constructing fault diagnosis criteria;
S5, deep learning is carried out on the standardized database, and deep features of faults are extracted from the standardized database;
S6, constructing a fault diagnosis criterion by adopting a multi-feature fusion technology according to the high-dimensional statistical features extracted in the S4 and the deep features extracted in the S5;
s7, constructing a fault diagnosis model according to the fault diagnosis criterion, obtaining effective fault diagnosis information from the real-time data of the power distribution network through the fault diagnosis model, and diagnosing the fault in real time according to the effective fault diagnosis information.
Further, in S4, the statistical properties of the high-dimensional statistical features are analyzed by the stochastic matrix theory analysis tool.
further, the S5 includes the following steps:
S51, initializing deep learning, distributing a training set, a verification set and a test set according to the standardized database, configuring hyper-parameters of a deep neural network, training the deep neural network according to the training set, and verifying the performance of the deep neural network by using the verification set;
s52, testing the performance of the deep neural network obtained in the S51 by adopting the test set;
s53, repeating the steps S51-S52 until an ideal effect is obtained, and establishing a fault deep layer network model;
and S54, applying the fault deep layer network model obtained in the step S53 to the real-time data of the power distribution network, and extracting deep layer characteristics of the fault from the data.
Further, fitting and generalization are performed on the fault network model obtained in S54.
Further, the S4 further includes: and performing primary fault diagnosis by applying the high-dimensional statistical characteristic construction process of the power distribution network fault to the real-time data of the power distribution network.
Compared with the prior art, the invention has the following beneficial effects:
The invention provides a power distribution network fault diagnosis method based on a random matrix and deep learning, which is characterized in that a fault high-dimensional data set is processed by introducing two basic tools, namely a random matrix theory and a deep learning technology, wherein the random matrix theory has strict and flexible mathematical analysis capability in a high-dimensional space, the deep learning has excellent high-dimensional data modeling capability, high-dimensional features of a fault are extracted through the random matrix theory and the deep learning technology, and a fault criterion is formed by adopting a multi-feature fusion technology according to the extracted fault high-dimensional features; and constructing a fault diagnosis model according to the process, obtaining effective fault diagnosis information from the real-time data of the power distribution network through the fault diagnosis model, and diagnosing the fault in real time according to the effective fault diagnosis information, so that the accuracy and the intelligent degree of fault diagnosis of the power distribution network are improved.
Drawings
in order to more clearly illustrate the technical solution of the present invention, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are only one embodiment of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on the drawings without creative efforts.
FIG. 1 is a flow chart of a method for diagnosing faults of a power distribution network based on a random matrix and deep learning according to the invention;
Fig. 2 is a specific flowchart of a power distribution network fault diagnosis method based on a random matrix and deep learning according to the present invention.
Detailed Description
The technical solutions in the present invention are clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
As shown in fig. 1 and fig. 2, the method for diagnosing the fault of the power distribution network based on the random matrix and the deep learning provided by the invention comprises the following steps:
And S1, acquiring the original electrical measurement data and the fault report which mainly comprise fault recording.
And S2, cleaning, preprocessing, labeling and structuring the data obtained in the S1.
and S3, establishing a standardized database with labels according to the data obtained in the S2.
s4, selecting a data set from the standardized database to establish a random matrix model, analyzing the random matrix model based on a random matrix theory analysis tool, and extracting high-dimensional statistical characteristics of the power distribution network faults from the random matrix model; analyzing the statistical properties of the high-dimensional statistical characteristics by a random matrix theory analysis tool, and observing the high-dimensional statistical characteristics by combining historical fault data to preliminarily ensure that the statistical properties of the high-dimensional statistical characteristics meet the requirements for constructing fault diagnosis criteria; the method comprises the following steps of acting the high-dimensional statistical characteristic construction process of the power distribution network fault on the real-time data of the power distribution network to carry out preliminary fault diagnosis:
S41, modeling a standard database with labels by adopting a random matrix theory, time sequence analysis, a model, free probability and other tools to obtain a random matrix model for the operation of the power distribution network;
s42, analyzing the random matrix model based on a random matrix theory analysis tool, and extracting high-dimensional statistical characteristics of the power distribution network fault from the random matrix model; and analyzing the statistical properties of the high-dimensional statistical characteristics by using a random matrix theory analysis tool, and observing the high-dimensional statistical characteristics by combining historical fault data to preliminarily ensure that the statistical properties of the high-dimensional statistical characteristics meet the requirements for constructing fault diagnosis criteria. The correlation process of the specific high-dimensional statistical features is as follows:
s420, the high-dimensional eigenvalue statistic is a Linear Eigenvalue Statistic (LES), which is represented by the following formula:
In the formula (1), the reaction mixture is,λ is a characteristic value, N is a spatial dimension, for a continuous test function (testing function);
s421, the law of maxima of LES:converging on probability:
in the formula (2), ρ (λ) is a Probability Density Function (PDF) of a matrix eigenvalue;
s422, central limit of LES:
given a non-Hermitian NxT rectangular random matrix X, whose elements are Xij satisfies the standard normal independent iso-distribution (i.i.d.); m is a covariance matrix of xOrder test functionSatisfy the conditions of continuityN, T → ∞ and c ═ N/T ≦ 1, constructed according to formula (2)The value distribution converges on a gaussian random variable with a mean of 0 and a variance of:
In the formula (3), the reaction mixture is,is a 4 th order cumulative amount of x elements;θ、θ1、θ2Is an integral independent variable;
S423, calculating an expected mean value of the LES according to a law of majority equation (2), and calculating an expected variance of the LES according to a central limit theorem equation (3); deviation of the expected value from the variance (theoretical value) is ΔMean value＝O(N-1)，ΔVariance (variance)＝O(N-2).；
And S43, acting the high-dimensional statistical characteristic construction process of the power distribution network fault on the real-time data of the power distribution network, comparing the obtained characteristic value with the expectation and variance, and judging whether the expected and variance are out of the set range so as to obtain a preliminary fault diagnosis.
s5, selecting a data set from the standardized database for deep learning, and extracting fault deep features of the data set from the data set; the method specifically comprises the following steps:
s51, initializing deep learning, allocating a training set, a verification set and a test set according to the database of S3, configuring hyper-parameters of a deep neural network, training the network according to the training set, and verifying the performance of the network by using the verification set;
S52, testing the performance of the network obtained in the S51 by adopting a test set;
s53, repeating S51-S52 until an ideal effect is obtained, establishing a fault deep layer network model, evaluating the fitting and generalization performance of the fault deep layer network model, and further realizing model optimization;
And S54, applying the fault deep layer network model obtained in the S53 to the real-time fault data, and extracting fault deep layer characteristics from the real-time fault data.
s6, constructing a fault diagnosis criterion by adopting a multi-feature fusion technology according to the high-dimensional statistical features extracted in the S4 and the deep features extracted in the S5;
in the process of constructing the fault diagnosis model, the AdaBoost algorithm is utilized to realize multi-feature fusion of the fault, and the effect of fault attribute classification is improved, so that the accuracy of the final fault diagnosis criterion is improved.
The core idea of Adaboost is adaptive enhancement, and the adaptivity is represented by: the sample weight of the previous basic classifier is strengthened, so that the constructed next classifier focuses more on the processing of the wrong samples, and finally, a plurality of basic classifiers are integrated to form the strong classifier. Suppose there are N samples (x)i，yi) i 1, … N, the weight of the training data sample D1 is initialized first (w)11 w12 … w1N)，w1i1/N, (i ═ 1, … N), then the iterative process starts, for the mth iteration (m ═ 1, 2, …), the core process is as follows:
1) constructing an error functionwith emTraining a current classifier for the minimum target;
2) calculate the Current classifier Gmimportance of (-) by:
3) Updating weights
s7, constructing a fault diagnosis model according to the fault diagnosis criterion, obtaining effective fault diagnosis information from the real-time data (including data collected by a power distribution network sensor and the like) of the power distribution network through the fault diagnosis model, and then diagnosing the fault in real time according to the effective fault diagnosis information.
in summary, the invention provides a power distribution network fault diagnosis method based on a random matrix and deep learning, which applies high-dimensional statistics and deep learning to distribution automation, realizes self-adaptive value extraction of high-dimensional characteristics (high-dimensional statistics characteristics and deep characteristics) of power distribution network faults, adopts a multi-characteristic fusion technology to form fault criteria according to the extracted high-dimensional characteristics of the faults, further constructs a fault diagnosis model, can obtain effective fault diagnosis information from real-time data of a power distribution network through the fault diagnosis model, and carries out real-time fault diagnosis according to the effective fault diagnosis information, thereby improving the accuracy and the intelligent degree of power distribution network fault diagnosis.
The above disclosure is only for the specific embodiments of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily conceive of changes or modifications within the technical scope of the present invention, and shall be covered by the scope of the present invention.
Claims (4)
1. A power distribution network fault diagnosis method based on random matrix and deep learning is characterized in that: the method comprises the following steps:
s1, acquiring original electrical measurement data and a fault report which mainly comprise fault recording;
s2, cleaning, preprocessing, labeling and structuring the data obtained in the S1;
s3, establishing a standardized database with labels according to the data obtained in the S2;
s4, selecting a data set from the standardized database to establish a random matrix model, analyzing the random matrix model based on a random matrix theory analysis tool, and extracting high-dimensional statistical characteristics of the power distribution network faults from the random matrix model; analyzing the statistical properties of the high-dimensional statistical characteristics, and observing the high-dimensional statistical characteristics by combining historical fault data to preliminarily ensure that the statistical properties of the high-dimensional statistical characteristics meet the requirements for constructing fault diagnosis criteria;
s5, deep learning is carried out on the standardized database, and deep features of faults are extracted from the standardized database;
S6, constructing a fault diagnosis criterion by adopting a multi-feature fusion technology according to the high-dimensional statistical features extracted in the S4 and the deep features extracted in the S5;
S7, constructing a fault diagnosis model according to the fault diagnosis criterion, obtaining effective fault diagnosis information from the real-time data of the power distribution network through the fault diagnosis model, and diagnosing the fault in real time according to the effective fault diagnosis information.
2. the power distribution network fault diagnosis method based on the random matrix and the deep learning of claim 1, characterized in that: the S5 includes the steps of:
s51, initializing deep learning, distributing a training set, a verification set and a test set according to the standardized database, configuring hyper-parameters of a deep neural network, training the deep neural network according to the training set, and verifying the performance of the deep neural network by using the verification set;
S52, testing the performance of the deep neural network obtained in the S51 by adopting the test set;
S53, repeating the steps S51-S52 until an ideal effect is obtained, and establishing a fault deep layer network model;
And S54, applying the fault deep layer network model obtained in the step S53 to the real-time data of the power distribution network, and extracting deep layer characteristics of the fault from the data.
3. the power distribution network fault diagnosis method based on the random matrix and the deep learning of claim 2 is characterized in that: and fitting and generalizing the fault network model obtained in the step S54.
4. the power distribution network fault diagnosis method based on the random matrix and the deep learning of claim 1, characterized in that: the S4 further includes: and performing primary fault diagnosis by applying the high-dimensional statistical characteristic construction process of the power distribution network fault to the real-time data of the power distribution network.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910224664.9A CN110045227B (en) | 2019-03-23 | 2019-03-23 | power distribution network fault diagnosis method based on random matrix and deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910224664.9A CN110045227B (en) | 2019-03-23 | 2019-03-23 | power distribution network fault diagnosis method based on random matrix and deep learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110045227A CN110045227A (en) | 2019-07-23 |
| CN110045227B true CN110045227B (en) | 2019-12-17 |
Family
ID=67275042
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910224664.9A Active CN110045227B (en) | 2019-03-23 | 2019-03-23 | power distribution network fault diagnosis method based on random matrix and deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110045227B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110673060B (en) * | 2019-09-25 | 2021-07-06 | 山东大学 | Power distribution network fault diagnosis method based on synchronous phasor measurement and random matrix theory |
| CN111292200A (en) * | 2020-01-17 | 2020-06-16 | 国网浙江嘉善县供电有限公司 | Distribution network typical fault studying and judging method based on operation data |
| CN113783710B (en) * | 2020-06-09 | 2024-01-02 | 南京南瑞继保电气有限公司 | Process layer network fault positioning method and device based on self-learning criteria |
| CN112365759A (en) * | 2020-07-06 | 2021-02-12 | 贵州电网有限责任公司 | High-voltage power equipment test training system based on virtual reality |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106022529A (en) * | 2016-05-26 | 2016-10-12 | 中国电力科学研究院 | Abnormal data detection method for power distribution network based on high dimensional random matrix |
| CN106093703A (en) * | 2016-06-07 | 2016-11-09 | 湖南大学 | The identification of a kind of intelligent distribution network fault and localization method |
| EP3181637A1 (en) * | 2011-09-21 | 2017-06-21 | Borealis AG | Moulding composition |
| CN108196165A (en) * | 2018-01-09 | 2018-06-22 | 贵州大学 | Power grid abnormal state detection method based on sample covariance matrix maximum eigenvalue |
| CN108828405A (en) * | 2018-06-06 | 2018-11-16 | 西南交通大学 | A kind of electric transmission line fault detection method based on random matrix |
| CN108898028A (en) * | 2018-07-06 | 2018-11-27 | 成都大象分形智能科技有限公司 | It is related to the neural network model encryption protection system and method for iteration and accidental enciphering |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109061387A (en) * | 2018-08-23 | 2018-12-21 | 国网内蒙古东部电力有限公司电力科学研究院 | A kind of power grid abnormality method of discrimination based on annulus theorem |
| CN110208647B (en) * | 2019-05-27 | 2021-07-06 | 国网上海市电力公司 | Power distribution network single-phase earth fault section positioning method based on random matrix |
-
2019
- 2019-03-23 CN CN201910224664.9A patent/CN110045227B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP3181637A1 (en) * | 2011-09-21 | 2017-06-21 | Borealis AG | Moulding composition |
| CN106022529A (en) * | 2016-05-26 | 2016-10-12 | 中国电力科学研究院 | Abnormal data detection method for power distribution network based on high dimensional random matrix |
| CN106093703A (en) * | 2016-06-07 | 2016-11-09 | 湖南大学 | The identification of a kind of intelligent distribution network fault and localization method |
| CN108196165A (en) * | 2018-01-09 | 2018-06-22 | 贵州大学 | Power grid abnormal state detection method based on sample covariance matrix maximum eigenvalue |
| CN108828405A (en) * | 2018-06-06 | 2018-11-16 | 西南交通大学 | A kind of electric transmission line fault detection method based on random matrix |
| CN108898028A (en) * | 2018-07-06 | 2018-11-27 | 成都大象分形智能科技有限公司 | It is related to the neural network model encryption protection system and method for iteration and accidental enciphering |
Non-Patent Citations (1)
| Title |
|---|
| 随机矩阵理论在配电网早期故障检测中的;郑贤舜等;《电气应用》;20190315;第38卷(第3期);第58-63页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110045227A (en) | 2019-07-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110045227B (en) | power distribution network fault diagnosis method based on random matrix and deep learning | |
| Yuan et al. | Detection of spatial defect patterns generated in semiconductor fabrication processes | |
| CN107590506A (en) | A kind of complex device method for diagnosing faults of feature based processing | |
| Hwang et al. | Variational deep clustering of wafer map patterns | |
| WO2023273249A1 (en) | Tsvm-model-based abnormality detection method for automatic verification system of smart electricity meter | |
| CN111126820A (en) | Electricity stealing prevention method and system | |
| CN111833175A (en) | Internet financial platform application fraud behavior detection method based on KNN algorithm | |
| CN114266289A (en) | Complex equipment health state assessment method | |
| CN115293577A (en) | Method for analyzing chemical control factors of underground water in alpine-cold flow region based on machine learning | |
| Li et al. | GMM-HMM-based medium-and long-term multi-wind farm correlated power output time series generation method | |
| CN112183459B (en) | Remote sensing water quality image classification method based on evolution multi-objective optimization | |
| CN114200245A (en) | Construction method of line loss abnormity identification model of power distribution network | |
| CN111090679B (en) | Time sequence data representation learning method based on time sequence influence and graph embedding | |
| CN117219124A (en) | Switch cabinet voiceprint fault detection method based on deep neural network | |
| CN110673577B (en) | Distributed monitoring and fault diagnosis method for complex chemical production process | |
| CN110244690B (en) | Multivariable industrial process fault identification method and system | |
| CN111737993A (en) | Method for extracting health state of equipment from fault defect text of power distribution network equipment | |
| CN115452957B (en) | Small sample metal damage identification method based on attention prototype network | |
| CN116630989A (en) | Visual fault detection method and system for intelligent ammeter, electronic equipment and storage medium | |
| CN116431988A (en) | Resident trip activity time sequence generation method based on activity mode-Markov chain | |
| CN115308705A (en) | Multi-pose extremely narrow pulse echo generation method based on generation countermeasure network | |
| CN115600753A (en) | Hybrid life prediction method for complex electromechanical equipment | |
| CN104573727A (en) | Dimension reduction method of handwritten digital image | |
| Nair et al. | Maximum likelihood uncertainty estimation: Robustness to outliers | |
| Maume-Deschamps et al. | Detecting the stationarity of spatial dependence structure using spectral clustering |
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 |