CN109784247A - A kind of low-frequency oscillation parameter identification method based on Random Decrement and blind source separating - Google Patents
A kind of low-frequency oscillation parameter identification method based on Random Decrement and blind source separating Download PDFInfo
- Publication number
- CN109784247A CN109784247A CN201910001443.5A CN201910001443A CN109784247A CN 109784247 A CN109784247 A CN 109784247A CN 201910001443 A CN201910001443 A CN 201910001443A CN 109784247 A CN109784247 A CN 109784247A
- Authority
- CN
- China
- Prior art keywords
- signal
- matrix
- low
- blind source
- oscillation
- 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
Landscapes
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
- Complex Calculations (AREA)
Abstract
The invention discloses a kind of low-frequency oscillation parameter identification method based on Random Decrement and blind source separating obtains free damping signal comprising steps of 1) handling using Random Decrement Technique the noise like electric power signal under collected environmental excitation；2) blind source separation algorithm is used to free damping signal, decomposites different single mode signals；3) Hilbert transform is carried out to different single mode signals, solves frequency of oscillation and attenuation coefficient, and use recognizer, determines optimal model results.The present invention breaks through existing recognition methods and realizes the parameter of the low-frequency oscillation in advance identification of efficiently and accurately using blind source separating and Random Decrement algorithm based on the subsequent identification disadvantage under obvious disturbance.
Description
Technical field
The present invention relates to the technical fields of pattern-recognition and low-frequency oscillation, refer in particular to a kind of based on Random Decrement and blind source
Isolated low-frequency oscillation parameter identification method.
Background technique
With " transferring electricity from the west to the east ", " national network " strategy development, the electric system scale in China constantly expands, forms
Each department interconnection, hands over the ultra-large power grid of straight mixed connection.The interconnection of power grid is conducive to distributing rationally for every resource, but simultaneously
The problem of stability of power system can be brought, wherein low-frequency oscillation problem becomes increasingly conspicuous.Accurately and rapidly identify low-frequency oscillation
Parameter is to guarantee one of the important foundation of power system stability to implement effective braking measure.
Currently, the work of low-frequency oscillation parameter identification is mainly based upon obvious disturbance (such as short circuit, load is widely varied)
When system response signal carry out.The deficiency of such methods is: 1, occurring obviously to disturb in electric system actual motion
Probability very little is not construed as manipulating, and data volume is limited.2, such methods are simply possible to use in subsequent adjusting, can not be in advance to low frequency
Parameter of oscillation is identified.
Based on this, propose a kind of low-frequency oscillation parameter identification method for noise-like signal, can not occur it is low
Frequency vibration identifies low-frequency oscillation parameter in the case where swinging.In order to solve the problems, such as that noise-like signal convergent response is extracted, use certainly
Right exciting technique reliably, rapidly extracts free damping response signal；Free damping signal is handled with blind source separation algorithm,
Decomposite different single mode signals；Hilbert transform is carried out to different single mode signals, frequency of oscillation is solved and declines
Subtract coefficient, and use recognizer, determines optimal model results；This method can rapidly and accurately identify that low-frequency oscillation is joined
Number is that the control strategy of power oscillation damping improves effective information.
Summary of the invention
The purpose of the present invention is to overcome the shortcomings of the existing technology and deficiency, proposes a kind of based on Random Decrement and blind source
Isolated low-frequency oscillation parameter identification method is broken through existing recognition methods based on the subsequent identification disadvantage under obvious disturbance, is utilized
Blind source separating and Random Decrement algorithm realize the parameter of the low-frequency oscillation in advance identification of efficiently and accurately.
To achieve the above object, technical solution provided by the present invention are as follows: a kind of based on Random Decrement and blind source separating
Low-frequency oscillation parameter identification method, comprising the following steps:
1) the noise like electric power signal under collected environmental excitation is handled using Random Decrement Technique, is obtained certainly
By deamplification；
2) blind source separation algorithm is used to free damping signal, decomposites different single mode signals；
3) Hilbert transform is carried out to different single mode signals, solves frequency of oscillation and attenuation coefficient, and use
Recognizer determines optimal model results.
In step 1), using Random Decrement Technique to the noise like electric power signal under collected environmental excitation at
Reason, obtains free damping signal, comprising the following steps:
1.1) activation threshold value is chosen, take -0.6 times of noise-like signal standard deviation, the trigger condition of use are as follows:
Y > a and
Wherein, y is signal value, and a is threshold value；
1.2) length of window of every section of intercept signal is determined, sampling number is 6 sampling cycle lengths；
1.3) N is accumulated and divided by from N number of signal that each identical initial conditions are truncated to, obtains free damping signal.
In step 2), blind source separation algorithm is used to free damping signal, decomposites different single mode signals, including
Following steps:
2.1) the covariance matrix R of observation signal matrix X (t) is first found outX(t), and to it Eigenvalues Decomposition R is carried outX(t)
=EDET, wherein E and D respectively indicates eigenvectors matrix and eigenvalue matrix, and T represents matrix transposition；
2.2) pre -whitening processing is carried out to observation signal matrix X (t), obtains whitening matrix Z (t)=D(-1/2)ETX (t)=
WmX (t), and then know whitening matrix Wm=D(-1/2)ET；
2.3) one group of time delay covariance matrix R of prewhitening matrix Z (t) is soughtZ(τ), τ represent time delay covariance matrix
Delay parameter；
2.4) this group of time delay covariance matrix R of joint approximate diagonalization technology simultaneous diagonalization is utilizedZ(τ), is acquired just
Hand over normalization matrixTo obtain solving mixed matrix
In step 3), Hilbert transform is carried out to different single mode signals, solves frequency of oscillation and decaying system
Number, and iterative program and recognizer are used, determine optimal model results, comprising the following steps:
3.1) Hilbert transform is carried out to single mode signal:
Wherein, y indicates that original signal, H () indicate Hilbert transform, and i indicates i-th of low frequency oscillation mode, fiIt (t) is the
The instantaneous frequency of i oscillation mode, AmpiIt (t) is the attenuation coefficient of i-th of oscillation mode；
3.2) to instantaneous frequency fi(t) it averages, the average oscillation frequency of i-th of low frequency oscillation mode can be obtained；
To (Ampi(t)) time graph of imaginary part, i.e. ln (Ampi(t)) curve of-t carries out the fitting of linear polynomial, obtained fitting
The slope of straight line represents the mean attenuation coefficient of i-th of low frequency oscillation mode；
3.3) calculating parameter assessment errors dav(m-1, j), j are recognition result number under every kind of mode, and m is different rank,
Find out davThe corresponding line number m of minimum value in each column, m value is best model order, it is notable that works as davCertain
When NaN element in one column is greater than 3, which is not considered as Critical inertial modes, and NaN is each pattern in different rank
The number of lower appearance.
Compared with prior art, the present invention have the following advantages that with the utility model has the advantages that
1, simple using Random Decrement principle, processing noise-like signal ability is strong invention introduces Random Decrement Technique
The characteristics of, accurate rapidly extracting free damping response signal is realized, the influence of noise is eliminated.
2, the present invention realizes the identification of parameter in advance of low-frequency oscillation, strategy can be inhibited to be provided with for electricity grid oscillating
The reference information of effect.
3, the present invention is directed to the disadvantage of traditional mode recognition methods noiseproof feature difference, blind source separation algorithm is sampled, noisy
In the case of, it also can fast and accurately identification parameter.
4, the shortcomings that present invention is difficult, there are spurious patterns for traditional low-frequency parameter of oscillation recognition methods model order,
Using recognizer, solve the problems, such as that dominant pattern determines.
5, the method for the present invention has an extensive use space in the identification of low-frequency oscillation parameter, and recognition speed is fast, noise immunity
Can be good, there are bright prospects in pattern-recognition.
Detailed description of the invention
Fig. 1 is logical flow diagram of the present invention.
Fig. 2 is noise-like signal figure.
Fig. 3 is Random Decrement Technique treated free damping signal graph.
Fig. 4 is blind source separation algorithm flow chart.
Specific embodiment
The present invention is further explained in the light of specific embodiments.
As shown in Figures 1 to 4, known provided by the present embodiment based on Random Decrement and the low-frequency oscillation parameter of blind source separating
Other method, comprising the following steps:
1) the noise like electric power signal under collected environmental excitation is handled using Random Decrement Technique, is obtained certainly
By deamplification, the specific steps of which are as follows:
1.1) activation threshold value is chosen, -0.6 times of noise-like signal standard deviation, the trigger condition of use are taken are as follows:
Y > a and
Wherein, y is signal value, and a is threshold value；
1.2) length of window of every section of intercept signal is determined, sampling number is 6 sampling cycle lengths；
1.3) N is accumulated and divided by from N number of signal that each identical initial conditions are truncated to, obtains free damping signal.
2) blind source separation algorithm is used to free damping signal, decomposites different single mode signals, specific steps are such as
Under:
2.1) the covariance matrix R of observation signal matrix X (t) is first found outX(t), and to it Eigenvalues Decomposition R is carried outX(t)
=EDET, wherein E and D respectively indicates eigenvectors matrix and eigenvalue matrix, and T represents matrix transposition；
2.2) pre -whitening processing is carried out to observation signal matrix X (t), obtains whitening matrix Z (t)=D(-1/2)ETX (t)=
WmX (t), and then know whitening matrix Wm=D(-1/2)ET；
2.3) one group of time delay covariance matrix R of prewhitening matrix Z (t) is soughtZ(τ), τ represent time delay covariance matrix
Delay parameter；
2.4) this group of time delay covariance matrix R of joint approximate diagonalization technology simultaneous diagonalization is utilizedZ(τ), is acquired just
Hand over normalization matrixTo obtain solving mixed matrix
3) Hilbert transform is carried out to different single mode signals, solves frequency of oscillation and attenuation coefficient, and use
Recognizer determines optimal model results, the specific steps of which are as follows:
3.1) Hilbert transform is carried out to single mode signal:
Wherein, y indicates that original signal, H () indicate Hilbert transform, and i indicates i-th of low frequency oscillation mode, fiIt (t) is the
The instantaneous frequency of i oscillation mode, AmpiIt (t) is the attenuation coefficient of i-th of oscillation mode；
3.2) to instantaneous frequency fi(t) it averages, the average oscillation frequency of i-th of low frequency oscillation mode can be obtained；
To (Ampi(t)) time graph of imaginary part, i.e. ln (Ampi(t)) curve of-t carries out the fitting of linear polynomial, obtained fitting
The slope of straight line represents the mean attenuation coefficient of i-th of low frequency oscillation mode；
3.3) calculating parameter assessment errors dav(m-1, j), j are recognition result number under every kind of mode, and m is different rank,
Find out davThe corresponding line number m of minimum value in each column, m value is best model order, it is notable that works as davCertain
When NaN element in one column is greater than 3, which is not considered as Critical inertial modes.
Embodiment described above is only the preferred embodiments of the invention, and but not intended to limit the scope of the present invention, therefore
All shapes according to the present invention change made by principle, should all be included within the scope of protection of the present invention.
Claims (4)
1. a kind of low-frequency oscillation parameter identification method based on Random Decrement and blind source separating, which is characterized in that including following step
It is rapid:
1) the noise like electric power signal under collected environmental excitation is handled using Random Decrement Technique, obtains freely declining
Cut signal；
2) blind source separation algorithm is used to free damping signal, decomposites different single mode signals；
3) Hilbert transform is carried out to different single mode signals, solves frequency of oscillation and attenuation coefficient, and using identification
Program determines optimal model results.
2. a kind of low-frequency oscillation parameter identification method based on Random Decrement and blind source separating according to claim 1,
It is characterized in that: in step 1), the noise like electric power signal under collected environmental excitation being carried out using Random Decrement Technique
Processing, obtains free damping signal, comprising the following steps:
1.1) activation threshold value is chosen, -0.6 times of noise-like signal standard deviation, the trigger condition of use are taken are as follows:
Y > a and
Wherein, y is signal value, and a is threshold value；
1.2) length of window of every section of intercept signal is determined, sampling number is 6 sampling cycle lengths；
1.3) N is accumulated and divided by from N number of signal that each identical initial conditions are truncated to, obtains free damping signal.
3. a kind of low-frequency oscillation parameter identification method based on Random Decrement and blind source separating according to claim 1,
It is characterized in that: in step 2), blind source separation algorithm being used to free damping signal, decomposites different single mode signals, is wrapped
Include following steps:
2.1) the covariance matrix R of observation signal matrix X (t) is first found outX(t), and to it Eigenvalues Decomposition R is carried outX(t)=
EDET, wherein E and D respectively indicates eigenvectors matrix and eigenvalue matrix, and T represents matrix transposition；
2.2) pre -whitening processing is carried out to observation signal matrix X (t), obtains whitening matrix Z (t)=D(-1/2)ETX (t)=WmX
(t), and then whitening matrix W is knownm=D(-1/2)ET；
2.3) one group of time delay covariance matrix R of prewhitening matrix Z (t) is soughtZ(τ), τ represent the time delay of time delay covariance matrix
Parameter；
2.4) this group of time delay covariance matrix R of joint approximate diagonalization technology simultaneous diagonalization is utilizedZ(τ) acquires orthogonal normalizing
Change matrixTo obtain solving mixed matrix
4. a kind of low-frequency oscillation parameter identification method based on Random Decrement and blind source separating according to claim 1,
It is characterized in that: in step 3), Hilbert transform being carried out to different single mode signals, solves frequency of oscillation and decaying system
Number, and iterative program and recognizer are used, determine optimal model results, comprising the following steps:
3.1) Hilbert transform is carried out to single mode signal:
Wherein, y indicates that original signal, H () indicate Hilbert transform, and i indicates i-th of low frequency oscillation mode, fi(t) it is i-th
The instantaneous frequency of oscillation mode, AmpiIt (t) is the attenuation coefficient of i-th of oscillation mode；
3.2) to instantaneous frequency fi(t) it averages, the average oscillation frequency of i-th of low frequency oscillation mode can be obtained；It is right
(Ampi(t)) time graph of imaginary part, i.e. ln (Ampi(t)) curve of-t carries out the fitting of linear polynomial, and obtained fitting is straight
The slope of line represents the mean attenuation coefficient of i-th of low frequency oscillation mode；
3.3) calculating parameter assessment errors dav(m-1, j), j are recognition result number under every kind of mode, and m is different rank, is found out
davThe corresponding line number m of minimum value in each column, m value is best model order, it is notable that works as davA certain column
In NaN element when being greater than 3, which is not considered as Critical inertial modes, and NaN is that each pattern goes out under different rank
Existing number.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910001443.5A CN109784247A (en) | 2019-01-02 | 2019-01-02 | A kind of low-frequency oscillation parameter identification method based on Random Decrement and blind source separating |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910001443.5A CN109784247A (en) | 2019-01-02 | 2019-01-02 | A kind of low-frequency oscillation parameter identification method based on Random Decrement and blind source separating |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109784247A true CN109784247A (en) | 2019-05-21 |
Family
ID=66499704
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910001443.5A Pending CN109784247A (en) | 2019-01-02 | 2019-01-02 | A kind of low-frequency oscillation parameter identification method based on Random Decrement and blind source separating |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109784247A (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103198184A (en) * | 2013-03-27 | 2013-07-10 | 深圳大学 | Low-frequency oscillation character noise-like identification method in electric power system |
| CN106202977A (en) * | 2016-08-17 | 2016-12-07 | 华南理工大学 | A kind of low frequency oscillation mode based on blind source separation algorithm analyzes method |
| US20170353189A1 (en) * | 2016-06-06 | 2017-12-07 | Richwave Technology Corp. | Subsampling Motion Detector for Detecting Motion of Object Under Measurement |
-
2019
- 2019-01-02 CN CN201910001443.5A patent/CN109784247A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103198184A (en) * | 2013-03-27 | 2013-07-10 | 深圳大学 | Low-frequency oscillation character noise-like identification method in electric power system |
| US20170353189A1 (en) * | 2016-06-06 | 2017-12-07 | Richwave Technology Corp. | Subsampling Motion Detector for Detecting Motion of Object Under Measurement |
| CN106202977A (en) * | 2016-08-17 | 2016-12-07 | 华南理工大学 | A kind of low frequency oscillation mode based on blind source separation algorithm analyzes method |
Non-Patent Citations (1)
| Title |
|---|
| 张安琪: ""基于盲源分离算法的电力系统低频振荡模式分析"", 《中国优秀硕士学位论文全文数据库 工程科技II辑》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110852201B (en) | Pulse signal detection method based on multi-pulse envelope spectrum matching | |
| CN102760444B (en) | Support vector machine based classification method of base-band time-domain voice-frequency signal | |
| WO2019161592A1 (en) | Method for automatically extracting structural modal parameters by clustering | |
| CN103837891A (en) | High-precision pickup method of first arrival of microseism | |
| CN103487788B (en) | The fast automatic extracting method of a kind of train pulse signal | |
| CN103968933A (en) | Fuzzy domain characteristics based optical fiber vibration signal identifying method | |
| CN110110738A (en) | A kind of Recognition Method of Radar Emitters based on multi-feature fusion | |
| CN110134976A (en) | A kind of airborne laser sounding method for extracting signal and system | |
| CN103616661A (en) | Robust far-field narrowband signal source number estimation method | |
| CN105652325A (en) | Exponential fit-adaptive Kalman-based ground-air electromagnetic data de-noising method | |
| US20190376874A1 (en) | A method of estimating the number of modes for the sparse component analysis based modal identification | |
| CN109342828A (en) | A kind of lightening pulse signal detecting method based on frequency domain constant false alarm | |
| CN106448694B (en) | A kind of deficient time-frequency list source point extracting method determined in blind source separating based on the detection of multiple angle | |
| CN109506767A (en) | A kind of real-time detection method causing sound field exception to underwater intrusion target | |
| CN112688324B (en) | Power system low-frequency oscillation mode identification method based on FastICA and TLS-ESPRIT | |
| CN105184264B (en) | Radio-frequency fingerprint feature extracting method based on Wigner diagonal slice spectrum | |
| CN105743756A (en) | Frame detection method based on Adaboost algorithm in Wi-Fi system | |
| Zhang et al. | Modified subspace method based on convex model for underdetermined blind speech separation | |
| CN109784247A (en) | A kind of low-frequency oscillation parameter identification method based on Random Decrement and blind source separating | |
| CN117169346A (en) | High-altitude building damage identification method based on wavelet packet energy spectrum analysis | |
| CN110133680A (en) | A kind of airborne laser sounding receives waveform useful signal Determination of Initial and system | |
| CN115826042B (en) | Edge cloud combined distributed seismic data processing method and device | |
| CN109580146A (en) | A kind of Vibration Parameters recognition methods based on improvement Sparse Component Analysis | |
| CN104316929B (en) | Power spectrum identification method and apparatus | |
| CN110880012B (en) | Inter-pulse agile radar radiation source frequency information correlation method for multi-reconnaissance platform |
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 |