The content of the invention
The defects of for prior art, it is an object of the invention to provide it is a kind of independent of standard utensil based on higher-dimension with
The electronic mutual inductor error state appraisal procedure of machine matrix, it is ensured that electronic mutual inductor error state appraisal procedure can
Realisation, validity and accuracy.
The invention provides a kind of electronic mutual inductor error state appraisal procedure based on higher-dimension random matrix, including under
State step:
S1:The output valve of electronic mutual inductor is acquired based on time slip-window, and built according to the data of collection
Original matrix;
S2:Kalman filter based on single quantity of state is extended to the original matrix, obtains higher-dimension random matrix;
S3:The higher-dimension random matrix is standardized, make its be converted to row vector average be 0, variance be 1
Non- Hermite Matrix;
S4:Error state evaluating matrix is obtained according to the non-Hermite Matrix;
S5:The evaluation index of electric mutual inductor error state is obtained according to the error state evaluating matrix, and according to institute
State error state evaluating matrix and the evaluation index is assessed the error state of electric mutual inductor.
The present invention exports only according to electronic mutual inductor independent of standard utensil, judges this using higher-dimension random matrix
Whether a little parameters meet known rules, so as to which whether the statistical distribution of the measurement error to electronic mutual inductor has anomalous variation to do
Go out to judge, carry out error state assessment accordingly.
Further, in step sl, original matrix D is built1Based on time slip-window, in the state estimation of interception
Between in window, T times to N number of measured state parameter measurement of equipment, using measured state parameter number N as space dimensionality, measurement
Number T is as time dimension, and forms are with step-length TWIt is mobile.The original matrix of structure isWherein, xijValue of i-th of parameter in j measurement is represented, i is can
Survey state parameter sequence number, i=1,2 ... ... N, N be state parameter number, j be measurement sequence number, j=1,2 ... ... T, T
For pendulous frequency.
Further, in step s 2, based on Kalman filter, the square built according to electronic mutual inductor output valve
Battle array D1, using single state Kalman filter equation, Kalman filter output is obtained, using the output of wave filter as matrix
OK, with the dimension of extended matrix, matrix D is obtained2.The higher-dimension random matrix obtained after extensionN' be extension after state parameter number, N' value model
Enclose satisfaction
Further, in step s3, non-Hermite MatrixWhereinxi=(xi1,xi2,…,
xiT), 1≤i≤N' is matrix D2Row vector,Represent sample xiAverage value, σ (xij) represent sample xiStandard deviation.
Further, step S4 is specially:Calculate the singular value equivalent matrice D of the non-Hermite Matrixu;According to institute
State singular value equivalent matrice calculating matrix product Z;Error state evaluating matrix Z is obtained according to the matrix product Z2。
Further, the error state evaluating matrixWherein,
zi=(zi1,zi2,…,ziT), 1≤i≤N ',Represent error state evaluating matrix Z2Element,σ
(zi) represent ziStandard deviation.
Further, in step s 5, the evaluation index includes central moment αkWith moment of the orign βk;The central momentThe moment of the orignWherein, n is characterized value number, and E (V) is characterized value sample
It is expected.
Further, using 2 rank central moment α2With 1 rank moment of the orign β1Assessment as electronic mutual inductor error state
Criterion.
Further, after step s 5, also comprise the following steps:Judge to assess whether the time terminates, if so, then obtaining
Obtain electronic mutual inductor error state assessment result;If it is not, then t=t+TWIt is back to step S1;Wherein TWMoved for time slip-window
Dynamic step-length, t are the assessment time.
The present invention, can be real only according to the data of electronic mutual inductor itself output without using standard utensil at the scene
" self-evaluating " of existing error state, has been greatly reduced the cost assessed, and the O&M for being advantageous to improve electronic mutual inductor is horizontal.
The present invention need not establish any physical model, without assumed condition and simplified condition, when the parameter of power system changes
The assessment result of the present invention will not be had any impact.No matter undergoing mutation property of error state failure or occur gradually changeable therefore
Barrier, the present invention can make accurate and effective assessment according to the regularity of distribution of characteristic value.
Embodiment
In order to make the purpose , technical scheme and advantage of the present invention be clearer, it is right below in conjunction with drawings and Examples
The present invention is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, and
It is not used in the restriction present invention.
The present invention exports only according to electronic mutual inductor, judges whether these parameters meet using higher-dimension random matrix
Know rule, so as to which whether the statistical distribution of the measurement error to electronic mutual inductor has anomalous variation to judge, and error
Statistical distribution and error state are directly related, carry out error state assessment accordingly.
In the present invention, the electronic mutual inductor error state appraisal procedure based on higher-dimension random matrix, comprise the following steps:
Step 1:The on-line monitoring of electronic mutual inductor output valve, based on time slip-window, built using output valve original
Matrix;
Electronic mutual inductor output is the Serial No. of time coherence, and measured state parameter contains the width of fundamental wave harmonic
The information such as value, phase, frequency, wherein, harmonic wave highest number is 13 times.In order to realize the real-time assessment of state, one can be used
The method of kind time slip-window, i.e. previous moment time window terminate the initial time that the time is subsequent time time window, constantly
The output valve at current time is obtained from source data, builds original matrix, time slip-window length span is 1000~∞
(s), it is preferable that time slip-window length can be selected as 3600s.
In the state estimation time window of a certain interception, to N number of measured state parameter measurement of electronic mutual inductor output
T times, using measured state parameter number N as space dimensionality, pendulous frequency T is as time dimension, and forms are with step-length TWIt is mobile.
State parameter N span is 1~39, it is preferable that N can be selected as 1;Pendulous frequency T span is 100~∞,
Preferably, T can be selected as 360;In the time window of interception, all measurement data may be constructed original matrix D1:
Wherein, xijRepresent i-th of parameter in j
The value of secondary measurement, i be measured state parameter sequence number, i=1,2 ... ... N, j be measurement sequence number, j=1,2 ... ... T.
Step 2:Kalman filter based on single quantity of state is to original matrix D1It is extended, obtains higher-dimension random matrix
D2;
When electronic mutual inductor error state based on higher-dimension Random Matrices Theory is assessed, due to characterizing electronic mutual inductor
The measured state parameter of error state is less, it is impossible to directly using measurement data structure higher-dimension random matrix.In order to overcome this
Problem, employ the Matrix extension method of the Kalman filter equation based on single quantity of state.
Estimate the accurate measurements of measuring system:Wherein, xkFor current time system
State space amount, xk+1For subsequent time system state space amount, ykFor system measurement;ξkIt is 0 mean value model noise;ηkIt is 0
Average measurement noise.
Change measurement noise ηkValue, the output valve of multigroup Kalman filter, η can be obtainedkSpan can be
0.1Vrms~10Vrms, wherein, VrmsFor the virtual value of Kalman filter input signal.Using the output of Kalman filter as
Row matrix, state parameter are changed into N' from N number of, and the dimension of technology transform, N' span needs to ensureIt is preferred that
Ground, N' can be selected as 150, so as to build higher-dimension random matrix D2:
Step 3:Higher-dimension random matrix is standardized, make its be converted to row vector average be 0, variance be 1
Non- Hermite Matrix;
To matrix D2It is changed into non-Hermite Matrix D after carrying out following normalizing operationstd:
Whereinxi
=(xi1,xi2,…,xiT), 1≤i≤N' is matrix D2Row vector,Represent sample xiAverage value, σ (xij) represent sample
xiStandard deviation.
So that the matrix D after normalizing operationstd=(yij)N'×TMeetWherein, yi=
(yi1,yi2,…,yiT),1≤i≤N’。
Step 4:Calculated by the calculating of singular value equivalent matrice, matrix product and evaluating matrix calculates link, obtain error
State estimation matrix;
First, the singular value equivalent matrice D of non-Hermite Matrix is asked foru:
Wherein,Representing matrix DstdConjugation means, U is Haar unitary matrice.
Then, calculating matrix product Z:
Wherein, Du,iIndependent singular value equivalent matrice is represented, L span is 1~∞, it is preferable that L is taken as 1.
Finally, based on matrix product Z, error state evaluating matrix Z is asked for2:
Wherein, zi=(zi1,
zi2,…,ziT), 1≤i≤N ',Represent error state evaluating matrix Z2Element,σ(zi) represent zi
Standard deviation.
Step 5:The evaluation index of electric mutual inductor error state is obtained, evaluation index includes central moment αkWith moment of the orign βk;
And according to error state evaluating matrix Z2The error state of electric mutual inductor is assessed with evaluation index;
The state of assessment equipment is judged according to the higher-dimension random matrix of structure, it is necessary to which whether test matrix meets some property
Matter, checkout procedure need to rely on some specific statistics.For monocyclic theorem, square statistic characteristic feature value can be used
The regularity of distribution, error state evaluating matrix Z2Characteristic value may be constructed characteristic value sample V={ V1,V2,…Vn, the k ranks of the sample
Central moment αkFor:Wherein, n is characterized value number, and E (V) is characterized the value sample phase
Hope, αkAs evaluation index J1, equally can be with the k rank moment of the origns β of defined feature value samplekAs evaluation index J2:
Preferably, 2 rank central moment α can be used2With 1 rank moment of the orign β1Assessment as electronic mutual inductor error state
Criterion.
Specific appraisal procedure:When electronic mutual inductor error state is normal, i.e., its error is in standard GB/T 20840.7-
2007《The part of transformer the 7th:Electronic type voltage transformer》With standard GB/T 20840.7-2007《The part of transformer the 8th:Electricity
Type current transformer》During defined electronic mutual inductor error limit requirement, measured state parameter keeps substantially constant constant,
Only make stable random fluctuation near average is horizontal, conversely, when electronic mutual inductor error state exception, measured state ginseng
There are unusual fluctuations in the measured value of amount, is no longer steady random fluctuation.According to monocyclic theorem, when electronic mutual inductor is in normal
During state, the singular value equivalent matrice for the random matrix that its state parameter measured value is formed will meet monocyclic theorem, characteristic value meeting
It is evenly distributed in the ring with specific interior outer radius;If abnormal, the square containing abnormal data occurs for electronic mutual inductor state
The feature Distribution value of battle array can change, and distribution is no longer uniform.Using electronic mutual inductor output valve, extract contain survey shape
State property, higher-dimension random matrix is built, obtains error state evaluating matrix, it is final to identify real-time electronic according to evaluation index
Formula transformer error state.After the completion of real-time electronic formula transformer error state estimation step, judge to assess whether the time ties
Beam, if so, then obtaining electronic mutual inductor error state assessment result;Continued executing with if it is not, being then back to step 1, during assessment
Between span be 1~7 day, it is preferable that assess the time can select as 1 day.
To further understand the present invention, relative theory in the present invention is briefly illustrated below:
(1) electronic mutual inductor output parameter Statistical Distribution Characteristics
Electronic mutual inductor gathers power network primary voltage and current information, by sampled data can be calculated frequency,
Phase, amplitude, power factor, harmonic wave, tri-phase unbalance factor etc., and can also be obtained between these parameters by mathematical operation
New parameter, i.e., almost all of operation of power networks parameter can directly or indirectly be observed by electronic mutual inductor.These electricity
The rule of gas parameter is accurate Analysis expression or can to obey known distribution etc., for example, the changing rule of node phase can parse
ForWherein, ω0For operation of power networks frequency,For the first phase of node phase.According to the defeated of electronic mutual inductor
Go out value, it can be determined that whether the parameter meets known rules, so as to which the statistical distribution of the measurement error to electronic mutual inductor is
No anomalous variation judges, and the statistical distribution and error performance of error are directly related, carries out error state assessment accordingly, this is
It is the general principle independent of standard utensil and physical model assessment electronics formula transformer error state.
(2) monocyclic theorem
Assuming that random matrix A=(aij)n×nIt is n rank Haar unitary matrice that A=PTQ, wherein P and Q, which can be decomposed into, and T is pair
Angle battle array, diagonal entry are A singular values.When certain conditions are met, the limit Spectral structure of matrix A is by the general of its singular value
Rate, which is estimated, to be uniquely determined, and characteristic value converges to annulus on a complex plane, and the interior outer radius of annulus is respectively:Wherein, v is the probability measure of the singular value of matrix A, and this is referred to as monocyclic theorem.
In actual applications, matrix is consideredFor non-Hermite Matrix, and the row vector of matrix A meets
Average is 0, variance 1.For multiple non-Hermite Matrix Ai, define matrix product A2ForWherein, Au,iFor Ai
Singular value equivalent matrice.By matrix A2It is standardized as Astd, it is met σ2(ai)=1/N, wherein, aiFor matrix AstdRow to
Measure, then AstdLimit Spectral structure convergence in (with)probability be to probability density function: Spectral distribution function.In formula (12),p,
M→∞。AstdCharacteristic value in the distribution of complex plane be an annulus, inner ring radius is (1-c)L/2, outer shroud radius is 1.Setting
In the case of standby state is normal, matrix AstdMeet following property:Singular value equivalent matrice converts what is obtained by Haar unitary matrice
Standardization product matrix should meet monocyclic theorem.
The invention will be further described with specific embodiment below in conjunction with the accompanying drawings.Embodiment is exemplary, it is intended to is used
It is of the invention in explaining, and be not considered as limiting the invention.
As shown in figure 1, the present invention is assessed the error state of electronic mutual inductor according to following steps:
(1) electronic mutual inductor error state Evaluation Platform as shown in Figure 2 is built, platform includes:AC 220V power supplys 1,
Isolating transformer 2, auto-transformer 3, voltage transformer 4, data acquisition unit 5, pulse per second (PPS) equipment 6 and computer 7.AC 220V
Power supply, isolating transformer, auto-transformer are used to provide signals to voltage transformer, because the error of electronic mutual inductor is not easy
Change, the simulation electronic formula transformer by the way of voltage transformer+data acquisition unit, pulse per second (PPS) equipment carries to data acquisition unit
Sampled for pps pulse per second signal with triggering.Voltage transformer data are passed through by being sent to computer, computer after data acquisition unit
Ethernet interface captures electronic mutual inductor packet, and original random matrix D is built according to electronic current mutual inductor output1。
(2) Kalman filter is based on, to matrix D1Higher-dimension random matrix D is formed after being extended2.Original matrix and expansion
The scale for opening up matrix is as shown in table 1.Due to only calculating electronic mutual inductor phase, the line number of original matrix is 1, and electronic type is mutual
Sensor phase calculates 1 time per 10s, and time slip-window length is 1h, calculates 360 times altogether, and this is the columns of original matrix, utilizes
After Kalman filter extension, the extended matrix of one 150 × 360 is constituted.
The higher dimensional matrix scale of table 1
(3) using formula (4) to matrix D2It is standardized and is converted to matrix Dstd。
(4) evaluating matrix Z is asked for using formula (6)~formula (8)2。
(5) method by being blocked in a computer to electronic mutual inductor data bits, simulation electronic formula mutual inductance
The normal condition and abnormal conditions of device error state, Fig. 3 are the characteristic value regularity of distribution for not blocking evaluating matrix in the case of digit,
Fig. 4 is the characteristic value regularity of distribution for blocking evaluating matrix in the case of digit is 10.From Fig. 3 and Fig. 4 contrasts as can be seen that when electricity
When minor transformer is in normal condition, the feature Distribution value of evaluating matrix is concentrated in an annulus;And work as electronic type mutual inductance
When device is in abnormality, characteristic value distribution starts discretization, and partial feature value fall into annulus.
Evaluation index is calculated according to formula (9)~formula (10), result of calculation is as shown in table 2.As can be seen from Table 2, when
When electronic mutual inductor is in abnormality, evaluation index J1And J2Value significantly become big, can be effective therefore according to evaluation index
Judge the error state of electronic mutual inductor.
The evaluation index of table 2 contrasts
Evaluation index |
Normal condition |
Abnormality |
J1 |
7.9128 |
10.7242 |
J2 |
0.6841 |
1.1660 |
In order to as it will be easily appreciated by one skilled in the art that the foregoing is merely illustrative of the preferred embodiments of the present invention, not
To limit the present invention, all any modification, equivalent and improvement made within the spirit and principles of the invention etc. all should
Within protection scope of the present invention.