Summary of the invention
It is an object of the invention to provide a kind of intrasonic harmonic detection method considering that negative frequency affects, its feature exists
In, comprise the following steps:
Step one: under operation of power networks normal condition, sets up the unifrequency discrete signal model of power network signal:Wherein AiFor the amplitude of power network signal, fiFor power network signal frequency andFor power network signal
Initial phase, n ∈ [0, N-1], N be intercept the total sampling number of discrete signal, fsFor power network signal sample frequency;
Step 2: x (n) is added normalization discrete Fourier transform (DFT) (the Discrete Fourier of rectangular window
Transform, DFT) obtainWherein k=0,1 ... N-1;
Step 3: set the frequency resolution of spectral line as Δ f, Δ f=fs/ N, and will formula (1) substitution formula (2) obtain
If normalized frequency is fi 1, fi 1=fi/ Δ f, is further simplified as formula (3): In formula
Step 3: take adjacent three spectral line km-1、kmAnd kmSpectrum X (the k of+1m-1)、X(km) and X (km+ 1), k herem≥
2 for integer and meet the spectral line of its correspondence for maximum spectral line, obtain through calculating abbreviation
H[fi 1-(km+ 1)]=H [fi 1-(km-1)]=H (fi 1-km) (5)
NoteHave according to formula (4):
(fi 1-km-p)-1H++(fi 1+km+p)-1H--X(km+ p)=0 (6), wherein p=-1,0,1, three the unknowns in formula
Number is H+、H-And fi 1, when p=-1,0,1, a ternary linear function group can be become with simultaneous, therefrom can obtain fi 1、H+And H-,
Wherein solve frequency normalization f to be askedi 1Expression formula be
Step 4: obtained the highest spectral line k by formula (4)mEquation:
In formulaSinc (x)=
Sin (π x)/(π x), takes formula (8) and is conjugated:Obtain fiAfter, simultaneous formula (8) and
Formula (9) composition equation group, only A inside equation groupiWithBeing unknown quantity, remaining is known quantity, therefore can solve:Step 5: formula (7) and formula (10) be consider negative frequency and interpolation ultralow between
The frequency of harmonic wave, amplitude and phase estimation formula.
Detection in full frequency band time harmonic frequency comprises the following steps:
1) signal collected by electrical network, intercepts one section with rectangular window and makees FFT (DFT) spectrum analysis;
2) for the m-Acetyl chlorophosphonazo component of more than 50Hz, regulation m-Acetyl chlorophosphonazo detected according to IEC61000-4-7 standard, with
M-Acetyl chlorophosphonazo group or the form of subgroup, measure between two harmonic waves and the m-Acetyl chlorophosphonazo parameter of frequency range between first-harmonic and second harmonic, and
Using the meansigma methods of two harmonic wave sequence numbers as corresponding m-Acetyl chlorophosphonazo group or the frequency of subgroup;
3) m-Acetyl chlorophosphonazo for 0~50Hz frequency range must take into the impact of negative frequency;Between the intrasonic of 0~10Hz frequency range humorous
Ripple calculates intrasonic m-Acetyl chlorophosphonazo parameter amplitude A according to intrasonic simple harmonic wave detection methodi, frequency fiAnd phase placeAnd 10~50Hz
The m-Acetyl chlorophosphonazo of frequency range calculates all spectral lines of residue except first-harmonic spectral line according to the method for IEC, removes the leakage of intrasonic intersegmental harmonic wave
The root-mean-square value A of harmonic amplitude after impactr;
4) step 3) inIt is the m-Acetyl chlorophosphonazo parameter of 0~50Hz subharmonic frequency range, according to IEC standard
Regulation, be designated as 0.5 m-Acetyl chlorophosphonazo;
5) combining step 2) and step 4) obtain the m-Acetyl chlorophosphonazo parameter of full frequency band.
In an embodiment of the present invention, the k that intrasonic m-Acetyl chlorophosphonazo produces occursmIt is i.e. k on Δ f that spectral line falls just in frequencym
=1, (km-1) spectral line falls on direct current spectral line, carries out filtering DC component D every straight process before carrying out FFT spectrum analysisc,
Then X (km-1)=0 so that formula (6) can not be write as the form of ternary secondly equation group, now may select km-1 is-1, corresponding X
(km-1)=X*(km), " * " expression takes conjugation, and at this moment, the spectral line serial number (-1,1,2) of utilization, corresponding spectrum is (X*(1)、
X (1), X (2)), formula fi=fi 1Δ f can be reduced to:
The present invention is different from traditional interpolation correction algorithm, and this algorithm considers negative frequency to intrasonic m-Acetyl chlorophosphonazo spectral line
Interference, becomes equation with three unknowns group to solve intrasonic m-Acetyl chlorophosphonazo frequency by three spectral line value simultaneous, and according to the frequency of maximum spectral line
Spectrum and conjugate thereof solve amplitude and the phase place of m-Acetyl chlorophosphonazo;Intrasonic quickly and is accurately extracted in achieving the short time
The relevant feature parameters of m-Acetyl chlorophosphonazo, it is ensured that the real-time of algorithm and effectiveness requirement time actually detected.The present invention derives simultaneously
Parameter correction formula when m-Acetyl chlorophosphonazo frequency is less than 1 frequency resolution Δ f, the actual m-Acetyl chlorophosphonazo frequency that can solve is low
To 0.3 below Δ f, greatly breach tradition and limited by frequency resolution based on DFT harmonic analysis, shorten analysis letter
Number length, this is that other algorithm institute is irrealizable.And the algorithm of the present invention is not affected by DC component and to system noise
Sound is insensitive ensure that its accuracy and robustness.
Detailed description of the invention
Consider the voltage of electrical network in practical power systems and frequency always in change, and the detection of m-Acetyl chlorophosphonazo is always and electrical network
Voltage closely related with frequency, according to voltage and the frequency transformation period speed of electrical network, signal model in actual electric network is entered
Row classification: 1) long time stable state, the voltage magnitude of power network signal and frequency expected value long period constant stationary random signal;2)
Stable in short-term, the voltage magnitude of power network signal and frequency expected value short period constant stationary random signal;3) astable shape
State, during fault, voltage magnitude and the frequency of power network signal are the non-stationary signal of time-varying.And the foundation of the model of the present invention is pin
For be analyzed under system normal operating condition, using the constant desired value of voltage magnitude in a period of time and frequency as reality
E [A under voltage magnitude and frequency, i.e. a period of time steady statue in the electrical network of borderi(t)]=Ai, E [fi(t)]=fi;Consider simultaneously
Noise signal in actual electric network is white Gaussian noise signal v (t) separate with each frequency component of electrical network.
The foundation power network signal mathematical model containing intrasonic m-Acetyl chlorophosphonazo:
A in formulai、fiWithFor the amplitude of each component, frequency and initial phase, D in power network signalcFor actual electric network signal
The middle direct-flow offset weight that may comprise, v (t) is the white Gaussian noise signal separate with each frequency component of electrical network, and M is
The frequency component number comprised in power network signal;Time t ∈ [t0,t0+ T], T is the intercepting time period of signal analysis.
It is discrete series x in view of the signal collected in actual electric network*N (), therefore by the mathematical model in formula (12)
Change discrete form into:
F in formulasFor the sample frequency of power network signal, n ∈ [0, N-1], N are the total sampling number of discrete signal intercepted.
It is pointed out that the when that the sampled signal to actual electric network carrying out spectrum analysis, typically can first carry out every directly
Process and filter DC component Dc, it is therefore an objective to that improves computer integer FFT does spectrum precision, prevents windowing (except rectangular window) from producing
Raw Frequency and interfering frequency.First ignore system noise part v (n) when carrying out the mathematical analysis of model, after analyze again
Containing during system noise testing result being affected situation;It is also noted that the linear superposition that signal is each component in formula (12),
Still this overlaying relation is met when it is carried out DFT (FFT) spectrum analysis.Therefore, for without loss of generality, analyze unifrequency from
Scattered signal model:
From formula (14) it can be seen that the signal that reality gathers from electrical network is discrete real sequence x (n), n ∈ [0, N-1],
Not single-frequency complex exponential signal sequence, adds the normalization discrete Fourier transform (DFT) (Discrete of rectangular window to it
Fourier Transform, DFT)
K=0 in formula, 1 ..., N-1.
If the frequency resolution of spectral line is Δ f, Δ f=fs/ N, substitutes into formula (14) in formula (15):
In practical situation, General N value is bigger, therefore sin [π (fi/Δf-k)/N]≈π(fi/Δf-k)/N.If normalization
Frequency is fi 1, fi 1=fi/ Δ f, is further simplified as formula (15):
In formula
It can be seen that after actual electric network sampled signal carries out DFT spectrum analysis, frequency spectrum there will be two from formula (17)
Frequency component fiWith-fi, wherein Section 2 is the negative frequency components composition of frequency spectrum.Tradition algorithm mould based on DFT spectrum analysis
Have ignored the negative frequency impact on spectral line in type, directly take Section 1 in formula (17), do so has certain reason, reason
Being that actual m-Acetyl chlorophosphonazo component works as distance 0 frequency spectrum enough far, negative frequency is micro-on the impact of actual m-Acetyl chlorophosphonazo component
It is micro-, but for intrasonic m-Acetyl chlorophosphonazo, negative frequency is the biggest on its impact, and this is also that general algorithm is at pin
The reason losing efficacy the detection of ultralow m-Acetyl chlorophosphonazo, reality have to consider negative frequency pair when detecting power network signal intrasonic m-Acetyl chlorophosphonazo
Its impact.Negative frequency affects schematic diagram as shown in Figure 1 to frequency spectrum.
Consider the discrete spectrum of negative frequency impact, take adjacent three spectral line km-1、kmAnd kmSpectrum X (the k of+1m-1)、X(km)
With X (km+ 1), k herem>=2 for integer and meet the spectral line of its correspondence for maximum spectral line.So
In like manner can verify
H[fi 1-(km+ 1)]=H [fi 1-(km-1)]=H (fi 1-km) (19)
Therefore, noteHave according to formula (15):
(fi 1-km-p)-1H++(fi 1+km+p)-1H--X(km+ p)=0 (20)
In formula, three unknown numbers are H+、H-And fi 1, when p=-1,0,1, a ternary linear function group can be become with simultaneous,
And equation existence and unique solution, therefrom can obtain fi 1、H+And H-.For the convenience solved, formula (20) is considered as linear homogeneous
Equation group, untrivialo solution vector is { H+, H-,-1}T, then its unique conditional existed is that the determinant of coefficient matrix C of equation group is
Zero det (C)=0.
Coefficient matrix C is:
Frequency normalization f to be asked can be solved by det (C)=0i 1Explicit Expression formula be
In order to ensure the accuracy calculated, the k in formula (22) need to be ensuredmSpectral line is the highest spectral line, then km-1 and km+ 1 respectively
For the most adjacent spectral line, then one of them must be second highest spectral line, then consider the intrasonic m-Acetyl chlorophosphonazo of negative frequency impact
Frequency estimation is
fi=fi 1·Δf (23)
In formula, Δ f is frequency resolution.
Consider the impact of system dc component: reality probably occurs the k that intrasonic m-Acetyl chlorophosphonazo producesmSpectral line is lucky
Falling in frequency is i.e. k on Δ fm=1, so (km-1) spectral line falls on direct current spectral line.Owing to system may have direct current to divide originally
The interference of amount, carried out filtering DC component D every straight process before carrying out FFT spectrum analysisc, i.e. X (km-1)=0 so that formula
(20) can not be write as the form of ternary secondly equation group, now be may select km-1 is-1, corresponding X (km-1)=X*(km), " * "
Represent and take conjugation.At this moment, the spectral line serial number (-1,1,2) of utilization, corresponding spectrum is (X*(1), X (1), X (2)), formula (22)
Can be with abbreviation:
In order to ensure signal to noise ratio and the computational solution precision of actual signal spectral line, carry out intrasonic m-Acetyl chlorophosphonazo amplitude and phase place
Estimate time similar with other interpolation algorithm, need selection locally the highest spectral line to calculate.By formula (17) the highest spectral line kmEquation
Formula:
In formula
Wherein sinc (x)=sin (π x)/(π x).
Formula (25) is taken and is conjugated:
F is obtained by formula (22)iAfter, simultaneous formula (25) and formula (26) composition equation group, only A inside equation groupiWithIt is
Unknown quantity, remaining is known quantity, therefore can solve:
Formula (22) and formula (27) are consideration negative frequency and the frequency of ultralow m-Acetyl chlorophosphonazo of interpolation, amplitude and phase estimation are public
Formula.It should be noted that above-mentioned model is built upon what signal was intercepted by rectangular window (i.e. not windowing), this also meets
The IEC61000-4-7 testing requirement to m-Acetyl chlorophosphonazo.
Have been detected as intrasonic m-Acetyl chlorophosphonazo for above-mentioned theory, and combine IEC61000-4-7 and GB GB/T
The regulation that m-Acetyl chlorophosphonazo is detected by 24337-2009, can form a whole set of m-Acetyl chlorophosphonazo detection scheme being applicable to Whole frequency band, its stream
Journey block diagram is as shown in Figure 2.Its detailed step is:
1) signal collected by electrical network, intercepts one section (such as 10 cycles of IEC regulation, 200ms, correspondence with rectangular window
Frequency resolution is 5Hz) make FFT (DFT) spectrum analysis.
2) for the m-Acetyl chlorophosphonazo component of more than 50Hz, regulation m-Acetyl chlorophosphonazo detected according to IEC61000-4-7, humorous with
Ripple group or the form of subgroup, measure between two harmonic waves and the m-Acetyl chlorophosphonazo parameter of frequency range between first-harmonic and second harmonic, and with two
The meansigma methods of harmonic wave sequence number as corresponding m-Acetyl chlorophosphonazo group or subgroup frequency (as between 2 subharmonic and 3 subharmonic m-Acetyl chlorophosphonazo remember
It is 2.5 m-Acetyl chlorophosphonazo group or subgroups).
3) m-Acetyl chlorophosphonazo (i.e. subharmonic) for 0~50Hz frequency range must take into the impact of negative frequency, the frequency of subharmonic
The least, affected the most serious by negative frequency;M-Acetyl chlorophosphonazo quantity in view of same frequency range will not be too many, and electrical network exists frequency simultaneously
Rate and the fluctuation of amplitude, as a example by the 5Hz frequency resolution that IEC specifies, the intrasonic m-Acetyl chlorophosphonazo of 0~10Hz frequency range is according to herein
In the algorithm mentioned calculate intrasonic m-Acetyl chlorophosphonazo parameter (amplitude Ai, frequency fiAnd phase place), and the m-Acetyl chlorophosphonazo of 10~50Hz frequency ranges
Remain that all spectral lines (without first-harmonic spectral line) remove after intrasonic intersegmental harmonic wave leakage contribution mean square is calculated according to the method for IEC
Root Ar。
4) the 3rd) in stepIt is the m-Acetyl chlorophosphonazo parameter of 0~50Hz subharmonic frequency range, marks according to IEC
Accurate regulation, is designated as 0.5 m-Acetyl chlorophosphonazo.
5) combining step 2) and step 4) obtain the m-Acetyl chlorophosphonazo parameter of full frequency band.
The harmonic detection method proposed for the present invention carries out Digital Simulation, and simulated environment is Matlab 2012b,
The 0.2s sampled signal length (synchronized sampling) specified with IEC61000-4-7 in emulation embodiment, corresponding 5Hz frequency resolution,
If the frequency of intrasonic m-Acetyl chlorophosphonazo is 3.5Hz, it is 1% relative to the amplitude coefficient of fundamental frequency, the inspection of various algorithms under noise circumstance
Survey result is as shown in table 1:
The accuracy of detection of each algorithm in table 1 noise circumstance
Note: when having noise, carries out 100 Monte Carlo simulation experiments and asks for the meansigma methods of result;" " represents correspondence
Algorithm is without correct testing result.
As can be seen from Table 1, window function and interpolation algorithm based on DFT lost efficacy when detecting intrasonic m-Acetyl chlorophosphonazo, and reason is to be subject to
Restriction to frequency resolution;Prony algorithm has high accuracy of detection when noiseless, but trickle noise can be made
Become precision drastically drop to cannot detect, its robustness is bad;The method carried herein can have in bigger noise circumstance
Higher accuracy of detection, it is ensured that algorithm real system detect in robustness and real-time.
The present invention and the comparison of other algorithms:
(1) algorithm and IEC standard method comparison herein
The m-Acetyl chlorophosphonazo group algorithm detection subharmonic that IEC61000-4-7 specifies, it is assumed that the relative magnitude of low-frequency inter-harmonic is Ai
=0.1, frequency is fi, phase place is 0.Sampling time is 0.2s, works as fiWith the step-length of 0.05Hz from 0~40Hz change time, corresponding
Testing result is as shown in Figure 3
As seen from Figure 3, IEC algorithm is 5Hz in frequency resolution, when range of allowable error is within ± 5%, it is possible to
The minimum m-Acetyl chlorophosphonazo frequency of detection is about 7.5Hz, simultaneously takes account of the phase angle variations of intrasonic m-Acetyl chlorophosphonazo and stablizing of algorithm
Property, the m-Acetyl chlorophosphonazo IEC algorithm of 0~10Hz frequency range cannot detect exactly.
And algorithm harmonic detecting intersegmental to intrasonic result such as Fig. 4 shows herein:
As seen from Figure 4, for the sampled signal of 0.2s length, doing frequency resolution corresponding to spectrum analysis is 5Hz,
Algorithm herein can IEC standard regulation ± 5% range of error in detect that minimum m-Acetyl chlorophosphonazo frequency is that 1Hz is (the most right
Answering m-Acetyl chlorophosphonazo phase place is 0), greatly shake off the restriction of frequency resolution, it is achieved that detect rapidly and accurately in the short period
Excess of export low-frequency inter-harmonic.
Under noiseless disturbed condition, IEC regulation ± 5% range of tolerable variance in, the phase angle variations scope of m-Acetyl chlorophosphonazo is 0
~2 π, obtain the minimum m-Acetyl chlorophosphonazo frequency that algorithm and IEC standard algorithm can detect when unlike signal length herein, enumerate
Numerical value and changing trend diagram thereof are respectively as shown in table 2 and Fig. 5.
The minimum m-Acetyl chlorophosphonazo frequency that in table 2 range of tolerable variance, algorithm can identify
By table 2 and Fig. 5 it can be seen that herein algorithm can detect intrasonic m-Acetyl chlorophosphonazo in a short period of time, and nothing
Need to increase signal length, thus quickly detect for the m-Acetyl chlorophosphonazo of ULF and provide new way.
(2) algorithm compares with Prony algorithm herein
As a example by 10 signal cycles of IEC standard regulation, m-Acetyl chlorophosphonazo frequency is in 0~10Hz change, its amplitude coefficient (phase
For fundamental frequency) it is 1%, carry out 100 Monte Carlo simulation experiments under different noise circumstances and tie with Prony algorithm detection
Fruit contrast is as shown in table 3:
Each algorithm accuracy of detection contrast situation during table 3 noise jamming
Note: " " represents the algorithm of correspondence cannot detect result.
In different noise circumstances, the accuracy of detection variation tendency of Prony algorithm and herein algorithm as indicated with 6:
By table 3 and Fig. 6 it can be seen that Prony algorithm is in noise-free environment, there is at a relatively high accuracy of detection
(reach 10-11The order of magnitude), and along with the increase (i.e. signal to noise ratio diminishes) of noise, its accuracy of detection drastically drops to detect
(being approximately higher than the noise of 50dB intensity).And voltage and frequency due to electrical network is always being continually changing in practical power systems, with
Time sampled signal inevitably introduce random noise, Prony algorithm is very sensitive to noise signal, is not suitable for power system
The detection of intrasonic m-Acetyl chlorophosphonazo.Simultaneously it can also be seen that this paper algorithm noise immunity is preferable, when noise strengthens, testing result essence
Degree change is little (stable 10-4~10-3The order of magnitude), it is possible to meet the demand that power system is actually detected.
In sum, the short time that can realize intrasonic m-Acetyl chlorophosphonazo of the present invention is quickly detected, and this algorithm is not simultaneously
Affected by DC component and insensitive to system noise, it is ensured that this algorithm robustness in actually detected is with in real time
Property.
The foregoing is only presently preferred embodiments of the present invention, all impartial changes done according to scope of the present invention patent with
Modify, all should belong to the covering scope of the present invention.