CN106597100A - Interpolation FFT estimation method of dynamic frequency of wide area power grid - Google Patents

Abstract

The invention discloses an interpolation FFT estimation method of a dynamic frequency of a wide area power grid. The method comprises: a grid signal is sampled and weighting is carried out on the sampling signal by using a second-order maximum side-lobe decaying window; FFT is carried out on the weighted signal to obtain a discrete spectrum; a three-spectral-line interpolation algorithm is used for processing to obtain the frequency of the window; and then a frequency of a next window is calculated by using a moving window function and then a frequency changing rate is calculated, thereby realizing dynamic frequency estimation of a grid signal. With the method disclosed by the invention, accurate grid frequency measurement can be realized while the grid frequency fluctuates; and the calculation load is low.

Description

A kind of interpolation FFT method of estimation of the Wide Area Power dynamic frequency

Technical field

The present invention relates to signal measurement field, specifically a kind of interpolation FFT method of estimation of the Wide Area Power dynamic frequency.

Background technology

With various types of nonlinear loads (high-power rectifying equipment, converter, electric arc furnace, electric railway with And home electric etc.) be continuously increased, voltage, current sinusoidal waveform wave phenomenon are on the rise in power system.In actual feelings In condition, mains frequency is difficult to keep expected standard frequency, but there is certain fluctuation range.Therefore, dynamic sinusoidal signal The accurate estimation of frequency components is the crucial problem for being badly in need of solving.

Dynamic sinusoidal signal intermediate frequency rate method of estimation is progressively turned to Digital Signal Processing from the analog detecting method of early stage Russian Market based on technology, such as wavelet transformation, Short Time Fourier Transform, artificial neural network, support vector machine and FFT etc..Method based on FFT is widely used in spectrum analyses, matched filtering, digital communication, image because its realization is simple The various fields such as process.And it is exactly effects of the FFT in terms of spectrum analyses that this patent is adopted, but when measured signal frequency When generation conversion causes non-synchronous sampling or carries out non-integer to signal and block, there is fence effect and spectral leakage in FFT Phenomenon so that the signal frequency for calculating is forbidden, it is impossible to meet the requirement of accurately measurement.For this purpose, can select the window of function admirable Function and enter non-integer of the row interpolation amendment to reduce non-synchronous sampling or data to result of calculation and block the error for causing.

The content of the invention

The invention aims in view of the shortcomings of the prior art, there is provided one kind is easily understood, precision is higher, The faster frequency estimating methods of calculating speed.

Solution proposed by the present invention is：First, power network signal is sampled and using second order maximum side lobe attenuation Window is weighted to sampled signal；Then FFT is done to the signal after weighting and obtains discrete spectrum, obtained using three spectral line interpolation algorithms To the frequency of the window, the frequency of next window is calculated finally by mobile window function, and then calculate frequency change rate, realize electricity The dynamic frequency of net signal is estimated.

The interpolation FFT method of estimation of the Wide Area Power dynamic frequency proposed by the present invention is comprised the following steps：

Use sample frequency f_sPower network signal is sampled, discrete sampling sequence x (n) that length is N is obtained；Using length Second order maximum side lobe attenuation window w (m) for M is weighted to x (n) and blocks, and obtains the weighting sequence x that length is M_w(m)=x (m)w(m)；Then to weighting sequence x_wM () is FFT and obtains discrete spectrum | X_w(m)|；When obtaining this by three spectral line interpolation algorithms Between power network signal in window frequency；The frequency of future time window is calculated by mobile window function, so as to be calculated frequency Rate rate of change, realizes electrical network Dynamic Frequency Measurement.

Described method, length is that the frequency spectrum function of the second order maximum side lobe attenuation window of M is

Described method, it is by amplitude near fundamental frequency in lookup discrete spectrum that three spectral line interpolation algorithms calculate frequency Maximum spectral line, remembers that its position is l, then its two adjacent position of spectral line is respectively l-1 and l+1, thus can calculate interpolation coefficient η：Being calculated frequency using interpolation coefficient η is：

Described method, by mobile window function the power network signal frequency of future time window is calculated, and is calculated and is obtained frequency Rate of change, is that the second order maximum side lobe attenuation window by length for M is moved rearwards by M/2 point, is calculated using three spectral line interpolation algorithms The power network signal frequency of future time window, be so as to calculate frequency change rate：

L in formula_k, η_kK-th time window power network signal spectral magnitude maximum spectral line position and three spectral lines are represented respectively Interpolation coefficient, wherein Expression is rounded downwards.

The interpolation FFT method of estimation of the Wide Area Power dynamic frequency of the present invention, with being easily understood, calculating speed faster, essence The features such as degree is higher, using the frequency of windowed interpolation method signal calculated, so as to reduce fence effect and spectral leakage to frequency The impact of calculating, by mobile window function frequency change rate is calculated, and accurate electrical network frequency is realized when mains frequency occurs and fluctuating Rate is measured.

Description of the drawings

Fig. 1 is the flow chart of the interpolation FFT method of estimation that the present invention realizes the Wide Area Power dynamic frequency；

Fig. 2 is the schematic diagram that the present invention realizes window function mobile computing frequency change rate.

Specific embodiment

The present invention proposes a kind of interpolation FFT method of estimation of the Wide Area Power dynamic frequency.Make detailed below in conjunction with accompanying drawing Explanation：

In the present embodiment, using the signal generator Agilent 33120A of Agilent company of the U.S. the signal of input is provided And it is connected to data collecting card.Data collecting card selects the NI USB-6210 of National Instruments, and its sample frequency is arranged For 4KHz, data conversion digit is for 16 and includes 16 signal input channels, selects one of input channel to believe with input Number be connected, NI USB-6210 are connected with computer by USB.Computer adopts association's Thinkpad X220 notebook computers, National Instruments' LabVIEW softwares are installed, version number is 12.0, and computer is connected by USB with NI USB-6210 Afterwards, running LabVIEW softwares can receive discrete sampling sequence x (m) that the length for collecting is N=10240.

As shown in figure 1, determining sample frequency f_s=4kHz, length for M=1024 second order maximum side lobe attenuation window to from Scattered sample sequence is weighted, and obtains the weighting sequence x that length is M_w(m).Then FFT is carried out to weighting sequence and obtains discrete frequency Rate | X_w(m) |, by the spectral line of amplitude maximum near fundamental frequency in lookup discrete spectrum, remember that its position is l, then its is adjacent Two position of spectral line are respectively l-1 and l+1, thus calculate interpolation coefficient η, and specific formula for calculation is： Thus the frequency of the time-ofday signals is obtained, specific formula for calculation is：

Such as Fig. 2, window function central instant is t_k, moment interpolation coefficient η is calculated by above-mentioned three spectral line interpolations algorithm_k =0.574, frequency f_k=52.2250, window function is moved into M/2 point, obtain t_k+1The interpolation coefficient η at moment_k+1=0.580, Frequency f_k+1=52.2391, it is ν=(f by the definition of frequency change rate_k+1-f_k)/(t_k+1-t_k), arrangement is obtained

Then ν=0.11.

Claims (4)

1. a kind of interpolation FFT method of estimation of the Wide Area Power dynamic frequency, it is characterised in that：Use sample frequency f_sTo power network signal Sampled, obtained discrete sampling sequence x (n) that length is N；Using second order maximum side lobe attenuation window w (m) that length is M to x N () is weighted and blocks, obtain the weighting sequence x that length is M_w(m)=x (m) w (m), then to weighting sequence x_wM () is FFT Obtain discrete spectrum | X_w(m)|；The frequency of the power network signal in the time window is obtained by three spectral line interpolation algorithms；By moving Dynamic window function calculates the frequency of future time window, so as to be calculated frequency change rate, realizes electrical network Dynamic Frequency Measurement.

2. method according to claim 1, it is characterised in that the length for the second order maximum side lobe attenuation window of M frequency spectrum Function is

3. method according to claim 1, it is characterised in that it is by looking into that three described spectral line interpolation algorithms calculate frequency The neighbouring amplitude maximum spectral line of fundamental frequency in discrete spectrum is looked for, remembers that its position is l, then its two adjacent position of spectral line is respectively L-1 and l+1, thus can calculate interpolation coefficient η：Frequency is calculated using interpolation coefficient η For：

$f=f_s/M\·(l+\frac{-3+\sqrt{9+4{\mathrm{\η}}^2}}{\mathrm{\η}}).$

4. method according to claim 1, it is characterised in that described calculates future time window by mobile window function Power network signal frequency, calculate obtain frequency change rate, be by length for M second order maximum side lobe attenuation window be moved rearwards by M/2 It is individual, the power network signal frequency of future time window is calculated using three spectral line interpolation algorithms, be so as to calculate frequency change rate：

$\mathrm{\ν}=\frac{2{f_s}^2}{M^2}\[l_k-l_{k+1}+\frac{-3+\sqrt{9+4{{\mathrm{\η}}_k}^2}}{{\mathrm{\η}}_k}-\frac{-3+\sqrt{9+4{{\mathrm{\η}}_{k+1}}^2}}{{\mathrm{\η}}_{k+1}}\]$

Wherein l_k, η_kK-th time window power network signal spectral magnitude maximum spectral line position and three spectral line interpolations are represented respectively Coefficient, wherein k=1,2 ..., Expression is rounded downwards.