CN102636775A - Wind profile radar echo spectrum reconfiguration method based on fuzzy logic recognition - Google Patents

Abstract

The invention discloses a wind profile radar echo spectrum reconfiguration method based on fuzzy logic recognition. The wind profile radar echo spectrum reconfiguration method includes the specific steps: rapidly performing median filtering on wind profile radar echo power spectra; extracting turbulent echo by means of fuzzy logic recognition according to difference among turbulent signals in echo signals and clutter (ground clutter, intermittent clutter and radio frequency interference) signals in terms of spectral continuity, spectral intensity, spectral position and spectral width; fuzzing subordinate functions of the spectral intensity and the spectral continuity for the echo power spectra, converting the subordinate functions into a spectral intensity fuzzy basis and a spectral continuity fuzzy basis and recognizing a fuzzy logic system by the aid of the fuzzy bases serving as substitute values of actually detected power spectra; and calculating wind profile radar turbulent echo signal moment data including noise level, signal-to-noise ratio, first moment and second moment by means of performing Gaussian least square fitting on turbulent echo spectral peaks recognized by the aid of fuzzy logic. By the aid of the wind profile radar echo spectrum reconfiguration method, turbulent echo signals without clutter interference can be acquired, and the detection height of an atmospheric wind field is effectively increased.

Description

Wind profile radar echo spectrum reconstructing method based on fuzzy logic identification

Technical field:

The present invention relates to signal Processing and data analysis technique field, be specifically related to a kind of wind profile radar echo spectrum reconstructing method based on fuzzy logic identification.

Background technology:

Wind profile radar is a detected object with the clear sky atmosphere, utilizes atmospheric turbulence that the detection of atmospheric wind and turbulent flow is carried out in electromagnetic scattering process, has the advantage of unmanned continuous working, high-spatial and temporal resolution; Comprised abundant turbulent flow information in the echoed signal; Have obvious fluctuating fluctuation characteristics, spectrum width broad simultaneously and with multiple clutter; Mainly contain ground clutter, intermittence clutter, radio frequency interference etc. such as trees, the turbulent flow echoed signal is mixed in the various clutters.This characteristics have determined that Radar Signal Processing is from various clutters, to extract weak useful signal and make useful signal have the process of statistical average meaning.In order to discern turbulence signal, will carry out clutter and suppress to handle.To the time-frequency characteristics of clutter, disposal route is divided into time domain and frequency domain is handled.The time domain noise signal is handled and is adopted dsp board to embed bank of filters, and processing speed is fast, is convenient to radar and handles in real time, can better suppress ground clutter, but powerless to the intermittence clutter, the easy simultaneously details filtering with the atmosphere echoed signal.The frequency domain clutter suppresses disposal route and considers the characteristics of clutter in frequency domain; Such as near the echo spectrum symmetry zero-frequency; Under ground clutter spectrum and clear sky atmosphere echo separation case, it is fine that clutter suppresses effect, but when the two overlaps, be easy near the cancellation of the spectrum of the part atmosphere the zero-frequency; A.D.Siggia and R.E.Passarelli propose Gauss model self-adaptive processing clutter suppression method, in clutter reduction, have well avoided the unnecessary cancellation of atmosphere echo, but have not considered strong echo disturbed condition.Above method all is the situation that are applicable to that single clutter kind with certain characteristics exists, but therefore common multiple interference and depositing in the radar real work is badly in need of a kind of recognition methods that in echoed signal, comprises turbulent flow echo under the multiple noise jamming situation.

Summary of the invention:

The purpose of this invention is to provide a kind of wind profile radar echo spectrum reconstructing method, under the condition that multiple clutter exists, can accurate recognition go out the turbulent flow echo, and estimate echo spectral moment data through the method for Gauss's least square fitting based on fuzzy logic identification.

For solving the problems of the technologies described above, the technical scheme that the present invention adopts is:

Based on the wind profile radar echo spectrum reconstructing method of fuzzy logic identification, concrete steps are:

Step 1: wind profile radar echo power spectrum is carried out pre-service; Adopt the Fast Median Filtering method, be specially: the medium filtering window that to set up a width be 2N+1, and move along digital signal sequences, step-length is 1; Before window moves, obtain the intermediate value of 2N+1 sequence in the window earlier; After window moves, find out the position in the sequence that the data object in the new immigration window sequenced in front, obtain the intermediate value of asking.Window width is got 5 points, i.e. N=2 observes the variation of data in the window when moving for adjacent twice of medium filtering window, when medium filtering window during in first position; Burst is { x (i+1), x (i+2), x (i+3), x (i+4) in the window; X (i+5) }, establish its size ordering and be { w (i+1), w (i+2), w (i+3); W (i+4), w (i+5) }, then intermediate value is w (3).Window moves right after one, and data are { x (i+2), x (i+3) in the window; X (i+4), x (i+5), x (i+6) }; At first get rid of x (i+1) relative { w (j), j=1,2; ...; 5} is then with x (i+6) and remaining { w (j), j=1; 2; ..., each object contrast ordering in the 5} sequence, that obtain thus in this sequence of new sequence

is the intermediate value of asking.

Step 2: turbulence signal and the noise signal difference aspect spectrum continuity, spectral intensity, spectral position and spectrum width: the turbulent flow echo has continuity, spectral position frequency displacement variation continuously, spectral intensity and spectrum width big; The ground clutter spectral position is positioned near the zero-frequency; The intermittence clutter does not have continuity, and spectral position is big at some range bin appearance, spectral intensity and spectrum width; The radio frequency interference spectral position is fixed, and spectral intensity and spectrum width are little.Through going out the intermittence clutter to the analysis of echo spectrum is distinguishable; Can distinguish radio frequency interference to spectral intensity and spectrum width analysis; Can slacken the ground clutter influence to spectral position about the zero-frequency symmetry analysis.Can tell the turbulent flow echo through above-mentioned analysis-by-synthesis.

Utilize the fuzzy logic recognition methods to extract the turbulent flow echo then.Set up subordinate function characteristic quantity is carried out Fuzzy processing, judge whether belong to the turbulent flow echo according to the fuzzy basic shared weight of each characteristic.Specifically be meant:

1) the spectral intensity normalization that the power spectrum symmetry is disappeared mutually, the spectral intensity after the normalization are P _Y(f _Ij), f _IjRepresent the corresponding Doppler shift in j position on the i range bin.P _Y(f _Ij) big more, the degree that is under the jurisdiction of the turbulent flow echo is big more; Otherwise it is more little.Set up " S " type subordinate function on this basis to the spectral intensity Fuzzy processing:

$S\left[P_Y\left(f_{\mathrm{ij}}\right)\right]=\left\{\begin{array}{c}0,P_Y\left(f_{\mathrm{ij}}\right)\∈[-\mathrm{1,0.3}]\\ \frac{10}{3}\·P_Y\left(f_{\mathrm{ij}}\right)-1,P_Y\left(f_{\mathrm{ij}}\right)\∈\left[\mathrm{0.3,0.6}\right]\\ 1,P_Y\left(f_{\mathrm{ij}}\right)\∈\left[\mathrm{0.6,1}\right]\end{array}\right.---\left(1\right)$

2) set up the Gaussian subordinate function to composing the continuity Fuzzy processing:

$G(w=j-[\stackrel{\‾}{f_{i-1}}\left]\right)=\exp [-\ln \left(\frac{5}{3}\right)\·w^2/\mathrm{\Δ}{\stackrel{\‾}{f}}^2],\frac{n_S}{2}-\left[\stackrel{\‾}{f_{i-1}}\right]\≤w\≤\frac{n_S}{2}+1-\left[\stackrel{\‾}{f_{i-1}}\right]---\left(2\right)$

3) through after the subordinate function Fuzzy processing, convert wind profile radar echo power spectrum to two fuzzy bases: the fuzzy base of spectral intensity (P _Y(f _Ij)) and the fuzzy base of spectrum continuity

In fuzzy logic system, the identification that utilizes fuzzy base to replace the substitution value of the power spectrum of actual detection to carry out system is handled, so problem concentrates in the foundation of subordinate function and rule of fuzzy system.Suppose in the radar power spectrum separately between the turbulent flow echo and various clutter, concerning two fuzzy bases, represent with the one dimension subordinate function:

$T_{\mathrm{ij}}=w_1\·S\left[P_Y\left(f_{\mathrm{ij}}\right)\right]+w_2\·G(j-[\stackrel{\‾}{f_{i-1}}\left]\right),\underset{i=1}{\overset{2}{\mathrm{\Σ}}}{w}_i=1---\left(3\right)$

Adopt the method for non-equal weight, wherein, w ₁, w ₂Represent spectral intensity and the successional weight factor of spectrum respectively, for the T of i range bin _IjBe worth big more, the P that it is corresponding _Y(f _Ij) be under the jurisdiction of the turbulent flow echo degree big more.

Step 3: after identifying the turbulent flow target echo through fuzzy logic, need to demarcate turbulent flow echo signal bandwidth so that carry out the square data estimation.Because the echoed signal generation mechanism of wind profile radar is mainly turbulent scattering; Suppose to be in the radar pulse volume RANGE IN HOMOGENEOUS ISOTROPIC TURBULENCE; When the statistical property of research turbulent flow percent ripple; Its probability distribution is approximately Gaussian, therefore uses Gauss's least-square fitting approach match turbulent flow echo spectrum to distribute:

$P_i\left(f\right)=P_{i0}\exp [-{(f-\stackrel{\‾}{f_i})}^2/{2\mathrm{\σ}}_i^2]+P_N---\left(4\right)$

P _N, P _I0,

Peak power, first moment and second moment for noise level, i range bin.About the boundary discussion of turbulent flow echo and noise jamming overlapping, introduce the degree of tilt analysis:

$g\left(f_{\mathrm{ij}}\right)=\frac{\frac{\mathrm{dp}\left(f_{\mathrm{ij}}\right)}{\mathrm{df}}}{\max \left(\right|\frac{\mathrm{dp}\left(f_{\mathrm{ij}}\right)}{\mathrm{df}}\left|\right)}$

Spectrum peak position left side g (f _Ij)＜0, right side g (f _Ij)＞0 is judged the boundary of turbulent flow echo and clutter thus.Draft unpolluted turbulent flow echo spectrum point on each range bin, carry out least square fitting based on these spectrum points:

P(f)＝Af ²+Bf+C (5)

By the least square curve fitting principle, parameter A, B, C confirm that method is:

$\left[\begin{array}{ccc}\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{f}_i^2& \underset{i=1}{\overset{n}{\mathrm{\Σ}}}{f}_i& n\\ \underset{i=1}{\overset{n}{\mathrm{\Σ}}}{f}_i^3& \underset{i=1}{\overset{n}{\mathrm{\Σ}}}{f}_i^2& \underset{i=1}{\overset{n}{\mathrm{\Σ}}}{f}_i\\ \underset{i=1}{\overset{n}{\mathrm{\Σ}}}{f}_i^4& \underset{i=1}{\overset{n}{\mathrm{\Σ}}}{f}_i^3& \underset{i=1}{\overset{n}{\mathrm{\Σ}}}{f}_i^2\end{array}\right]\left[\begin{array}{c}A\\ B\\ C\end{array}\right]=\left[\begin{array}{c}\underset{i=1}{\overset{n}{\mathrm{\Σ}}}P\left(f_i\right)\\ \underset{i=1}{\overset{n}{\mathrm{\Σ}}}{f}_i\·P\left(f_i\right)\\ \underset{i=1}{\overset{n}{\mathrm{\Σ}}}{f}_i^2\·P\left(f_i\right)\end{array}\right]---\left(6\right)$

N representes that the turbulent flow echo spectrum of treating match counts, and obtains each square parameter thus and is respectively:

$P_{i0}={10}^{(\frac{B^2}{4A}-C)/10};$ ${\stackrel{\‾}{f}}_i=-\frac{B}{2A};$ ${\mathrm{\σ}}_i^2=\frac{5\log 10\left(e\right)}{A}$

To can obtaining turbulent flow echoed signal power through the signal spacing integration after the least square fitting, the ratio of signal power and noise power can draw the turbulent flow echo signal to noise ratio (S/N ratio) that accuracy has improved.

Compared with present technology, good effect of the present invention is:

The fuzzy logic of utilizing that the present invention proposes is discerned the turbulent flow echo, after the subordinate function Fuzzy processing, converts wind profile radar echo power spectrum to identification processing that two fuzzy bases carry out system.The present invention not only can eliminate single influence of planting clutter to the turbulent flow echoed signal, also can effectively reject at useful signal and various ways clutter simultaneously to have the noise jamming under the situation simultaneously; Gauss's least-square fitting approach carries out square to the turbulent flow echo spectrum to be estimated, can calculate echoed signal power, noise power, signal to noise ratio (S/N ratio), first moment and second moment more accurately, and not receive noise jamming.Handle through the present invention, obtain pure turbulent flow information, and can synthesize atmospheric wind more accurately.

Description of drawings:

Fig. 1 is based on fuzzy logic method identification turbulent flow target echo process flow diagram among the present invention.

The left figure of Fig. 2 is the original echo spectrogram that on June 4th, 2008,10:42:16 east wave beam detected, and right figure is the echo spectrogram that utilizes behind the fuzzy logic recognition methods impurity elimination ripple.

The left figure of Fig. 3 is the WIND FIELD EVOLUTION figure that on June 4th, 2008, original echo calculated, and right figure is the WIND FIELD EVOLUTION figure after utilizing the present invention to handle.

Embodiment:

Based on the wind profile radar echo spectrum reconstructing method of fuzzy logic identification, concrete steps are:

Step 1: wind profile radar echo power spectrum is carried out pre-service; Adopt the Fast Median Filtering method, be specially: the medium filtering window that to set up a width be 2N+1, and move along digital signal sequences, step-length is 1; Before window moves, obtain the intermediate value of 2N+1 sequence in the window earlier; After window moves, find out the position in the sequence that the data object in the new immigration window sequenced in front, obtain the intermediate value of asking.Window width is got 5 points, i.e. N=2 observes the variation of data in the window when moving for adjacent twice of medium filtering window, when medium filtering window during in first position; Burst is { x (i+1), x (i+2), x (i+3), x (i+4) in the window; X (i+5) }, establish its size ordering and be { w (i+1), w (i+2), w (i+3); W (i+4), w (i+5) }, then intermediate value is w (3).Window moves right after one, and data are { x (i+2), x (i+3) in the window; X (i+4), x (i+5), x (i+6) }; At first get rid of x (i+1) relative { w (j), j=1,2; ...; 5} is then with x (i+6) and remaining { w (j), j=1; 2; ..., each object contrast ordering in the 5} sequence,

that obtain thus in this sequence of new sequence

is the intermediate value of asking.

Step 2: turbulence signal and the noise signal difference aspect spectrum continuity, spectral intensity, spectral position and spectrum width: the turbulent flow echo has continuity, spectral position frequency displacement variation continuously, spectral intensity and spectrum width big; The ground clutter spectral position is positioned near the zero-frequency; The intermittence clutter does not have continuity, and spectral position is big at some range bin appearance, spectral intensity and spectrum width; The radio frequency interference spectral position is fixed, and spectral intensity and spectrum width are little.Through going out the intermittence clutter to the analysis of echo spectrum is distinguishable; Can distinguish radio frequency interference to spectral intensity and spectrum width analysis; Can slacken the ground clutter influence to spectral position about the zero-frequency symmetry analysis.Can tell the turbulent flow echo through above-mentioned analysis-by-synthesis.

Utilize the fuzzy logic recognition methods to extract the turbulent flow echo then.Set up subordinate function characteristic quantity is carried out Fuzzy processing, judge whether belong to the turbulent flow echo according to the fuzzy basic shared weight of each characteristic.Specifically be meant:

1) the spectral intensity normalization that the power spectrum symmetry is disappeared mutually, the spectral intensity after the normalization are P _Y(f _Ij), f _IjRepresent the corresponding Doppler shift in j position on the i range bin.P _Y(f _Ij) big more, the degree that is under the jurisdiction of the turbulent flow echo is big more; Otherwise it is more little.Set up " S " type subordinate function on this basis to the spectral intensity Fuzzy processing:

$S\left[P_Y\left(f_{\mathrm{ij}}\right)\right]=\left\{\begin{array}{c}0,P_Y\left(f_{\mathrm{ij}}\right)\∈[-\mathrm{1,0.3}]\\ \frac{10}{3}\·P_Y\left(f_{\mathrm{ij}}\right)-1,P_Y\left(f_{\mathrm{ij}}\right)\∈\left[\mathrm{0.3,0.6}\right]\\ 1,P_Y\left(f_{\mathrm{ij}}\right)\∈\left[\mathrm{0.6,1}\right]\end{array}\right.---\left(1\right)$

2) set up the Gaussian subordinate function to composing the continuity Fuzzy processing:

$G(w=j-[\stackrel{\‾}{f_{i-1}}\left]\right)=\exp [-\ln \left(\frac{5}{3}\right)\·w^2/\mathrm{\Δ}{\stackrel{\‾}{f}}^2],\frac{n_S}{2}-\left[\stackrel{\‾}{f_{i-1}}\right]\≤w\≤\frac{n_S}{2}+1-\left[\stackrel{\‾}{f_{i-1}}\right]---\left(2\right)$

3) through after the subordinate function Fuzzy processing, convert wind profile radar echo power spectrum to two fuzzy bases: the fuzzy base of spectral intensity (P _Y(f _Ij)) and the fuzzy base of spectrum continuity

In fuzzy logic system, the identification that utilizes fuzzy base to replace the substitution value of the power spectrum of actual detection to carry out system is handled, so problem concentrates in the foundation of subordinate function and rule of fuzzy system.Suppose in the radar power spectrum separately between the turbulent flow echo and various clutter, concerning two fuzzy bases, represent with the one dimension subordinate function:

$T_{\mathrm{ij}}=w_1\·S\left[P_Y\left(f_{\mathrm{ij}}\right)\right]+w_2\·G(j-[\stackrel{\‾}{f_{i-1}}\left]\right),\underset{i=1}{\overset{2}{\mathrm{\Σ}}}{w}_i=1---\left(3\right)$

Adopt the method for non-equal weight, wherein, w ₁, w ₂Represent spectral intensity and the successional weight factor of spectrum respectively, for the T of i range bin _IjBe worth big more, the P that it is corresponding _Y(f _Ij) be under the jurisdiction of the turbulent flow echo degree big more.

Step 3: after identifying the turbulent flow target echo through fuzzy logic, need to demarcate turbulent flow echo signal bandwidth so that carry out the square data estimation.Because the echoed signal generation mechanism of wind profile radar is mainly turbulent scattering; Suppose to be in the radar pulse volume RANGE IN HOMOGENEOUS ISOTROPIC TURBULENCE; When the statistical property of research turbulent flow percent ripple; Its probability distribution is approximately Gaussian, therefore uses Gauss's least-square fitting approach match turbulent flow echo spectrum to distribute:

$P_i\left(f\right)=P_{i0}\exp [-{(f-\stackrel{\‾}{f_i})}^2/{2\mathrm{\σ}}_i^2]+P_N---\left(4\right)$

P _N, P _I0,

Peak power, first moment and second moment for noise level, i range bin.About the boundary discussion of turbulent flow echo and noise jamming overlapping, introduce the degree of tilt analysis:

$g\left(f_{\mathrm{ij}}\right)=\frac{\frac{\mathrm{dp}\left(f_{\mathrm{ij}}\right)}{\mathrm{df}}}{\max \left(\right|\frac{\mathrm{dp}\left(f_{\mathrm{ij}}\right)}{\mathrm{df}}\left|\right)}$

Spectrum peak position left side g (f _Ij)＜0, right side g (f _Ij)＞0 is judged the boundary of turbulent flow echo and clutter thus.Draft unpolluted turbulent flow echo spectrum point on each range bin, carry out least square fitting based on these spectrum points:

P(f)＝Af ²+Bf+C (5)

By the least square curve fitting principle, parameter A, B, C confirm that method is:

$\left[\begin{array}{ccc}\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{f}_i^2& \underset{i=1}{\overset{n}{\mathrm{\Σ}}}{f}_i& n\\ \underset{i=1}{\overset{n}{\mathrm{\Σ}}}{f}_i^3& \underset{i=1}{\overset{n}{\mathrm{\Σ}}}{f}_i^2& \underset{i=1}{\overset{n}{\mathrm{\Σ}}}{f}_i\\ \underset{i=1}{\overset{n}{\mathrm{\Σ}}}{f}_i^4& \underset{i=1}{\overset{n}{\mathrm{\Σ}}}{f}_i^3& \underset{i=1}{\overset{n}{\mathrm{\Σ}}}{f}_i^2\end{array}\right]\left[\begin{array}{c}A\\ B\\ C\end{array}\right]=\left[\begin{array}{c}\underset{i=1}{\overset{n}{\mathrm{\Σ}}}P\left(f_i\right)\\ \underset{i=1}{\overset{n}{\mathrm{\Σ}}}{f}_i\·P\left(f_i\right)\\ \underset{i=1}{\overset{n}{\mathrm{\Σ}}}{f}_i^2\·P\left(f_i\right)\end{array}\right]---\left(6\right)$

N representes that the turbulent flow echo spectrum of treating match counts, and obtains each square parameter thus and is respectively:

$P_{i0}={10}^{(\frac{B^2}{4A}-C)/10};$ ${\stackrel{\‾}{f}}_i=-\frac{B}{2A};$ ${\mathrm{\σ}}_i^2=\frac{5\log 10\left(e\right)}{A}$

To can obtaining turbulent flow echoed signal power through the signal spacing integration after the least square fitting, the ratio of signal power and noise power can draw the turbulent flow echo signal to noise ratio (S/N ratio) that accuracy has improved.

As shown in Figure 1; Discern turbulent flow target echo process flow diagram based on fuzzy logic method among the present invention; To carry out spectral intensity and spectrum continuity subordinate function Fuzzy processing through the spectrum of the echo power after the Fast Median Filtering; Convert it into fuzzy base of spectral intensity and the fuzzy base of spectrum continuity, come judgment signal to be under the jurisdiction of the degree of turbulent flow echo according to the weight of two fuzzy bases.T is big more, the p (f that it is corresponding _Ij) be under the jurisdiction of the turbulent flow echo degree big more, through training test spectrum data processing, when T＞0.6, can think the p (f that it is corresponding _Ij) be the turbulent flow target echo.

Like the left figure of Fig. 2 is the original echo spectrogram that on June 4th, 2008,10:42:16 east wave beam detected, and right figure is the echo spectrogram that utilizes behind the fuzzy logic recognition methods impurity elimination ripple, and horizontal ordinate is that Fourier counts, and ordinate is a range bin.Can find out, through fuzzy logic identification, reject near the ground clutter the low range bin and zero-frequency among the left figure, also having removed with fixing frequency displacement simultaneously is the radio frequency interference of characteristic, obtains the cleaner turbulent flow echo that does not receive noise jamming.After utilizing fuzzy logic to identify turbulence signal, utilize Gauss's least square fitting demarcation signal bandwidth calculation square data, thereby carry out the synthetic of atmospheric wind.

Like the left figure of Fig. 3 is the WIND FIELD EVOLUTION figure that on June 4th, 2008, original echo calculated, and right figure utilizes this method to handle the WIND FIELD EVOLUTION figure that calculate the back.Horizontal ordinate is time (was the interval with one hour), and ordinate is a height.Handle through the present invention, not only make wind field average survey height by before 10 kms bring up to 13 kms afterwards, filled up " blank " that six km height places in the original graph do not detect wind simultaneously.

Claims (5)

1. based on the wind profile radar echo spectrum reconstructing method of fuzzy logic identification, it is characterized in that may further comprise the steps:

1) wind profile radar echo power spectrum is carried out pre-service;

2) according to turbulence signal in the echoed signal and clutter (like ground clutter, intermittence clutter, radio frequency interference) signal in the difference aspect spectrum continuity, spectral intensity, spectral position and the spectrum width, utilize the fuzzy logic recognition methods to extract the turbulent flow echo;

3) calculate the square data of wind profile radar turbulent flow echoed signal through Gauss's least-square fitting approach: noise level, signal to noise ratio (S/N ratio), first moment (Doppler shift) and second moment (spectrum width).

2. the wind profile radar echo spectrum reconstructing method based on fuzzy logic identification according to claim 1; It is characterized in that: the Fast Median Filtering method is adopted in the pre-service described in the step 1; Be specially: the medium filtering window that to set up a width be 2N+1; And move along digital signal sequences, step-length is 1; Before window moves, obtain the intermediate value of 2N+1 sequence in the window earlier; After window moves, find out the position in the sequence that the data object in the new immigration window sequenced in front, obtain the intermediate value of asking.

3. the wind profile radar echo spectrum reconstructing method based on fuzzy logic identification according to claim 1 is characterized in that: the turbulence signal described in the step 2 and noise signal refer in the difference of composing aspect continuity, spectral intensity, spectral position and the spectrum width: it is big that the turbulent flow echo has continuity, spectral position frequency displacement variation continuously, spectral intensity and a spectrum width; The ground clutter spectral position is positioned near the zero-frequency; The intermittence clutter does not have continuity, and spectral position occurs at some range bin; The spectral position of radio frequency interference is fixed, and spectral intensity and spectrum width are little.

4. the wind profile radar echo spectrum reconstructing method based on fuzzy logic identification according to claim 1 is characterized in that: the fuzzy logic recognition methods described in the step 2 is extracted the turbulent flow echo and is meant:

1) spectral intensity that the power spectrum symmetry is disappeared is mutually carried out normalization, and the spectral intensity after the normalization is P _Y(f _Ij), P _Y(f _Ij) value big more, the degree that is under the jurisdiction of the turbulent flow echo is big more; Otherwise P _Y(f _Ij) more little, the degree that is under the jurisdiction of the turbulent flow echo is more little; Set up " S " type subordinate function on this basis to the spectral intensity Fuzzy processing:

$S\left[P_Y\left(f_{\mathrm{ij}}\right)\right]=\left\{\begin{array}{c}0,P_Y\left(f_{\mathrm{ij}}\right)\∈[-\mathrm{1,0.3}]\\ \frac{10}{3}\·P_Y\left(f_{\mathrm{ij}}\right)-1,P_Y\left(f_{\mathrm{ij}}\right)\∈\left[\mathrm{0.3,0.6}\right]\\ 1,P_Y\left(f_{\mathrm{ij}}\right)\∈\left[\mathrm{0.6,1}\right]\end{array}\right.---\left(1\right)$

2) set up the Gaussian subordinate function to composing the continuity Fuzzy processing:

$G(w=j-[\stackrel{\‾}{f_{i-1}}\left]\right)=\exp [-\ln \left(\frac{5}{3}\right)\·w^2/\mathrm{\Δ}{\stackrel{\‾}{f}}^2],\frac{n_S}{2}-\left[\stackrel{\‾}{f_{i-1}}\right]\≤w\≤\frac{n_S}{2}+1-\left[\stackrel{\‾}{f_{i-1}}\right]---\left(2\right)$

3), convert wind profile radar echo power spectrum to two fuzzy bases, the i.e. fuzzy base of spectral intensity (P through after the subordinate function Fuzzy processing _Y(f _Ij)) and the fuzzy base of spectrum continuity

Suppose between turbulent flow echo in the radar power spectrum, ground clutter, intermittence clutter, the radio frequency interference separately, concerning two fuzzy bases, represent with the one dimension subordinate function:

$T_{\mathrm{ij}}=w_1\·S\left[P_Y\left(f_{\mathrm{ij}}\right)\right]+w_2\·G(j-[\stackrel{\‾}{f_{i-1}}\left]\right),\underset{i=1}{\overset{2}{\mathrm{\Σ}}}{w}_i=1---\left(3\right)$

w ₁, w ₂Represent spectral intensity and the successional weight factor of spectrum, the T of i range bin respectively _IjBe worth big more, the P that it is corresponding _Y(f _Ij) be under the jurisdiction of the turbulent flow echo degree big more.

5. the wind profile radar echo spectrum reconstructing method based on fuzzy logic identification according to claim 1 is characterized in that: step 3 uses Gauss's least-square fitting approach match turbulent flow echo spectrum to distribute:

$P_i\left(f\right)=P_{i0}\exp [-{(f-\stackrel{\‾}{f_i})}^2/{2\mathrm{\σ}}_i^2]+P_N---\left(4\right)$

P _N, P _I0,

Be peak power, first moment and the second moment of noise level, i range bin, fitting formula is: P (f)=Af ²+ Bf+C, parameter A, B, C are confirmed by following formula:

$\left[\begin{array}{ccc}\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{f}_i^2& \underset{i=1}{\overset{n}{\mathrm{\Σ}}}{f}_i& n\\ \underset{i=1}{\overset{n}{\mathrm{\Σ}}}{f}_i^3& \underset{i=1}{\overset{n}{\mathrm{\Σ}}}{f}_i^2& \underset{i=1}{\overset{n}{\mathrm{\Σ}}}{f}_i\\ \underset{i=1}{\overset{n}{\mathrm{\Σ}}}{f}_i^4& \underset{i=1}{\overset{n}{\mathrm{\Σ}}}{f}_i^3& \underset{i=1}{\overset{n}{\mathrm{\Σ}}}{f}_i^2\end{array}\right]\left[\begin{array}{c}A\\ B\\ C\end{array}\right]=\left[\begin{array}{c}\underset{i=1}{\overset{n}{\mathrm{\Σ}}}P\left(f_i\right)\\ \underset{i=1}{\overset{n}{\mathrm{\Σ}}}{f}_i\·P\left(f_i\right)\\ \underset{i=1}{\overset{n}{\mathrm{\Σ}}}{f}_i^2\·P\left(f_i\right)\end{array}\right]---\left(5\right)$

N representes that the turbulent flow echo spectrum of treating match counts, and obtains each square parameter thus and is respectively:

$P_{i0}={10}^{(\frac{B^2}{4A}-C)/10};$ ${\stackrel{\‾}{f}}_i=-\frac{B}{2A};$ ${\mathrm{\σ}}_i^2=\frac{5\log 10\left(e\right)}{A};$

To the signal spacing integration after the least square fitting is turbulent flow echoed signal power, and the ratio of signal power and noise power can draw the turbulent flow echo signal to noise ratio (S/N ratio) that accuracy has improved.

Images