CN103760531A - Method for restraining noise of aerial target echoes of narrow-band radar - Google Patents

Abstract

The invention belongs to the technical field of restraining of noise of aerial target echoes of narrow-band radar and discloses a method for restraining the noise of the aerial target echoes of the narrow-band radar. The method comprises the steps that the narrow-band radar receives an aerial target echo signal and conducts sliding window processing on the aerial target echo signal, and then a post-sliding window processing matrix is obtained; the average power M of a noise signal in a time domain is estimated; an autocorrelation matrix R of the post-sliding window processing matrix E is obtained; the feature vectors of the R are calculated; a feature vector matrix U is formed by the feature vectors of the R; a noise threshold thresh _ Alpha is set and then is subtracted from Lambda r; if Lambda (r-1) - thresh _ Alpha > 0 and Lambda r - thresh _ Alpha < 0 when the value of r is P, a matrix Us is formed by the first P columns of the U; please see the specification for the expression of Es; the aerial target echo signal x which is processed with noise restraining is obtained according to elements of the first column of the matrix Es and elements of the last row of the matrix Es.

Description

A kind of noise suppressing method of aerial target echo of Narrow-band Radar

Technical field

The invention belongs to the noise reduction techniques field of Narrow-band Radar aerial target echo, particularly a kind of noise suppressing method of aerial target echo of Narrow-band Radar.

Background technology

In modern war, due to the needs of combat duty and environment, helicopter, propeller aeroplane and jet plane are being born different tasks, and the threaten degree having is different, therefore realizes the detection of this three classes Aircraft Targets is had very important significance.

So far, in the noise suppressing method of existing aerial target echo, conventionally use training data and test data, and the signal to noise ratio (S/N ratio) condition of training data and test data is more or less the same.But in actual conditions, training data is all by some cooperative experiments or the high s/n ratio data obtained by emulation conventionally; Yet test data is in most cases all obtained under some noncooperative conditions, for example, in war process, the distance of the complex environment in battlefield, weather conditions and target range radar all can cause the difference of noise intensity in test data.Like this, bring certain difficulty will to the squelch of aerial target echo.

Summary of the invention

The object of the invention is to propose a kind of noise suppressing method of aerial target echo of Narrow-band Radar.The noise suppressing method of the aerial target echo of this Narrow-band Radar can carry out squelch effectively to aerial target echo, can be used for Narrow-band Radar aerial target echo to carry out pre-service.

For realizing above-mentioned technical purpose, the present invention adopts following technical scheme to be achieved.

A noise suppressing method for the aerial target echo of Narrow-band Radar, comprises the following steps:

S1: described Narrow-band Radar receives the echoed signal of aerial target, the pulse accumulation number of the echoed signal of described aerial target is N, and the echoed signal of described aerial target is carried out to slide window processing, window length is

draw matrix E after slide window processing,

represent that N/2 rounds downwards; The average power M of estimated noise signal in time domain;

S2: draw the autocorrelation matrix R of matrix E after slide window processing,

wherein, E ^hthe complex-conjugate transpose matrix that represents E; Then eigenwert and the proper vector of calculating R, the eigenwert of R is followed successively by λ from big to small ₁extremely

with the proper vector composition characteristic vector matrix U of R, the r of U classifies the r proper vector u of R as _r, r get 1 to

u _rfor with λ _rcharacteristic of correspondence vector;

S3: noise gate thresh_ α is set, uses λ _rdeduct thresh_ α; When if r gets P, λ _r-1-thresh_ α >0, λ _r-thresh_ α <0, forms matrix U with the front P row of U _s;

S4: structural matrix E _s,

extract matrix E _smiddle first row element is as column vector e ₁, at matrix E _sin the element of middle last column, extract secondary series to the element of last row as row vector

draw the echoed signal x of the aerial target after squelch is processed,

Feature of the present invention and further improvement are:

In step S1, the echoed signal of the aerial target that described Narrow-band Radar receives is expressed as matrix S, S=[S ₁..., S _n] ^t, wherein, S _kthe echoed signal of aerial target is at k signal constantly, and k gets 1 to N; Matrix after slide window processing

wherein,

for matrix S is carried out the signal obtaining after the L time slide window processing,

l get 1 to

In step S1, the echoed signal of described aerial target is carried out to discrete Fourier transformation, the echoed signal that makes described aerial target is frequency-region signal, according to described frequency-region signal, the average power of estimated noise signal in frequency domain; Then by described noise signal, the average power in frequency domain is counted divided by discrete Fourier transformation, draws the average power M of noise signal in time domain.

In step S3,

represent that the card side that degree of freedom is 2 is distributed in the upside fractile on horizontal α.

Beneficial effect of the present invention is: the noise suppressing method of the aerial target echo of Narrow-band Radar of the present invention can carry out squelch effectively to aerial target echo, can be used for Narrow-band Radar aerial target echo to carry out pre-service.

Accompanying drawing explanation

Fig. 1 is the schematic flow sheet of noise suppressing method of aerial target echo of a kind of Narrow-band Radar of the embodiment of the present invention;

The Doppler frequency spectrum schematic diagram that Fig. 2 a is jet plane when Noise not;

Fig. 2 b is the Doppler frequency spectrum schematic diagram of jet plane when Noise;

Fig. 2 c is the Doppler frequency spectrum schematic diagram of jet plane after using the present invention to suppress noise signal;

The Doppler frequency spectrum schematic diagram that Fig. 3 a is propeller aeroplane when Noise not;

Fig. 3 b is the Doppler frequency spectrum schematic diagram of propeller aeroplane when Noise;

Fig. 3 c is for being used the present invention noise signal to be suppressed to the Doppler frequency spectrum schematic diagram of rear screw shaft aircraft;

The Doppler frequency spectrum schematic diagram that Fig. 4 a is helicopter when Noise not;

Fig. 4 b is the Doppler frequency spectrum schematic diagram of helicopter when Noise;

Fig. 4 c is the Doppler frequency spectrum schematic diagram of helicopter after using the present invention to suppress noise signal;

Fig. 5 be under different threshold condition average reconstructed error with the robustness curve synoptic diagram of signal to noise ratio (S/N ratio).

Embodiment

Below in conjunction with accompanying drawing, the invention will be further described:

With reference to Fig. 1, it is the schematic flow sheet of the noise suppressing method of the aerial target echo of a kind of Narrow-band Radar of the embodiment of the present invention.The noise suppressing method of the aerial target echo of this Narrow-band Radar comprises the following steps:

S1: by sliding window method, process the echoed signal of the aerial target of Narrow-band Radar reception, and the average power of estimated noise signal in time domain.Detailed process is as follows:

Narrow-band Radar receives the echoed signal of aerial target, and the echoed signal of the aerial target that for example Narrow-band Radar receives is Low SNR signal.The echoed signal of the aerial target that Narrow-band Radar is received is expressed as matrix S, S=[S ₁..., S _n] ^t, wherein, S _kthe echoed signal of aerial target is at k signal constantly, and k gets 1 to N, the pulse accumulation number of the echoed signal that N is aerial target.

The echoed signal of above-mentioned aerial target is carried out to slide window processing, draw matrix E after slide window processing; Window length is

draw matrix E after slide window processing, represent that N/2 rounds downwards; Matrix after slide window processing

wherein,

for matrix S is carried out the signal obtaining after the L time slide window processing,

l get 1 to

[.] ^tthe transposition of representing matrix.

The echoed signal of above-mentioned aerial target is carried out to discrete Fourier transformation, the echoed signal that makes above-mentioned aerial target is frequency-region signal, according to above-mentioned frequency-region signal, estimate the average power of corresponding noise signal in frequency domain, above-mentioned noise signal is arranged in the echoed signal of above-mentioned aerial target.Average power by above-mentioned noise signal in frequency domain is counted divided by discrete Fourier transformation, draws the average power M of noise signal in time domain, M=σ ².

S2: draw the autocorrelation matrix of matrix after slide window processing, and draw eigenwert and the proper vector of this autocorrelation matrix.Its detailed process is as follows:

First according to matrix E after slide window processing, draw the autocorrelation matrix R of matrix E after slide window processing,

wherein, E ^hcomplex-conjugate transpose matrix for E.Then autocorrelation matrix R is made to Eigenvalues Decomposition, draw eigenwert and the proper vector of this autocorrelation matrix R.The eigenwert of autocorrelation matrix R is followed successively by λ from big to small ₁extremely

now,

also can be expressed as

c and d all get 1 to

and c>d.

The proper vector composition characteristic vector matrix U of autocorrelation matrix R,

the columns of U is

the r of U classifies the r proper vector u of R as _r, r get 1 to

u _rfor with λ _rcharacteristic of correspondence vector.Now can also draw characteristic of correspondence value matrix Λ,

S3: noise gate is set, and determines the number of principal component according to this noise gate.Its detailed process is as follows:

First noise gate thresh_ α is set, and noise gate thresh_ α can take various ways, for example

the upside fractile that the card side that expression degree of freedom is 2 is distributed on horizontal α (is χ ²(2) the upside fractile on horizontal α).The following describes thresh_ α is set to

reason: suppose the real part of above-mentioned noise signal and the Gaussian distribution that imaginary part is all obeyed zero-mean, w=w ₁+ jw ₂, wherein, w ₁for the real part of above-mentioned noise signal, w ₂imaginary part for above-mentioned noise signal.? $w_1~N(0,\frac{{\mathrm{\&sigma;}}^2}{2}),w_2~N(0,\frac{{\mathrm{\&sigma;}}^2}{2}),$ After deriving, can draw:

$\frac{w_1^2+w_2^2}{{\mathrm{\&sigma;}}^2/2}~{\mathrm{\&chi;}}^2\left(2\right)$

$P(\frac{w_1^2+{\mathrm{<figure-callout\; id="2"\; label="w"\; filenames="HDA0000461040910000022.png,HDA0000461040910000024.png"\; state="\{\{state\}\}">w</figure-callout>}}_2^2}{{\mathrm{\&sigma;}}^2/2}>w)=\mathrm{\&alpha;}$

$w_1^2+{\mathrm{<figure-callout\; id="2"\; label="w"\; filenames="HDA0000461040910000022.png,HDA0000461040910000024.png"\; state="\{\{state\}\}">w</figure-callout>}}_2^2=\frac{{\mathrm{\&sigma;}}^2}{2}{\mathrm{\&chi;}}_{\mathrm{\&alpha;}}^2\left(2\right)$

Hence one can see that, and above-mentioned noise signal is obeyed the χ that degree of freedom is 2 ²(2) distribute, according to χ ²(2) distribute that noise gate thresh_ α is set is comparatively reasonable.

Then, use successively λ ₁extremely deduct thresh_ α, until draw an eigenvalue λ that is less than thresh_ α _p, i.e. λ _p-1-thresh_ α >0, λ _p-thresh_ α <0, like this, the number that P is principal component (or signal subspace dimension).

Then, with the front P row of U, form matrix U _s, U _s=[u ₁..., u _p].

S4: draw the echoed signal (namely signal being projected to signal subspace) of the aerial target after squelch is processed, its detailed process is as follows:

Structural matrix E _s,

for U _sassociate matrix.

Wherein, e _ijfor matrix E _sthe element of the capable j of i row, i get 1 to

j get 1 to

Extract matrix E _smiddle first row element is as column vector e ₁, at matrix E _sin the element of middle last column, extract secondary series to the element of last row as row vector

according to column vector e ₁and row vector

draw the echoed signal x of the aerial target after squelch is processed,

Below by emulation experiment, the invention will be further described:

The noisy measured signal of helicopter, propeller aeroplane and jet plane that in emulation experiment, packet used receives containing conventional Narrow-band Radar.In emulation experiment, use the present invention to carry out squelch to the echoed signal of above-mentioned three class aircrafts, wherein, and when noise gate is set, α=0.10 or α=0.01.After use the present invention carries out squelch to the echoed signal of above-mentioned three class aircrafts, calculate the average reconstructed error of all samples.With reference to Fig. 2 a, for the Doppler frequency spectrum schematic diagram of jet plane when the Noise not, with reference to Fig. 2 b, be the Doppler frequency spectrum schematic diagram of jet plane when the Noise, with reference to Fig. 2 c, it is the Doppler frequency spectrum schematic diagram of jet plane after using the present invention to suppress noise signal; With reference to Fig. 3 a, for the Doppler frequency spectrum schematic diagram of propeller aeroplane when the Noise not, with reference to Fig. 3 b, be the Doppler frequency spectrum schematic diagram of propeller aeroplane when the Noise, with reference to Fig. 3 c, for using the present invention noise signal to be suppressed to the Doppler frequency spectrum schematic diagram of rear screw shaft aircraft; With reference to Fig. 4 a, for the Doppler frequency spectrum schematic diagram of helicopter when the Noise not, with reference to Fig. 4 b, be the Doppler frequency spectrum schematic diagram of helicopter when the Noise, with reference to Fig. 4 c, it is the Doppler frequency spectrum schematic diagram of helicopter after using the present invention to suppress noise signal.From above-mentioned figure, can find out: the noise suppressing method of the aerial target echo by Narrow-band Radar of the present invention, can suppress preferably to the noise containing in the echoed signal of above-mentioned three class aircrafts.

With reference to Fig. 5, for average reconstructed error under different threshold condition is with the robustness curve synoptic diagram of signal to noise ratio (S/N ratio).In Fig. 5, the α of thresholding 1 correspondence is 0.10, and the α of thresholding 2 correspondences is 0.01.As can be seen from Figure 5, utilize the signal after squelch is processed that the present invention obtains more close to Noise signal not, average reconstructed error is less., from Fig. 5, it can also be seen that, the selection of thresholding is larger on result impact meanwhile, therefore, when the method for using the present invention to introduce, select suitable thresholding just can obtain more desirable result.

Obviously, those skilled in the art can carry out various changes and modification and not depart from the spirit and scope of the present invention the present invention.Like this, if within of the present invention these are revised and modification belongs to the scope of the claims in the present invention and equivalent technologies thereof, the present invention is also intended to comprise these changes and modification interior.

Claims (4)

1. a noise suppressing method for the aerial target echo of Narrow-band Radar, is characterized in that, comprises the following steps:

S1: described Narrow-band Radar receives the echoed signal of aerial target, the pulse accumulation number of the echoed signal of described aerial target is N, and the echoed signal of described aerial target is carried out to slide window processing, window length is draw matrix E after slide window processing,

represent that N/2 rounds downwards; The average power M of estimated noise signal in time domain;

S2: draw the autocorrelation matrix R of matrix E after slide window processing,

wherein, E ^hthe complex-conjugate transpose matrix that represents E; Then eigenwert and the proper vector of calculating R, the eigenwert of R is followed successively by λ from big to small ₁extremely

with the proper vector composition characteristic vector matrix U of R, the r of U classifies the r proper vector u of R as _r, r get 1 to u _rfor with λ _rcharacteristic of correspondence vector;

S3: noise gate thresh_ α is set, uses λ _rdeduct thresh_ α; When if r gets P, λ _r-1-thresh_ α >0, λ _r-thresh_ α <0, forms matrix U with the front P row of U _s;

S4: structural matrix E _s,

extract matrix E _smiddle first row element is as column vector e ₁, at matrix E _sin the element of middle last column, extract secondary series to the element of last row as row vector

draw the echoed signal x of the aerial target after squelch is processed,

2. the noise suppressing method of the aerial target echo of a kind of Narrow-band Radar as claimed in claim 1, is characterized in that, in step S1, the echoed signal of the aerial target that described Narrow-band Radar receives is expressed as matrix S, S=[S ₁..., S _n] ^t, wherein, S _kthe echoed signal of aerial target is at k signal constantly, and k gets 1 to N; Matrix after slide window processing

wherein,

for matrix S is carried out the signal obtaining after the L time slide window processing,

l get 1 to

3. the noise suppressing method of the aerial target echo of a kind of Narrow-band Radar as claimed in claim 1, it is characterized in that, in step S1, the echoed signal of described aerial target is carried out to discrete Fourier transformation, the echoed signal that makes described aerial target is frequency-region signal, according to described frequency-region signal, the average power of estimated noise signal in frequency domain; Then by described noise signal, the average power in frequency domain is counted divided by discrete Fourier transformation, draws the average power M of noise signal in time domain.

4. the noise suppressing method of the aerial target echo of a kind of Narrow-band Radar as claimed in claim 1, is characterized in that, in step S3,

represent that the card side that degree of freedom is 2 is distributed in the upside fractile on horizontal α.

Images