CN104166129A - Real beam radar iteration minimum mean square error angle super-resolution method - Google Patents

Abstract

The invention discloses a real beam radar iteration minimum mean square error angle super-resolution method. The problem that a traditional real beam radar bearing angle is low in resolution is solved. A minimum mean square error objective function of orientation echo signals is established and is solved, the computing relation of orientation signal power and a covariance matrix is used for establishing an iteration expression, estimation on an orientation target is gradually close to a real target position by iterative computing with the method, and real beam radar azimuth angle super resolution is achieved.

Description

A kind of real Beam radar iteration least mean-square error angle ultra-resolution method

Technical field

The invention belongs to Radar Technology field, be specifically related to real beam scanning radar angle super-resolution.

Background technology

Real Beam radar is because the restriction that is not subject to platform motor pattern can be widely used in military and and the civil areas such as target tracking, terrain-avoidance and rescue at sea.

Real Beam radar wave beam scans and is detected region equably in orientation, and obtains the reflective information of target on dimensional orientation.As a kind of conventional radar working system, real Beam radar is not restricted by platform motor pattern, all has broad application prospects in civil and military field.But the angular resolution of real Beam radar mainly determines by radar aperture, aperture is larger, and resolution is higher.And in practical application, limited by platform size and manufacture craft, the aperture of real Beam radar is limited.Therefore, the angular resolution of real Beam radar often can not meet the demand of types of applications.

Improve the angular resolution of radar and can improve target detection and follow the trail of precision, improve ability and the rescue at sea ability of landform identification, have great importance to realizing that complex-terrain in precise guidance, aircraft travel track is evaded etc.

For the platform of regular motion, can adopt the technology of synthetic aperture to form virtual large aperture, thereby obtain high angular resolution.Document: Oliver C J.Synthetic aperture radar imaging[J] .Journal of Physics D:Applied Physics, 1989,22 (7): 871., the second order doppler phase that utilizes the relative motion of regular motion platform and side-looking imaging region internal object to produce, use matched filtering technique to realize orientation to angle high-resolution, but for platform motion forward sight or rear viewed area, and static or irregular movement platform, the method is inapplicable.

Monopulse technology be another kind be applicable to motion platform pattern orientation to angle ultra-resolution method, this technology based on difference beam angle measuring principle, can be applied to platform forward sight and rear viewed area.Document: " airborne radar monopulse forward sight imaging algorithm " (" Chinese image graphics " 2010,15 (3): P462 469) adopts monopulse angle measurement technique to carry out orientation to angle super-resolution imaging.But this technology is only applicable to isolated strong point target, for there are multiple scattering centers, particularly multiple scattering centers are distributed in the situation in single real wave beam, and monopulse processing can cause serious angle scintillations phenomenon to occur.

Deconvolution Method is that the third orientation that is applicable to motion platform pattern is to angle ultra-resolution method, this technology is utilized the statistical property establishing target function of the interior noise of imaging region and target, by solving and construct iterative formula, realize the high angular resolution of original object is restored.Document: Daolin Z, Yulin H, Jianyu Y.Radar angular superresolution algorithm based on Bayesian approach[C] //Signal Processing (ICSP), 2010 IEEE 10th International Conference on.IEEE, 2010:1894 1897. adopt Deconvolution Method to carry out orientation to angle super-resolution processing, but the method is to noise-sensitive, be only applicable to high s/n ratio environment, in the time that echoed signal signal to noise ratio (S/N ratio) is lower, orientation significantly declines to angle super-resolution performance.

For static platform, contrary synthetic aperture technique is to realize a kind of effective ways of position angle super-resolution.Document: Prickett M J, Chen C C.Principles of inverse synthetic aperture radar/ISAR/imaging[C] //EASCON'80; Electronics and Aerospace Systems Conference.1980,1:340 345., utilize doppler phase that tangential motion produces between static platform and moving target to realize position angle high-resolution, surveyed but the method is not suitable for static target high-resolution.

Summary of the invention

Of the present invention for the problems referred to above, a kind of real Beam radar iteration least mean-square error angle ultra-resolution method has been proposed, solve real Beam radar orientation to the low technical barrier of angular resolution, and by iterative processing, under low signal-to-noise ratio environment, still can realize noise is effectively suppressed, prevent the appearance of false target.

The real Beam radar iteration least mean-square error of one provided by the invention angle ultra-resolution method, specifically comprises the following steps:

S1: imaging system two dimension echo generates, and specifically comprises the following steps:

S11: echo is carried out to coherent demodulation:

$S(t,\mathrm{\τ})=\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{\mathrm{\σ}}_l\·\mathrm{\α}({\mathrm{\θ}}_l,\mathrm{\τ})\·\mathrm{rect}(t-\frac{2R_0}{c})\·\exp (-j\frac{4\mathrm{\π}}{\mathrm{\λ}}\·R_0)\·\exp \left(\mathrm{j\π}{K}_a{[t-\frac{{2R}_0}{c}]}^2\right)$

Wherein, S (t, τ) represents echo coherent demodulation signal, rect () represents Distance Time window, and α () represents orientation time window, and t represents that distance is to time variable, τ represents that orientation is to time variable, and λ represents the wavelength that transmits, R ₀represent the distance between motion platform and target, K _arepresent the time chirp rate transmitting, c represents the light velocity; θ _lrepresent the location parameter of target, σ _lrepresent range parameter;

S12: the orientation time arrow T of scanning radar imaging region _afor:

T _a＝[-PRI·N _a/2,-PRI·(N _a/2-1),…,PRI·(N _a/2-1)]；

Wherein, PRI represents indicating impulse recurrence interval, N _arepresent that echoed signal orientation is to sampling number;

S13: the Distance Time vector T of scanning radar imaging region _rfor:

T _r＝[-1/f _r·N _r/2,-1/f _r·(N _r/2-1),…,1/f _r·(N _r/2-1)]；

Wherein, f _rrepresent that distance is to sampling rate, N _rrepresent that echoed signal distance is to sampling number;

S2: echo data distance, to pulse compression, specifically comprises step by step following:

S21: echoed signal is carried out to distance to process of pulse-compression, obtain the echoed signal S (f of distance to frequency domain, orientation to time domain by distance to FFT _r, τ);

$S(f_r,\mathrm{\τ})=\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{\mathrm{\σ}}_l\·\mathrm{\α}({\mathrm{\θ}}_l,\mathrm{\τ})\·\mathrm{rect}\left(\frac{f_r}{B}\right)\·\exp \{-j\frac{4\mathrm{\π}(f_c+f_r)}{c}{R}_0\}\·\exp \left\{\mathrm{j\π}\frac{f_r^2}{K_a}\right\};$

S22: structure distance is to matched filtering function H (f _r);

$H\left(f_r\right)=\exp (-\mathrm{j\π}\frac{f_r^2}{K_a});$

S23: by H (f _r) multiply each other and obtain distance after the Range compress echoed signal S to frequency domain, orientation to time domain with echoed signal ₁(f _r, τ);

$S_1(f_r,\mathrm{\τ})=\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{\mathrm{\σ}}_l\·a({\mathrm{\θ}}_l,\mathrm{\τ})\·\mathrm{rect}\left(\frac{f_r}{B}\right)\·\exp \{-j\frac{4\mathrm{\π}(f_c+f_r)}{c}{R}_0\};$

S24: to S ₁(f _r, τ) carry out distance and obtain two-dimensional time-domain signal s to IFFT conversion ₂(t, τ);

$S_2(t,\mathrm{\τ})\≈\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{\mathrm{\σ}}_l\·a({\mathrm{\θ}}_l,\mathrm{\τ})\·\exp (-j\frac{4\mathrm{\π}}{\mathrm{\λ}}{R}_0)\·\sin c\left[B(\mathrm{\τ}-\frac{{2R}_0}{c})\right];$

S3: real Beam radar orientation is to signal modeling, obtains orientation and to echoed signal vector y is:

y＝H(θ)f+n；

Wherein, H (θ) represents direction matrix, and f represents the amplitude information of orientation to dispersive target, and n represents additional noise vector;

Described orientation is to echoed signal vector y=[y ₁..., y _k];

Described orientation is to the amplitude information f=[f of dispersive target ₁..., f _n];

S4: structure least mean-square estimate, specifically comprises step by step following:

S41: for orientation echoed signal, the weighting matrix w of structure K × N dimension, asks optimum weighting matrix:

minE{|f-w ^Hy| ²}；

Wherein, E{} represents mean value computation, () ^hthe conjugate transpose of representing matrix or vector, || ²represent the quadratic sum of vectorial each element,

S42: obtain the optimum solution of objective function about weighting matrix:

w＝(H(θ)PH(θ) ^H+R _n) ^-1H(θ)P

Wherein, H (θ) PH (θ) ^hrepresent the covariance matrix of signal, for noise covariance matrix, represent signal power spectrum matrix;

S42: solve the estimated value about orientation target

$\hat{f}={(H\left(\mathrm{\θ}\right)\mathrm{PH}{\left(\mathrm{\θ}\right)}^H+R_n)}^{-1}H\left(\mathrm{\θ}\right)\mathrm{Py};$

S5: iteration least mean-square estimate, by building iteration expression formula to realize the accurate estimation of orientation target;

f _σ+1＝(H(θ)P _σH(θ) ^H+R _n} ^-1H(θ)P _ky；

Wherein σ represents iterations,

S6: judge whether iteration restrains;

S7: real field angle super-resolution imaging.

Beneficial effect of the present invention: a kind of real Beam radar iteration least mean-square error of the present invention angle ultra-resolution method, utilize known antenna directional diagram and real beamformer system parameter information, establishing target function on the basis of weighted least mean square error criterion, accurately calculate and approach the angle super-resolution result that realistic objective distributes by differentiate and structure iterative algorithm, thereby solved the low problem of motion platform radar forward sight zone of action internal object azimuth resolution.Advantage of the present invention is compared with conventional angular ultra-resolution method, and the method is not restricted by platform motor pattern, and algorithm performance is sane under low signal-to-noise ratio.The present invention can be applied to the field such as target tracking, rescue at sea.

Attached caption

Fig. 1 is the inventive method FB(flow block).

Fig. 2 is the real beam scanning radar imagery schematic diagram of embodiment of the present invention.

Fig. 3 is the simulation objectives scene arrangenent diagram that the present invention adopts while specifically implementing.

Fig. 4 adds SNR=30dB white Gaussian noise figure apart from the data after pulse pressure.

Fig. 5 be after pulse pressure data add SNR=30dB white Gaussian noise along orientation to sectional view.

Fig. 6 implements in the present invention method to 3 results that point target is carried out angle super-resolution processing in Fig. 3.

Fig. 7 be the result of corresponding diagram 6 along orientation to sectional view.

Embodiment

, be described further content of the present invention below in conjunction with accompanying drawing.

First the present invention generates the signal echo model of real beam scanning radar, recycling echoed signal and antenna direction graphic sequence, on the basis of minimum mean square error criterion, set up the objective function distributing about solving imaging scene scattering coefficient, and derive about scattering coefficient distribute rough estimate operator expression formula, according to the pass series structure iterative equation of orientation target power spectrum and covariance matrix, process the position angle super-resolution imaging of finally having realized scanning radar by iteration self-adapting subsequently.

Be illustrated in figure 2 the real beam scanning radar imagery schematic diagram of embodiment of the present invention.

The present invention mainly adopts the method for emulation experiment to verify, institute in steps, conclusion all on Matlab2010 checking correct.With regard to embodiment, the present invention is described in further detail below.

As shown in Figure 1, concrete steps are as follows for the solution of the present invention process flow diagram:

First to imaging region arbitrfary point target, calculate target and motion platform apart from course, real beam scanning radar point target simulation parameter is set, corresponding radar imaging system parameter is as shown in table 1.

Table 1

Parameter	Symbol	Numerical value
			Carrier frequency	f c	10GHz
Bandwidth	B	20MHz
			Wide while transmitting	T	50μs

Transmitted signal bandwidth	B	40MHz
			Podium level	H	5Km
Impulse sampling frequency	PRF	1000Hz
			Antenna scanning speed	ω	30°/s
Antenna beamwidth	θ	3°
			Sweep limit	Φ	-8°～8°

S1: the position of real beam antenna be designated as (0,0, h), wherein, 0,0 and h be respectively the x-axis, y-axis and z-axis coordinate of receiving station; The position angle of the relative platform of target is designated as θ; Radar antenna downwards angle of visibility is designated as radar antenna sweep velocity is designated as ω; R ₀for the distance between motion platform and target, real Beam radar sweep limit is designated as Φ, and real Beam radar beam angle is designated as φ.

As shown in Figure 3, way orbicular spot is the point target of layout and ground 3 × 3 to the imaging scene that this example adopts, and along Y-axis positive dirction, amplitude is followed successively by 1,0.8,0.8, and interval is respectively 6 ° and 1.5 °; Be spaced apart 500m along X-direction.Texas tower initial time position coordinates is (0,0,5km).XOY plane internal object scattering function is designated as f (x, y).Point in t moment XOY plane and radar (x, y) platform distance are designated as R ₀.

Wherein, suppose in scanning area same distance R ₀amplitude all on the different azimuth sampling location at place, the location parameter that makes these targets is θ=(θ ₁, θ ₂... θ _l), range parameter is σ=(σ ₁, σ ₂..., σ _l), radar emission signal is linear FM signal, the echo of the scanning radar zone of action can be designated as S (t, τ) through coherent demodulation

S2: the echo matrix S (t, τ) of step S1 is carried out to distance to FFT, then according to transmitting chirp rate K _ato reference time t, construct distance to pulse pressure reference function at frequency domain with distance by S (f _r, τ) multiply each other with pulse pressure reference function, complete distance to pulse compression, the two-dimentional echo data of the distance after pulse pressure to frequency domain orientation to time domain is expressed as S ₁(f _r, τ).The row distance of going forward side by side obtains two-dimensional time-domain signal S to IFFT conversion ₂(t, τ).In order to simulate the actual conditions that have noise, at data S ₂in (t, τ), add the white Gaussian noise of SNR=30dB.Corresponding result as shown in Figure 4, along orientation to section as shown in Figure 5.

S3: utilize systematic parameters such as sweep velocity, pulse-recurrence time, antenna beamwidth that system arranges etc. to generate direction vector h (θ _n) and direction matrix H (θ).

S4: for orientation echoed signal, the weighting matrix w of structure K × N dimension, becomes problem and ask optimum weighting matrix under minimum mean square error criterion,

minE{|f-w ^Hy| ²} (8)

Wherein E () represents mean value computation, () ^hthe conjugate transpose of representing matrix or vector, || ²the quadratic sum that represents vectorial each element, solves about minE{|f-w ^hy| ²formula objective function is about the derivative of w and to make it be zero, can obtain objective function about the optimum solution of weighting matrix to be

w＝(E{yy ^H}) ^-1E{yf ^H}＝(E{yy ^H}) ^-1E{(Hf+n)f ^H}

Due to signal and noise separate, therefore, can abbreviation be

w＝(E{yy ^H}) ^-1HE{ff ^H}

Meanwhile, due to E{yy ^hrepresent the covariance matrix of echo, and

E{yy ^H}＝HPH ^H+R _n

Wherein HPH ^hrepresent the covariance matrix of signal, and for signal power spectrum matrix; for noise covariance matrix, therefore, the optimum solution of weighting matrix finally can be expressed as

w＝(HPH ^H+R _n) ^-1HP

The weighting matrix that utilization calculates, can solve the estimated value about orientation target;

S5: utilize formula calculate the rough estimate evaluation of covariance matrix, and by result of calculation substitution equation (E{yy ^h) ^-1in HPy, calculate the estimated value of orientation target due to can substitution in obtain new signal power spectrum matrix P ₁, therefore, utilize matrix P and equation $\hat{f}={({\mathrm{HPH}}^H+R_n)}^{-1}\mathrm{HPy}$ Structure iterative formula

f _σ+1＝(HP _σH ^H+R _n) ^-1HP _σy

Wherein, σ is iterations.

S6: minimal value ε is set, as σ+1 time iteration result f _σ+1deduct iteration result f σ time _σwhile being less than ε,

|f _σ+1-f _σ|≤ε

End iteration and record the result that this iteration result is this range unit.

S7: utilize step 3 line by line range unit processing one by one to be carried out in whole real beam scanning radar imagery region to the method for step 6, finally obtain whole motion platform scanning radar imaging region internal object angle super-resolution result.Imaging results as shown in Figure 6, Figure 7.

As can be seen from the figure, method provided by the invention can significantly improve real beam scanning radar target angular resolution.All effective to multiple targets in single target or same wave beam, the real Beam radar iteration least mean-square error of one provided by the invention angle ultra-resolution method, can realize the target angle super-resolution imaging in real beam scanning radar imagery region.Result has the good effect of improving for the angle information of target, multiple targets of differentiating in same wave beam.This area engineering technical personnel can make corresponding application according to technology disclosed by the invention, and relevant application is still within protection scope of the present invention.

Those of ordinary skill in the art will appreciate that, embodiment described here is in order to help reader understanding's principle of the present invention, should be understood to that protection scope of the present invention is not limited to such special statement and embodiment.For a person skilled in the art, the present invention can have various modifications and variations.Within the spirit and principles in the present invention all, any amendment of doing, be equal to replacement, improvement etc., within all should being included in claim scope of the present invention.

Claims (8)

1. a real velocity of wave radar iteration least mean-square error angle ultra-resolution method, is characterized in that, comprising:

S1: imaging system two dimension echo generates, and specifically comprises the following steps:

S11: echoed signal is carried out to coherent demodulation:

Wherein, S (t, τ) represents echo coherent demodulation signal, rect () represents Distance Time window, and α () represents orientation time window, and t represents that distance is to time variable, τ represents that orientation is to time variable, and λ represents the wavelength that transmits, R ₀represent the distance between motion platform and target, K _arepresent the time chirp rate transmitting, c represents the light velocity; L represents the number that transmits, θ _lrepresent the sensing position of antenna when l transmits, σ _lrepresent range parameter;

S12: the orientation time arrow T of scanning radar imaging region _afor:

T _a＝[-PRI·N _a/2,-PRI·(N _a/2-1),…,PRI·(N _a/2-1)]；

Wherein, PRI represents indicating impulse recurrence interval, N _arepresent that echoed signal orientation is to sampling number;

S13: the Distance Time vector T of scanning radar imaging region _rfor:

T _r＝[-1/f _r·N _r/2,-1/f _r·(N _r/2-1),…,1/f _r·(N _r/2-1)]；

Wherein, f _rrepresent that distance is to sampling rate, N _rrepresent that echoed signal distance is to sampling number;

S2: echo data distance, to pulse compression, specifically comprises step by step following:

S21: echoed signal is carried out to distance to process of pulse-compression, obtain the echoed signal S (f of distance to frequency domain, orientation to time domain by distance to FFT _r, τ):

S22: structure distance is to matched filtering function H (f _r):

S23: by H (f _r) multiply each other and obtain distance after the Range compress echoed signal S to frequency domain, orientation to time domain with echoed signal ₁(f _r, τ):

S24: to S ₁(f _r, τ) carry out distance and obtain two-dimensional time-domain signal S to IFFT conversion ₂(t, τ):

S3: real Beam radar orientation is to signal modeling, obtains orientation and to echoed signal vector y is:

y＝H(θ)f+n；

Wherein, H (θ) represents direction matrix, and f represents the amplitude information of orientation to dispersive target, and n represents additional noise vector;

Described orientation is to echoed signal vector y=[y ₁..., y _k];

Described orientation is to the amplitude information f=[f of dispersive target ₁..., f _n];

S4: structure least mean-square estimate, specifically comprises step by step following:

S41: for orientation echoed signal, the weighting matrix w of structure K × N dimension, asks optimum weighting matrix:

minE{|f-w ^Hy| ²}；

Wherein, E{} represents mean value computation, () ^hthe conjugate transpose of representing matrix or vector, || ²represent the quadratic sum of vectorial each element;

S42: obtain the optimum solution of objective function about weighting matrix:

w＝(H(θ)PH(θ) ^H+R _n) ^-1H(θ)P

Wherein, H (θ) PH (θ) ^hrepresent the covariance matrix of signal, represent signal power spectrum matrix, R _nfor noise covariance matrix;

S42: solve the estimated value about orientation target

S5: iteration least mean-square estimate, by building iteration expression formula to realize the accurate estimation of orientation target;

f _σ+1＝(H(θ)P _σH(θ) ^H+R _n) ^-1H(θ)P _ky；

Wherein σ represents iterations,

S6: judge whether iteration restrains;

S7: real field angle super-resolution imaging.

2. according to claim 1 real velocity of wave radar iteration least mean-square error angle ultra-resolution method, it is characterized in that, described in described described

3. according to claim 1 real velocity of wave radar iteration least mean-square error angle ultra-resolution method, it is characterized in that direction matrix described in S3

Wherein, h (θ _n) expression direction vector, [h ₁, h ₂..., h _d] ∈ R ^{l × 1}for antenna direction graphic sequence, D represents the sampling number of a beam angle, and K represents that orientation is to sampling number, and D represents single wave beam sampling number.

4. according to claim 3 real velocity of wave radar iteration least mean-square error angle ultra-resolution method, it is characterized in that K=N+D-1.

5. according to claim 4 real velocity of wave radar iteration least mean-square error angle ultra-resolution method, it is characterized in that, described orientation is to sampling number

Wherein, PRF indicating impulse repetition frequency, ω represents sweep velocity, Φ represents sweep limit.

6. according to claim 5 real velocity of wave radar iteration least mean-square error angle ultra-resolution method, it is characterized in that described single wave beam sampling number

7. according to claim 1 real velocity of wave radar iteration least mean-square error angle ultra-resolution method, it is characterized in that described signal power spectrum matrix .

8. according to claim 1 real velocity of wave radar iteration least mean-square error angle ultra-resolution method, it is characterized in that, described step S6 judges whether iteration restrains, and concrete determination methods is: a given minimal value ε is as threshold values, and if only if adjacent twice iteration result f _σ+1with f _σwhile meeting the following formula condition of convergence, judge iteration convergence,

|f _σ+1-f _σ|≤ε

When double iteration result | f _σ+1-f _σ| when > ε, return to step 5 and proceed iterative computation, until meet | f _σ+1-f _σ| till≤ε;

Wherein, σ represents iterations.