CN102628937B - Radar detection method based on generalized keystone transformation and non-coherent accumulation - Google Patents

Abstract

The invention discloses a radar detection method based on generalized keystone transformation and non-coherent accumulation, wherein the method mainly solves the detection problem of a high speed and highly accelerated target in a strong Gaussian white noise background. The method is implemented by the following steps: (1) generating radar target multiple-pulse echoes with Gaussian white noise added, and performing frequency domain pulse compression; (2) correcting range curvature of the pulse-compressed echoes by using the generalized keystone transformation; (3) performing fuzzy compensation on the curvature-corrected echoes by using a fuzzy factor; (4) performing non-coherent accumulation on the fuzzy-compensated echoes, and performing velocity compensation and slow time dimensional quadratic term compensation on the echoes by using a target velocity obtained in the non-coherent accumulation; (5) performing coherent accumulation on the slow time dimensional quadratic term compensated echoes to obtain the target to be detected. The method provided in the invention can eliminate effects of the range curvature and migration of the echoes on energy accumulation, can reduce computation of velocity search, and can be used in the detection of the target in the strong Gaussian white noise background.

Description

Radar detecting method based on General keystone transform and the accumulation of non-coherent

Technical field

The invention belongs to Radar Technology field, relate to object detection method, can be used for efficiently processing based on the detections of radar of coherent pulse echo repeatedly in white Gaussian noise environment.

Background technology

In Radar Targets'Detection field, can run into the situation that target to be detected is high-speed motion, this class target has very high speed and acceleration conventionally, traditional method is that a plurality of pulse echos are directly carried out to slow time dimension FFT, if can there is not range walk when target is low-speed motion, thereby the energy accumulation of each target echo can be got up reach by target detection object out.But because target is high-speed motion, make, at pulse internal object integration time, larger range walk to occur, and the high acceleration of target also can cause the range curvature of echo envelope, if paired pulses echo directly carries out slow time dimension FFT, the energy accumulation of same range unit can only be got up, signal gross energy is dispersed on a plurality of range units, the peak value of gained pulse accumulation result is much lower when there is not range walk and range curvature, thereby can not obtain good testing result, even can lose objects when serious, existing research introduction can be got rid of the range walk of target generation to accumulating the impact of result by keystone method, but for reaching good accumulation result, the range curvature of the echo envelope that the high acceleration while also needing target travel causes is removed the impact of accumulation result, also do not have at present good method to solve.

In the situation that movable information the unknown of target, if obtain the velocity information of target by the method for accumulation, usual method is, first set very large speed search scope and certain speed search step-length, the echo envelope that then can cause according to each search speed of the calculating number of walking about, paired pulses echo carries out envelope ring shift, again the pulse echo of carrying out after envelope ring shift is done to non-coherent accumulation, the result finally each possible search speed being accumulated compares, maximum speed corresponding to accumulation result is the velocity estimation value of target.The result of doing is like this that need to take very high search arithmetic amount be cost, therefore, is necessary to look for a kind of method that reduces operand.

Summary of the invention

The object of the invention is to the deficiency for above-mentioned prior art, a kind of radar detecting method based on General keystone transform and the accumulation of non-coherent is proposed, to reduce the operand of target velocity search, get rid of the impact of high acceleration on target energy accumulation, thereby be conducive to improve radar to the high detection performance of accelerating target of high speed.

For achieving the above object, radar target detection method of the present invention, comprises the steps:

1) approximate model and the noise model of radar high speed high acceleration moving target are expressed as

and d (l, t _m);

2), according to target approximate model and noise model, obtain radar pulse echo:

And paired pulses echo with the fast time for variable, be FFT and transform to frequency of distance territory, obtain echo apart from frequency domain data z (f, the t of dimension _m), wherein f represents frequency of distance;

3) by frequency domain data z (f, the t of distance dimension _m) along distance dimension, do frequency domain pulse compression, the data after pulse compression are designated as to z ' (f, t _m), and to the data z ' after this pulse compression (f, t _m) in slow time variable t _mwith new time variable τ _mreplace, variable alternate form is

the data that variable obtains after replacing are designated as z ' (f, τ _m);

4) setting fuzzy factor scope is [n _min, n _max], n wherein _minminimum, n _maxmaximum, N altogether, and be all positive integer, data z ' (f, τ after with the fuzzy factor within the scope of this, variable being replaced _m) carrying out fuzzy compensation, the echo data after compensation is designated as z ' _i(f, τ _m), i=1 ... N;

5) to the echo frequency domain data z ' after fuzzy compensation _i(f, τ _m), i=1 ... N carries out the time domain interpolation that in each range unit, p is ordered, and the data after time domain interpolation are designated as

i=1 ... N;

6) utilize the data after time domain interpolation, when calculating target velocity and being v, echo envelope range walk is counted the umber of pulse w in k and same range unit _k;

7) according to target friction speed ripple range walk number k and same range unit umber of pulse w next time _kdifference, the data after time domain interpolation are carried out to ring shift, and the N group echo data after ring shift are carried out to non-coherent accumulation, obtain N and organize non-coherent accumulation result, be designated as

i=1 ... N, wherein || expression be modulo operation;

8) N is organized to non-coherent accumulation result

i=1 ... N compares, and gets and wherein accumulates the velocity ambiguity factor of n that the fuzzy factor corresponding to result of peak value maximum is target _kestimated value, be designated as

and speed corresponding to the accumulation peak value of getting this maximum is fuzzy rear target initial velocity v ₀estimated value, be designated as

9) initial velocity estimated value after the objective fuzzy that utilization obtains

with fuzzy factor estimated value

calculate the estimated value of target initial velocity

then by the target initial velocity estimated value of estimating to obtain

to with fuzzy factor estimated value corresponding echo frequency numeric field data

carry out velocity compensation, then the echo data after compensation is transformed to apart from time domain, the distance time domain data after conversion is designated as

10) to converting later distance time domain data by the method for Dechirp, carry out the search of frequency modulation rate, obtain frequency modulation rate γ _aestimated value, be designated as

and then obtained the estimated value of target travel acceleration

11) use the estimated value of the aimed acceleration obtaining

to the data after velocity compensation

carry out the compensation of slow time dimension quadratic phase item, and the data after compensation are done to the coherent accumulation of slow time dimension, the peak value of coherent accumulation is the target of detection.

The present invention, owing to utilizing the method for pulse accumulation to carry out target detection, can effectively suppress noise; Simultaneously owing to utilizing General keystone transform correction target envelope range curvature, make the result of non-coherent accumulation after correction distance bending more approach target echo actual energy; In addition owing to utilizing the method for velocity ambiguity compensation to dwindle the hunting zone of target initial velocity, the non-coherent accumulation operand after General keystone transform is reduced greatly.

Below in conjunction with accompanying drawing, working of an invention is described in detail.

Accompanying drawing explanation

Fig. 1 is the general flow chart of realizing of the present invention;

Fig. 2 is the testing result figure that adopts existing conventional sense method;

Fig. 3 is the testing result figure that adopts existing optimum detection methodology;

Fig. 4 is the testing result figure that adopts the inventive method.

Embodiment

With reference to Fig. 1, specific implementation step of the present invention is as follows:

Step 1, the radar target multiple-pulse echo of generation plus noise.

1a) the approximate model of radar target multiple-pulse echo:

$s(\hat{t},t_m)=A_0\mathrm{rect}\left[\frac{t-\frac{2\hat{R}\left(t_m\right)}{c}}{T_p}\right]\exp \left[\mathrm{j\&pi;\&gamma;}{(\hat{t}-\frac{2R\left(t_m\right)}{c})}^2\right]\exp [-\mathrm{<figure-callout\; id="4"\; label="j"\; filenames="HDA0000155747140000031.png"\; state="\{\{state\}\}">j</figure-callout>}\frac{4\mathrm{\&pi;}{f}_{c}R\left(t_m\right)}{c}]\exp (-\mathrm{<figure-callout\; id="4"\; label="j"\; filenames="HDA0000155747140000031.png"\; state="\{\{state\}\}">j</figure-callout>}\frac{{4\mathrm{\&pi;f}}_{c}v\hat{t}}{c})$

In formula, $\mathrm{rect}\left(u\right)=\left\{\begin{array}{cc}1& \left|u\right|\&le;1/2\\ 0& \left|u\right|>1/2\end{array}\right.,$ T _pbe pulsewidth, γ is frequency modulation rate,

be the fast time, t is full-time, and m is integer, T _rpulse-recurrence time, f _cfor radar emission signal carrier frequency, t _m=mT _rthe slow time, A ₀that scattering point returns wave intensity,

t _mtarget is to the distance of radar, R constantly ₀for target initial time distance, v is target velocity, and a is aimed acceleration, and c is the light velocity;

1b) noise model:

By the noise in continuous N time echo take matrix representation as:

wherein, l is range unit label, and u (l) represents the noise power of l range unit, g (l, t _m) be the noise phase of l range unit in m pulse, column vector g (l) adds up the independently multiple Gaussian random variable of M dimension zero-mean with u (l), and M represents transponder pulse number, m=1,2 ..., M;

1c) utilize the approximate model of radar target multiple-pulse echo

and noise generate the radar target multiple-pulse echo of plus noise: $z(\hat{t},t_m)=s(\hat{t},t_m)+d(\hat{t},t_m).$

Step 2, is FFT to the radar target multiple-pulse echo of plus noise along distance dimension.

Because more Range cell migration and range curvature at the direct processing target in Distance Time territory are inconvenient, so paired pulses echo

will be with the fast time

for variable, be FFT and transform to frequency of distance territory, obtain frequency domain data z (f, t _m), what wherein FFT represented is Fast Fourier Transform (FFT), f represents frequency of distance.

Step 3, frequency domain data z (f, the t of paired pulses echo _m) carry out pulse compression.

3a) by frequency domain data z (f, the t of following formula paired pulses echo _m) do frequency domain pulse compression:

$H\left(\hat{t}\right)=\mathrm{rect}\left(\frac{\hat{t}}{T_p}\right)\exp (-\mathrm{j\&pi;\&gamma;}{\hat{t}}^2)$

In formula,

indicating impulse compression function, $\mathrm{rect}\left(u\right)=\left\{\begin{array}{cc}1& \left|u\right|\&le;1/2\\ 0& \left|u\right|>1/2\end{array}\right.,$

the fast time, T _pbe pulsewidth, γ is frequency modulation rate;

3b) frequency of distance of the pulse echo after pulse compression-slow time domain data are designated as to z ' (f, t _m), this packet has contained target echo and noise information, and this example only carries out theoretical analysis to target echo, therefore write out the concrete form of the later target echo signal of pulse pressure, is:

$s_1(f,t_m)=A^{\&prime;}\mathrm{rect}\left(\frac{f}{\mathrm{\&Delta;f}}\right)\exp (-j2\mathrm{\&pi;f}(\frac{{2R}_0}{c}+\frac{f_d}{\mathrm{\&gamma;}}))\exp (-j2\mathrm{\&pi;}\frac{f_d(f_c+f)}{f_c}{t}_m)$

$\&CenterDot;\exp (-j\frac{{\mathrm{\&pi;\&gamma;}}_a(f_c+f)}{f_c}{t}_m^2),$

In formula, A ' expression is done the echo strength after frequency pulse pressure to target echo, and Δ f is radar emission signal bandwidth,

for the Doppler frequency of target, λ is radar wavelength, and target is to the distance of radar

r ₀for target initial time distance, v is target velocity, and a is aimed acceleration, t _mthe slow time, f _cfor radar emission signal carrier frequency, c is the light velocity;

Due to target high-speed motion, make the envelope of target echo signal that the skew being directly proportional to Doppler frequency can occur, the distance of its skew is

but the distance of this skew is identical for each echo, therefore do not consider.

Step 4, utilizes General keystone transform correction target echo range curvature.

4a) due to the speed of target, and the repetition frequency of radar transmitted pulse is lower, therefore can cause target generation velocity ambiguity, therefore the Doppler frequency of target is expressed as:

f _d＝f _d0+n _kPRF ，

In formula, f _d0the later Doppler frequency of velocity ambiguity, n _kbe fuzzy factor, PRF is pulse repetition rate;

4b) by the expression of target echo after the expression substitution pulse compression of target Doppler frequency, the new expression of target echo obtaining is:

$s_2(f,t_m)=A^{\&prime;}\mathrm{rect}\left(\frac{f}{\mathrm{\&Delta;f}}\right)\exp (-j2\mathrm{\&pi;f}(\frac{{2R}_0}{c}+\frac{f_d}{\mathrm{\&gamma;}}))\exp (-j2\mathrm{\&pi;}\frac{f_{d0}(f_c+f)}{f_c}{t}_m)$

$\&CenterDot;\exp (-j\frac{{2\mathrm{\&pi;n}}_k\mathrm{fPRF}}{f_c}{t}_m)\exp (-j2\mathrm{\&pi;PRF}{n}_k{t}_m)\exp (-j\frac{{\mathrm{\&pi;\&gamma;}}_a(f_c+f)}{f_c}{t}_m^2),$

In formula, A ' expression is done the echo strength after frequency pulse pressure to target echo, and Δ f is radar emission signal bandwidth,

for the Doppler frequency of target, λ is radar wavelength, and target is to the distance of radar

r ₀for target initial time distance, v is target velocity, and a is aimed acceleration, γ _adoppler frequency rate, t _mthe slow time, f _cfor radar emission signal carrier frequency, c is the light velocity;

According to the phase place of the 4th exponential term in the new expression of target echo after pulse compression, be 4c) integral multiple of 2 π, its value is 1, therefore after being removed, obtain the reduced representation formula s that target echo is new ₃(f, t _m) be:

$s_3\left({f,t}_m\right)=A^{\&prime;}\mathrm{rect}\left(\frac{f}{\mathrm{\&Delta;f}}\right)\exp (-j2\mathrm{\&pi;f}(\frac{{2R}_0}{c}+\frac{f_d}{\mathrm{\&gamma;}}))\exp (-j2\mathrm{\&pi;}\frac{f_{d0}(f_c+f)}{f_c}{t}_m)$

$\&CenterDot;\exp (-j\frac{{2\mathrm{\&pi;n}}_k\mathrm{fPRF}}{f_c}{t}_m)\exp (-j\frac{{\mathrm{\&pi;\&gamma;}}_a(f_c+f)}{f_c}{t}_m^2);$

4c) because target travel acceleration is very large, therefore pulse echo easily produces range curvature, thereby can affect the result that pulse energy accumulates, for eliminating this impact, introduce General keystone transform, i.e. data z ' (f, t after paired pulses compression _m) slow time variable t _mintroduce variable alternate form pulse echo data after variable is replaced are designated as z ' (f, τ _m);

4d) target echo signal s after paired pulses compression ₃(f, t _m) carry out after General keystone transform the target echo signal s obtaining ₄(f, τ _m) expression be:

$s_4\left({f,t}_m\right)=A^{\&prime;}\mathrm{rect}\left(\frac{f}{\mathrm{\&Delta;f}}\right)\exp (-j2\mathrm{\&pi;f}(\frac{{2R}_0}{c}+\frac{f_d}{\mathrm{\&gamma;}}))\exp (-j2\mathrm{\&pi;}{f}_{d0}\frac{(f_c+f)}{f_c}\sqrt{\frac{f_c}{f_c+f}}{\mathrm{\&tau;}}_m)$

$\&CenterDot;\exp (-j\frac{{2\mathrm{\&pi;n}}_k\mathrm{fPRF}}{f_c}\sqrt{\frac{f_c}{f_c+f}}{\mathrm{\&tau;}}_m)\exp (-j{\mathrm{\&pi;\&gamma;}}_a{\mathrm{\&tau;}}_m^2).$

Step 5, carries out fuzzy compensation to the echo data after General keystone transform.

5a) paired pulses echo data z ' (f, τ _m) in target echo signal expression analyze knownly, its 3rd exponential term is velocity ambiguity item, needs it to compensate in this example;

5b) target setting fuzzy factor n _kscope [n _min, n _max], n wherein _minfor minimum fuzzy factor estimated value, n _maxfor maximum fuzzy factor estimated value, total N, and be all integer, utilize the fuzzy factor within the scope of this, constructed fuction G (f, τ _m; n _k) to the data z ' after General keystone transform (f, τ _m) utilize following formula to carry out fuzzy compensation:

$G(f,{\mathrm{\&tau;}}_m,n_k)=\exp \left(j\frac{{2\mathrm{\&pi;n}}_{k}f}{f_c}\sqrt{\frac{f_c}{{f+f}_c}}{\mathrm{PRF\&tau;}}_m\right),$

In formula, G (f, τ _m; n _k) expression fuzzy compensation function, n _krepresent fuzzy factor, τ _mbe the slow time after variable changes, f represents frequency of distance, f _cfor radar emission signal carrier frequency, PRF is the repetition frequency of radar emission signal, and the pulse echo data after fuzzy compensation are designated as to z ' _i(f, τ _m), i=1 ... N.

5c) with fuzzy compensation function G (f, τ _m; n _k) to the target echo signal s obtaining after General keystone transform ₄(f, τ _m) carry out fuzzy compensation, the target echo signal s after the fuzzy compensation obtaining ₅(f, τ _m) expression be:

$s_5(f,{\mathrm{\&tau;}}_m)=A^{\&prime;}\mathrm{rect}\left(\frac{f}{\mathrm{\&Delta;f}}\right)\exp (-j2\mathrm{\&pi;f}(\frac{{2R}_0}{c}+\frac{f_d}{\mathrm{\&gamma;}}))\exp (-j2\mathrm{\&pi;}{f}_{d0}\frac{(f_c+f)}{f_c}\sqrt{\frac{f_c}{f_c+f}}{\mathrm{\&tau;}}_m)$

$\&CenterDot;\exp (-j{\mathrm{\&pi;\&gamma;}}_a{\mathrm{\&tau;}}_m^2).$

Step 6, to the pulse echo data z ' after fuzzy compensation _i(f, τ _m), i=1 ... N carries out apart from time domain interpolation, is about to the data z ' of fuzzy compensation afterpulse echo _i(f, τ _m), i=1 ... N carries out the time domain interpolation of p point in each range unit, p=1, and 2,3, the data after time domain interpolation are designated as

i=1 ... N, time domain Interpolation Theory and method refer to < < radar imagery technology > >.

Step 7, when calculating target velocity is v, the range walk of echo envelope is counted the umber of pulse w in k and same range unit _k.

7a) the pulse echo data after General keystone transform are after carrying out velocity ambiguity compensation, and target velocity scope is 0≤v≤v _maxtherefore, only need estimate the speed within the scope of this, calculate the fuzzyyest speed of target

according to the fuzzyyest speed v obtaining _max, calculating target echo envelope range walk number is l=(p+1) * round (2v _maxf _s/ cPRF) the umber of pulse w and in same range unit _l=M/l, wherein PRF represents radar transmitted pulse repetition frequency, and λ represents radar wavelength, and what round represented is the operation that rounds rounding up, f _sfor signal sampling frequency;

7b) according to target echo envelope range walk, count the umber of pulse w in l and same range unit _l, determine speed search step delta v:

First, get two value l of arbitrary neighborhood within the scope of range walk number 0～l-1 ₁and l ₂, l ₁< l ₂;

Secondly, calculating generation range walk number is l ₁time corresponding target velocity

be l with there is range walk number ₂time corresponding target velocity $v_2=\frac{{\mathrm{cl}}_2\mathrm{PRF}(p+1)}{2\mathrm{\&Delta;fM}};$

Then, with there is range walk number, be l ₂time corresponding target velocity v ₂be l with there is range walk number ₁time corresponding target velocity v ₁difference as speed search step delta v;

7c) take Δ v as step-length, obtain 0≤v≤v _maxall speed v within the scope of this, and the range walk of computing velocity target echo envelope while being v counts k=(p+1) * round (2v/ (cPRF) Δ fM), c represents the light velocity, Δ f represents radar emission signal bandwidth;

While 7d) being v according to speed, the range walk of target echo envelope is counted k, obtains the umber of pulse w in same range unit _k=M/k, M indicating impulse number.

Step 8, to the echo data after time domain interpolation

i=1 ... N carries out ring shift:

First, take target velocity umber of pulse w in same range unit during as v _kfor segment length, M pulse is divided into k group;

Then, the k group data after dividing are carried out to the ring shift of 0～k-1 position successively along distance dimension.

Step 9, carries out non-coherent accumulation to the echo data after ring shift, and soon the mould value in all pulse respective distances unit of the data after N group ring shift is added

i=1 ... N, completes the accumulation of non-coherent, and now N organizes non-coherent accumulation result and can occur a peak value at a certain range unit.

Step 10, target velocity is estimated.

10a) will own [n _min, n _max] N that obtains after fuzzy factor compensation organizes non-coherent accumulation result

i=1 ... N compares, and selects fuzzy factor corresponding to the accumulation result of peak value maximum as the estimated value of the fuzzy factor of target

and get speed corresponding to this peak-peak as objective fuzzy after the estimated value of speed calculate the estimated value of target initial velocity

$\hat{v}=\frac{\mathrm{\&lambda;}(\hat{{\mathrm{<figure-callout\; id="0"\; label="f\; d"\; filenames="HDA0000155747140000021.png,HDA0000155747140000031.png"\; state="\{\{state\}\}">f</figure-callout>}}_d}+\widehat{n_k}\mathrm{PRF})}{2},$

In formula,

the estimated value that represents the Doppler frequency after target generation velocity ambiguity, λ represents radar wavelength,

the estimated value that represents fuzzy factor;

10b) by the estimated value with fuzzy factor

the data of corresponding fuzzy compensation afterpulse echo are designated as

Step 11, the echo data after fuzzy to General keystone transform compensating for doppler

carry out velocity compensation.

11a) after fuzzy to the crooked also compensating for doppler of echo data correction distance, by pulse echo data z ' (f, τ _m) in target echo expression can find out, there is linear coupling with the slow time in its second exponential term speed, the impact bringing for eliminating this coupling pulsed energy accumulation, is similar to

wherein o () represents that high-order is infinitely small, and desin speed penalty function is:

$P(f,{\mathrm{\&tau;}}_m;{\hat{f}}_{d0})=\exp \left(j\frac{\mathrm{\&pi;f}}{f_c}{\hat{f}}_{\mathrm{<figure-callout\; id="0"\; label="d"\; filenames="HDA0000155747140000021.png,HDA0000155747140000031.png"\; state="\{\{state\}\}">d</figure-callout>}0}{\mathrm{\&tau;}}_m\right),$

In formula,

represent velocity compensation function,

represent the just estimated value of Doppler frequency of fuzzy rear target, f represents frequency of distance, τ _mit is the slow time after variable is replaced;

11b) use velocity compensation function

pulse echo data are fallen in compensation

in frequency of distance and the exponential term of doppler velocity coupling, the pulse echo data after velocity compensation are designated as

11c) use velocity compensation function

to target echo signal s ₅(f, τ _m) carry out velocity compensation, obtain the target echo signal s after velocity compensation ₆(f, τ _m), be expressed as:

$s_6(f,{\mathrm{\&tau;}}_m)=A^{\&prime;}\mathrm{rect}\left(\frac{f}{\mathrm{\&Delta;f}}\right)\exp (-j2\mathrm{\&pi;f}(\frac{{2R}_0}{c}+\frac{f_d}{\mathrm{\&gamma;}}))\exp (-j2{\mathrm{\&pi;<figure-callout\; id="0"\; label="f\; d"\; filenames="HDA0000155747140000021.png,HDA0000155747140000031.png"\; state="\{\{state\}\}">f</figure-callout>}}_d{\mathrm{\&tau;}}_m)\exp \left({-\mathrm{j\&pi;\&gamma;}}_a{\mathrm{\&tau;}}_m^2\right),$

In formula, the echo strength of A ' expression target, Δ f is radar emission signal bandwidth, γ _adoppler frequency rate,

for the Doppler frequency of target, λ is radar wavelength, f _d0represent the Doppler frequency after fuzzy, R ₀for target initial time distance, v is target velocity, and a is aimed acceleration, τ _mthe slow time after variable is replaced, f _cfor radar emission signal carrier frequency, c is the light velocity;

11d) to the pulse echo data after velocity compensation

be distance dimension IFFT, convert the signal into apart from time domain, obtain the distance-slow time domain data of pulse echo, be designated as

wherein IFFT represents inverse fast Fourier transform;

11e) to the target echo signal s after velocity compensation ₆(f, τ _m) do the distance-slow time-domain signal of the target echo that distance dimension IFFT obtains

be expressed as:

$s_7(\hat{t},{\mathrm{\&tau;}}_m)=A^{\&prime;\&prime;}\sin c\left(\mathrm{\&Delta;f}\right[\hat{t}-(\frac{f_d}{\mathrm{\&gamma;}}+\frac{{2R}_0}{c})\left]\right)\exp (-j2{\mathrm{\&pi;<figure-callout\; id="0"\; label="f\; d"\; filenames="HDA0000155747140000021.png,HDA0000155747140000031.png"\; state="\{\{state\}\}">f</figure-callout>}}_d{\mathrm{\&tau;}}_m)\exp \left({-\mathrm{j\&pi;\&gamma;}}_a{\mathrm{\&tau;}}_m^2\right),$

In formula, A " represents target echo to do the echo strength of target after IFFT.

Step 12, with Dechirp method estimating target motion acceleration.

The specific implementation of this step: the Dechirp method described in list of references < < wide and narrow strip Radar Targets'Detection and imaging technique research > >, to the pulse echo data after velocity compensation

carry out the estimation of doppler frequency rate, the estimated value of doppler frequency rate is designated as

Step 13, utilizes the estimated value of doppler frequency rate

calculate the estimated value of target travel acceleration

$\hat{a}=\frac{\mathrm{\&lambda;}\widehat{{\mathrm{\&gamma;}}_a}}{2},$

In formula,

the estimated value that represents doppler frequency rate, λ represents radar wavelength.

Step 14, compensates the slow time dimension quadratic term of the echo data after velocity compensation.

14a) the distance of the echo signal echo in the pulse echo from velocity compensation-slow time domain expression can find out, also there is quadratic phase item in pulse echo on slow time dimension, for the energy of echo can effectively be accumulated, and the data with following formula after to velocity compensation

carry out the compensation of slow time dimension quadratic term:

$Q({\mathrm{\&tau;}}_m;{\widehat{\mathrm{\&gamma;}}}_a)=\exp \left(\mathrm{j\&pi;}{\widehat{\mathrm{\&gamma;}}}_a{\mathrm{\&tau;}}_m^2\right),$

In formula,

the penalty function that represents slow time dimension quadratic phase item, τ _mthe slow time after variable is replaced,

the estimated value that represents doppler frequency rate;

14b) the pulse echo data after slow time dimension quadratic term compensation are designated as

14c) use function

to target echo signal

carry out the target echo signal obtaining after slow time dimension quadratic term compensation

expression be:

$s_8(\hat{t},{\mathrm{\&tau;}}_m)=A^{\&prime;\&prime;}\sin c\left(\mathrm{\&Delta;f}\right[\hat{t}-(\frac{f_d}{\mathrm{\&gamma;}}+\frac{{2R}_0}{c})\left]\right)\exp (-j2\mathrm{\&pi;}{\mathrm{<figure-callout\; id="0"\; label="f\; d"\; filenames="HDA0000155747140000021.png,HDA0000155747140000031.png"\; state="\{\{state\}\}">f</figure-callout>}}_d{\mathrm{\&pi;}}_m)$

，

In formula, A " echo strength that represents target,

the fast time, τ _mbe the slow time after conversion, Δ f is radar emission signal bandwidth, for the Doppler frequency of target, λ is radar wavelength, f _d0represent the Doppler frequency after fuzzy, R ₀for target initial time distance, v is target velocity, τ _mbe the slow time after variable is replaced, c is the light velocity.

Step 15, the echo data after the compensation of slow time dimension quadratic term is done coherent accumulation.

15a) at paired pulses echo data

carry out successively after General keystone transform, velocity ambiguity compensation, velocity compensation and slow time dimension quadratic term compensation, to the pulse echo data that obtain

do the accumulation of multiple-pulse coherent, the peak value of coherent accumulation result is the target detecting;

15b) to target echo signal

carry out coherent accumulation, the echo signal that obtains detecting

expression be:

$s_9(\hat{t},f_{\mathrm{\&tau;}})=A^{\&prime;\&prime;}\sin c\left(\mathrm{\&Delta;f}\right[\hat{t}-(\frac{f_d}{\mathrm{\&gamma;}}+\frac{{2R}_0}{c})\left]\right)\sin c\left(T(f_{\mathrm{\&tau;}}-f_{d0})\right),$

In formula, A " echo strength that represents target,

the fast time, f _τwith τ _mfor the frequency after variable FFT conversion, FFT represents Fast Fourier Transform (FFT), and γ represents frequency modulation rate, and Δ f is radar emission signal bandwidth, for the Doppler frequency of target, λ is radar wavelength, and T is coherent integration time, f _d0represent the Doppler frequency after fuzzy, R ₀for target initial time distance, c is the light velocity.

Effect of the present invention further illustrates by following simulation comparison test:

1. experiment scene:

Experimental data comprises with the white Gaussian noise of science software for calculation matlab R2008a emulation generation and the radar return of high speed high acceleration moving target, target number is 1, the initial distance of the relative radar of target is 900 meters, and target radial speed is 3400m/s, and target radial acceleration is 200m/s ², radar emission carrier frequency is 1GHz, and range resolution is 2 meters, and radar pulse repetition frequency is 500Hz, and a coherent comprises 100 pulses in integration time, and signal to noise ratio (S/N ratio) is-10dB.

2. emulation content:

Utilize experimental data, carry out computer simulation experiment, provide respectively the testing result of conventional sense method, optimum detection methodology and detection method of the present invention as shown in Figure 2, Figure 3, Figure 4, wherein Fig. 2 is for adopting the testing result of conventional sense method, Fig. 3 is for adopting the testing result of optimum detection methodology, and Fig. 4 is for adopting the testing result of the method for the invention.

Conventional sense method refers to that paired pulses echo data is directly slow time dimension FFT, and optimum detection methodology refers to paired pulses echo range curvature and the range walk of pulse echo of having carried out the compensation of General keystone transform, velocity ambiguity and velocity compensation post-equalization, but the slow time dimension FFT of the pulse echo result of not carrying out slow time dimension quadratic term compensation.

3. analysis of simulation result:

As can be seen from Figure 2, while utilizing conventional sense method to detect, because target is high speed high acceleration moving, can cause range walk and the bending of pulse echo, therefore paired pulses echo data cannot accumulate out energy peak after directly adopting slow time dimension FFT, thereby target cannot be detected.

As can be seen from Figure 3, while utilizing optimum detection methodology to detect, although proofreaied and correct range curvature and the range walk of pulse echo, but do not carry out the compensation of slow time dimension quadratic phase item, thereby make target energy on slow time dimension, still disperse by energy accumulation, still target cannot to be detected.

As can be seen from Figure 4, utilize after the method for the invention, be that paired pulses echo data is after General keystone transform, velocity ambiguity compensation, velocity compensation, slow time dimension quadratic term compensation, the peak value of echo envelope has all been corrected to same range unit, thereby to being corrected to when the echoed signal of same range unit is carried out coherent accumulation, target echo energy efficient can be accumulated, thereby target detected.

In sum, when the high acceleration of high speed target is detected, the inventive method is obviously better than conventional sense method and optimum detection methodology.

Claims (8)

1. the radar target detection method based on General keystone transform and the accumulation of non-coherent, comprises the steps:

1) approximate model and the noise model of radar high speed high acceleration moving target are expressed as and d (l, t _m), wherein, l is range unit label,

represent the fast time, t _mrepresent slow time variable;

2), according to target approximate model and noise model, obtain radar pulse echo:

and paired pulses echo

with the fast time

for variable, be FFT and transform to frequency of distance territory, obtain echo apart from frequency domain data z (f, the t of dimension _m), wherein f represents frequency of distance;

3) by frequency domain data z (f, the t of distance dimension _m) along distance dimension, do frequency domain pulse compression, the data after pulse compression are designated as to z'(f, t _m), and to the data z'(f after this pulse compression, t _m) in slow time variable t _mwith new time variable τ _mreplace, variable alternate form is the data that variable obtains after replacing are designated as z'(f, τ _m), f wherein _cfor radar emission signal carrier frequency;

4) setting fuzzy factor scope is [n _min, n _max], n wherein _minminimum, n _maxmaximum, from minimum to maximum, N is individual altogether, and is all positive integer, the data z'(f after with the fuzzy factor within the scope of this, variable being replaced, τ _m) carrying out fuzzy compensation, the echo data after compensation is designated as

5) to the echo frequency domain data after fuzzy compensation pass the time domain interpolation that in each range unit of row, p is ordered, p=1,2,3, the data after time domain interpolation are designated as

6) utilize the data after time domain interpolation, when calculating target velocity and being v, echo envelope range walk is counted the umber of pulse w in k and same range unit _k;

7) according to target friction speed ripple range walk number k and same range unit umber of pulse w next time _kdifference, the data after time domain interpolation are carried out to ring shift, and the N group echo data after ring shift are carried out to non-coherent accumulation, obtain N and organize non-coherent accumulation result, be designated as

wherein || expression be modulo operation, M indicating impulse number;

8) N is organized to non-coherent accumulation result

compare, get and wherein accumulate the velocity ambiguity factor of n that the fuzzy factor corresponding to result of peak value maximum is target _kestimated value, be designated as

and speed corresponding to the accumulation peak value of getting this maximum is fuzzy rear target initial velocity v ₀estimated value, be designated as

9) initial velocity estimated value after the objective fuzzy that utilization obtains

with fuzzy factor estimated value

calculate the estimated value of target initial velocity then by the target initial velocity estimated value of estimating to obtain

to with fuzzy factor estimated value

corresponding echo frequency numeric field data

carry out velocity compensation, then the echo data after compensation is transformed to apart from time domain, the distance time domain data after conversion is designated as

10) to converting later distance time domain data

by the method for Dechirp, carry out the search of frequency modulation rate, obtain frequency modulation rate γ _aestimated value, be designated as

and then obtained the estimated value of target travel acceleration

11) use the estimated value of the aimed acceleration obtaining

to the data after velocity compensation carry out the compensation of slow time dimension quadratic phase item, and the data after compensation are done to the coherent accumulation of slow time dimension, the peak value of coherent accumulation is the target of detection.

2. method according to claim 1, wherein described in step 3) by frequency domain data z (f, the t of distance dimension _m) along distance dimension, do frequency domain pulse compression, by following formula, undertaken:

In formula,

indicating impulse compression function,

the fast time, T _pbe pulsewidth, γ is frequency modulation rate.

3. method according to claim 1, wherein step 4) data after described one group of fuzzy factor of use is replaced variable carry out fuzzy compensation, utilize following formula to carry out:

In formula, G (f, τ _m; n _k) expression fuzzy compensation function, n _krepresent fuzzy factor, τ _mbe the slow time after variable is replaced, f represents frequency of distance, f _cfor radar emission signal carrier frequency, PRF is the repetition frequency of radar emission signal.

4. method according to claim 1, the umber of pulse in echo envelope range walk number and same range unit during calculating target velocity and be v described in step 6) wherein, carry out as follows:

6a) calculate the fuzzyyest speed of target

according to the fuzzyyest speed v obtaining _max, calculating target echo envelope range walk number is l=(p+1) * round (2v _maxf _s/ cPRF) the umber of pulse w and in same range unit _l=M/l, wherein PRF represents radar transmitted pulse repetition frequency, and λ represents radar wavelength, and what round represented is the operation that rounds rounding up, f _sfor signal sampling frequency, p=1,2,3, c represents the light velocity, M indicating impulse number;

6b) according to target echo envelope range walk, count the umber of pulse w in l and same range unit _l, determine speed search step delta v:

First, get two value l of arbitrary neighborhood within the scope of range walk number 0～l-1 ₁and l ₂, l ₁< l ₂;

Secondly, calculate target velocity corresponding when range walk number occurring being l1 be l with there is range walk number ₂time corresponding target velocity

Δ f represents radar emission signal bandwidth;

Then, with there is range walk number, be l ₂time corresponding target velocity v ₂be l with there is range walk number ₁time corresponding target velocity v1 difference as speed search step delta v;

6c) take Δ v as step-length, obtaining target velocity scope is 0≤v≤v _maxall speed v within the scope of this, and the range walk of computing velocity target echo envelope while being v counts k=(p+1) * round (2v/ (cPRF) Δ fM), c represents the light velocity, Δ f represents radar emission signal bandwidth;

While 6d) being v according to speed, the range walk of target echo envelope is counted k, obtains the umber of pulse w in same range unit _k=M/k, M indicating impulse number.

5. method according to claim 1, wherein described in step 7) according to target at friction speed ripple range walk number and same range unit umber of pulse w next time _kdifference, the data after time domain interpolation are carried out to ring shift, be to the N group data after time domain interpolation

all proceed as follows:

First, take target velocity umber of pulse w in same range unit during as v _kfor segment length, M pulse is divided into k group;

Then, the k group data after dividing are carried out to the ring shift of 0～k-1 position successively along distance dimension.

6. method according to claim 1, calculates the estimated value of target initial velocity in wherein said step 9) and described step 10)

with the estimated value of calculating target travel acceleration

with following formula, calculate:

In formula,

the estimated value of first Doppler frequency after expression target generation velocity ambiguity, λ represents radar wavelength,

the estimated value that represents fuzzy factor, the estimated value that represents doppler frequency rate, PRF is the repetition frequency of radar emission signal.

7. method according to claim 1, wherein described in step 9) with estimating that the target initial velocity estimated value pair echo frequency numeric field data corresponding with fuzzy factor estimated value obtaining carry out velocity compensation, utilize following formula to carry out:

In formula, represent velocity compensation function, represent the just estimated value of Doppler frequency of fuzzy rear target, f represents frequency of distance, τ _mit is the slow time after variable is replaced.

8. according to the method described in right 1, the data of the estimated value of the aimed acceleration that wherein use described in step 11) obtains after to velocity compensation are carried out the compensation of slow time dimension quadratic phase item, utilize following formula to carry out:

In formula,

the penalty function that represents slow time dimension quadratic phase item, τ _mthe slow time after variable is replaced,

the estimated value that represents doppler frequency rate.

Images