CN104535973A - Target detection method by use of airborne early warning radar - Google Patents

Abstract

The invention belongs to the technical field of radar clutter inhibition and in particular relates to a target detection method by use of airborne early warning radar. The target detection method comprises the following specific steps: receiving space-time two-dimensional echo data by use of airborne early warning radar; determining normalized space-time two-dimensional frequency points corresponding to the positions of a clutter ridge curve related to the space-time two-dimensional echo data; constructing a clutter ridge steering vector basis matrix Psi according to the normalized space-time two-dimensional frequency points corresponding to the positions of the clutter ridge curve; constructing a matrix element error vector Es and a new clutter ridge steering vector basis matrix Psi', and performing iterative fitting on the data of the first unit to be tested in distance under the least square constraint by use of the new clutter ridge steering vector basis matrix Psi', thereby obtaining the fitted residual data of the first unit to be tested in distance; performing unit average constant false alarm rate detection on the fitted residual data of the first unit to be tested in distance, thereby obtaining a target detection result.

Description

A kind of airborne early warn ing radar object detection method

Technical field

The invention belongs to radar clutter suppression technology field, particularly a kind of airborne early warn ing radar object detection method, can be used for airborne early warn ing radar signal transacting.

Background technology

The main task of airborne early warn ing radar is the detection of a target position tracking to it in complex clutter background, and carries out effectively suppressing being the core means improving early warning radar serviceability to clutter.Space-time adaptive process (space-time adaptive processing, STAP) technology makes full use of spatial domain and time-domain information, while carrying out coherent accumulation to echo signal, by space-time adaptive process filtering ground clutter, realize the effective detection of airborne radar to target.E2-D airborne early warn ing radar as the U.S. just adopts this technology.In actual applications, mainly there are following two aspect problems in STAP technology: first, in clutter environment heterogeneous, obtain the abundant independent same distribution for estimate covariance matrix (independentand identically distributed, IID) training sample very difficult; Secondly, even if the demand of training sample is met, the excessive problem of full space time processing calculated amount can cause real-time to be difficult to ensure.For solving the problem, promoting STAP technology more practical, there has been proposed many innovative approachs or method.

The patent of invention " space-time adaptive processing method under non-homogeneous clutter environment " (number of patent application 201010129723.3, publication No. CN 101819269 A) of Tsing-Hua University's application discloses a kind of overcomplete sparse representation method of super-resolution estimation clutter space-time two-dimensional spectrum in non-homogeneous clutter environment.The method achieve when independent same distribution sample number deficiency, utilize single frames training sample to estimate clutter covariance matrix, thus avoid strong non-homogeneous clutter environment on the impact of self-adaptive processing effect.But, the main deficiency that the method still exists is: super complete radix order clutter spectrum being carried out to rarefaction representation is uncertain, but much larger than degree of freedom in system, and degree of freedom in system is usually thousands of in reality, operand required in the covariance matrix restructuring procedure of each range unit sample is like this very large, be unfavorable for real-time process, thus have influence on its practical engineering application effect.

Patent of invention " a kind of dimensionality reduction space-time adaptive processing method based on Covariance Matrix Weighting " (number of patent application 201210251589.3, publication No. CN 102778669 A) of Beijing Institute of Technology's application discloses a kind of method utilizing Covariance Matrix Weighting technology to estimate clutter covariance matrix.The method achieve broadening clutter recess adaptively and, to adapt to the clutter ridge in actual environment, thus make to exist clutter when revealing, also by STAP method clutter reduction effectively, improve STAP robustness in actual applications.But, the deficiency that the method still exists is: the wave filter notch width designed by clutter reduction is artificial setting, can not clutter ridge situation in perception real data adaptively, when broadening amount is excessive, Minimum detectable can be caused to become large, very unfavorable to the detection of low speed Small object.

Summary of the invention

The object of the invention is to propose a kind of airborne early warn ing radar object detection method, the clutter recognition performance of Adaptive Signal Processing can be improved, improve the detection probability of moving target.

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

A kind of airborne early warn ing radar object detection method comprises the following steps:

Step 1, utilizes airborne early warn ing radar antenna to receive space-time two-dimensional echo data;

Step 2, determines the normalization space-time two-dimensional frequency corresponding to clutter ridge curve location that space-time two-dimensional echo data is correlated with;

Step 3, the normalization space-time two-dimensional frequency structure clutter ridge steering vector basis matrix ψ corresponding to clutter ridge curve location;

Step 4, structure array element error vector E _s, wherein, ⊙ represents that Hadamard amasss; 1 _nrepresent the column vector that N is capable, 1 _nin element be 1 entirely; ρ represents array element range error vector, represent array element phase error vector; Construct new clutter ridge steering vector basis matrix ψ ':

${\mathrm{\ψ}}^{\′}=\mathrm{diag}(E_s\⊗1_M)\mathrm{\ψ}$

Wherein, represent Kronecker product, 1 _mrepresent the column vector that M is capable, 1 _min element be 1 entirely, represent with vector element be the diagonal matrix that the elements in a main diagonal is formed;

Step 5, new clutter ridge steering vector basis matrix ψ ' is utilized under least square constraint, to carry out iterative fitting to the data of l range unit to be detected, obtain the remaining data after the matching of l range unit to be detected, l=1,2, ... L, L are the range unit number to be detected of airborne early warn ing radar;

Step 6, carries out CA-CFAR detection to the remaining data after the matching of l range unit to be detected, draws object detection results.

Feature of the present invention and further improvement are:

Described step 1 specifically comprises following sub-step:

1a) set under airborne early warn ing radar is operated in positive side-looking mode;

1b) airborne early warn ing radar is utilized outwards to transmit; The receiving array of airborne early warn ing radar is the even linear array of N number of array element composition, within the coherent accumulation time of M pulse, utilizes the receiving array of airborne early warn ing radar to receive the space-time two-dimensional echo data of L range unit of ground return.

Described step 2 specifically comprises following sub-step:

2a) on interval [0,1], carry out uniform discrete, obtain multiple discrete point, the value of each discrete point is in interval [0,1], and discrete counting is expressed as num, arranges normalization spatial frequency normalization spatial frequency for the value of each discrete point after discretize;

2b) basis determine corresponding normalization Doppler frequency β=2vT _r/ d, wherein, v is the speed of carrier aircraft, T _rbe the pulse repetition time that airborne early warn ing radar transmits, d is the antenna spacing of the receiving array of airborne early warn ing radar;

2c) by normalization Doppler frequency with normalization spatial frequency be defined as the normalization space-time two-dimensional frequency corresponding to clutter ridge curve location.

Described step 3 specifically comprises following sub-step:

Normalization space-time two-dimensional frequency 3a) corresponding to clutter ridge curve location, obtains corresponding time domain steering vector and spatial domain steering vector

$\left\{\begin{array}{c}{s}_t\left({\stackrel{\‾}{f}}_d\right)={\left[\begin{array}{cccc}1& e^{j2\mathrm{\π}{\stackrel{\‾}{f}}_d}& ...& e^{j2\mathrm{\π}(M-1){\stackrel{\‾}{f}}_d}\end{array}\right]}^T\\ s_s\left({\stackrel{\‾}{f}}_s\right)={\left[\begin{array}{cccc}1& e^{j2\mathrm{\π}{\stackrel{\‾}{f}}_s}& ...& e^{j2\mathrm{\π}(N-1){\stackrel{\‾}{f}}_s}\end{array}\right]}^T\end{array}\right.$

Wherein, representing time domain steering vector, is the column vector that M is capable; representing spatial domain steering vector, is the column vector that N is capable; for normalization Doppler frequency, for normalization spatial frequency, () ^trepresent matrix transpose operation; N represents the array number of airborne early warn ing radar receiving array, and M represents the accumulation umber of pulse of airborne early warn ing radar Received signal strength;

3b) according to the coupled relation of space-time two-dimensional echo data when sky on bidimensional, ask for the space-time two-dimensional steering vector of corresponding empty time frequency point $s({\stackrel{\‾}{f}}_d,{\stackrel{\‾}{f}}_s)=s_t\left({\stackrel{\‾}{f}}_d\right)\⊗s_s\left({\stackrel{\‾}{f}}_s\right),$ Wherein, represent Kronecker product;

3c) by num discrete space-time two-dimensional steering vector of counting corresponding combine, form clutter ridge steering vector basis matrix ψ, the matrix of ψ to be size be MN × num.

Described step 5 specifically comprises following sub-step:

5a) set up the following 2 norm optimization models about fitting coefficient vector:

$\underset{{\mathrm{\α}}_l}{\min }{\left|\right|x_l-\mathrm{diag}(E_s\⊗1_M)\mathrm{\ψ}{\mathrm{\α}}_l\left|\right|}_2$

Wherein, || || ₂represent 2 norms asking vector or matrix, x _lfor the data of l in space-time two-dimensional echo data range unit to be detected, x _lit is the column vector that MN is capable; α _lfor l to be solved range unit to be detected to the fitting coefficient vector under basis matrix, α _lit is the column vector that num is capable; 1 _mrepresent the column vector that M is capable, 1 _min element be 1 entirely, represent Kronecker product operation;

Make array element error vector E _s=1 _n, 1 _nrepresent the column vector that N is capable, 1 _nin element be 1 entirely;

5b) by E _s=1 _nsubstitute into above-mentioned about in 2 norm optimization models of fitting coefficient vector, draw l range unit to be detected to the fitting coefficient vector α under basis matrix _l;

5c) according to sub-step 5b) the fitting coefficient vector α that draws _l, set up the following Optimized model about array element error vector:

$\underset{E_s}{\min }{\left|\right|x_l-\mathrm{diag}(E_s\⊗1_M)\mathrm{\ψ}{\mathrm{\α}}_l\left|\right|}_2$

Solve the above-mentioned Optimized model about array element error vector, draw and obtain the array element error vector E after upgrading _s;

5d) make iteration difference ${\mathrm{\δ}}_l=z_l-{\left|\right|x_l-\mathrm{diag}(E_s\⊗1_M)\mathrm{\ψ}{a}_l\left|\right|}_2,$ Make data norm z _lfor:

${\left|\right|x_l-\mathrm{diag}(E_s\⊗1_M)\mathrm{\ψ}{a}_l\left|\right|}_2;$

If 5e) z _lbe greater than σ _land δ _lbe greater than 0.01 σ _l, then sub-step 5b is back to), σ _lfor the noise level of setting; Otherwise, skip to sub-step 5f);

5f) by current E _svalue substitute in the computing formula of new clutter ridge steering vector basis matrix ψ ', draw the clutter ridge steering vector basis matrix ψ ' after matching, order matrix ψ ' ^*=ψ '; Obtain current E _sthe fitting coefficient vector α corresponding to value _l, by current E _sthe fitting coefficient vector α corresponding to value _lbe designated as the basis matrix optimal fitting coefficient of l range unit to be detected draw the remaining data x after the matching of l range unit to be detected _l-ψ ' ^*α _l ^*.

Beneficial effect of the present invention is:

1) the present invention is the airborne radar space-time adaptive processing method based on clutter ridge steering vector basis matrix, according to radar array flow pattern and systematic parameter, steering vector when to select corresponding to the empty time frequency point in clutter ridge curve location empty, construction data is for representing clutter ridge steering vector basis matrix, iterative fitting process is carried out to radar return data, thus the abundant matching realized strong ground clutter and suppression, substantially increase the detectability of mobile target in complex background.

2) the clutter ridge steering vector basis matrix used in the present invention be according to the mode of operation of radar and systematic parameter determined, when radar is operated under the battle array pattern of positive side, clutter in echo data can be represented by the steering vector ascertained the number, and this number is much smaller than degree of freedom in system.With this group steering vector structure basis matrix, the problem dimension that the problem dimension of optimized algorithm in the present invention is required in existing sparse restoration methods can be made, greatly reduce algorithm complex, be conducive to the real-time process to data.

3) the present invention is by using the mode of array element error and fitting coefficient being carried out to iterative least square estimation, can carry out good matching, thus fully suppress it, improve moving object detection performance to non-homogeneous clutter.Meanwhile, after considering array element error, there is good robustness to array element error.

Accompanying drawing explanation

Fig. 1 is the process flow diagram of a kind of airborne early warn ing radar object detection method of the present invention;

Fig. 2 is pulse Doppler (PD) the result schematic diagram of original space-time two-dimensional echo data in emulation experiment;

Fig. 3 is pulse Doppler (PD) the result schematic diagram of the fitting data drawn by the present invention in emulation experiment;

Fig. 4 a is the comparison diagram of array element range error and the true array element range error estimated through iteration optimization by the present invention in emulation experiment;

Fig. 4 b is the comparison diagram of array element phase error and the true array element phase error estimated through iteration optimization by the present invention in emulation experiment;

Fig. 5 is respectively by the target detection probability curve map that the present invention, GIP-EFA algorithm and GIP-JDL algorithm draw in emulation experiment.

Embodiment

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

With reference to Fig. 1, be the process flow diagram of a kind of airborne early warn ing radar object detection method of the present invention, the present invention is used for airborne early warn ing radar signal transacting, and its concrete implementation step is as follows:

Step 1, utilizes airborne early warn ing radar antenna to receive space-time two-dimensional echo data.

Step 1 specifically comprises following sub-step:

1a) set under airborne early warn ing radar is operated in positive side-looking mode.

1b) airborne early warn ing radar is utilized outwards to transmit; The receiving array of airborne early warn ing radar is the even linear array of N number of array element composition, within the coherent accumulation time of M pulse, utilizes the receiving array of airborne early warn ing radar to receive the space-time two-dimensional echo data of L range unit of ground return.

Step 2, determines the normalization space-time two-dimensional frequency corresponding to clutter ridge curve location that space-time two-dimensional echo data is correlated with.

Step 2 specifically comprises following sub-step:

2a) on interval [0,1], carry out uniform discrete, obtain multiple discrete point, the value of each discrete point is in interval [0,1], and discrete counting is expressed as num, arranges normalization spatial frequency normalization spatial frequency for the value of each discrete point after discretize; The discrete num of counting calculates according to following formula:

num＝N+β·(M-1)

Wherein, N represent airborne early warn ing radar receiving array array number, M represents the accumulation umber of pulse of airborne early warn ing radar Received signal strength, and β represents clutter ridge slope of a curve, and clutter ridge slope of a curve β is calculated as follows formula:

β＝2vT _r/d

Wherein, v is the speed of carrier aircraft, T _rbe the pulse repetition time that airborne early warn ing radar transmits, d is the antenna spacing of the receiving array of airborne early warn ing radar.

2b) normalization Doppler frequency with normalization spatial frequency the linear relationship of following formula is there is in positive side-looking even linear array radar:

${\stackrel{\‾}{f}}_d=\mathrm{\β}{\stackrel{\‾}{f}}_s$

Corresponding normalization Doppler frequency is determined according to above formula linear relationship

2c) by normalization Doppler frequency with normalization spatial frequency be defined as the normalization space-time two-dimensional frequency corresponding to clutter ridge curve location, because discrete counting is expressed as num, then the number of normalization space-time two-dimensional frequency is num.

Step 3, the normalization space-time two-dimensional frequency structure clutter ridge steering vector basis matrix ψ corresponding to clutter ridge curve location.

Step 3 specifically comprises following sub-step:

Normalization space-time two-dimensional frequency 3a) corresponding to clutter ridge curve location, obtains corresponding time domain steering vector and spatial domain steering vector as shown in the formula:

$\left\{\begin{array}{c}{s}_t\left({\stackrel{\‾}{f}}_d\right)={\left[\begin{array}{cccc}1& e^{j2\mathrm{\π}{\stackrel{\‾}{f}}_d}& ...& e^{j2\mathrm{\π}(M-1){\stackrel{\‾}{f}}_d}\end{array}\right]}^T\\ s_s\left({\stackrel{\‾}{f}}_s\right)={\left[\begin{array}{cccc}1& e^{j2\mathrm{\π}{\stackrel{\‾}{f}}_s}& ...& e^{j2\mathrm{\π}(N-1){\stackrel{\‾}{f}}_s}\end{array}\right]}^T\end{array}\right.$

Wherein, representing time domain steering vector, is the column vector that M is capable; representing spatial domain steering vector, is the column vector that N is capable; for normalization Doppler frequency, for normalization spatial frequency, () ^trepresent matrix transpose operation; N represents the array number of airborne early warn ing radar receiving array, and M represents the accumulation umber of pulse of airborne early warn ing radar Received signal strength.

3b) according to the coupled relation of space-time two-dimensional echo data when sky on bidimensional, ask for the space-time two-dimensional steering vector of corresponding empty time frequency point

$s({\stackrel{\‾}{f}}_d,{\stackrel{\‾}{f}}_s)=s_t\left({\stackrel{\‾}{f}}_d\right)\⊗s_s\left({\stackrel{\‾}{f}}_s\right)$

Wherein, represent Kronecker product operation; Can find out, space-time two-dimensional steering vector it is the column vector that MN is capable.

3c) by num discrete space-time two-dimensional steering vector of counting corresponding combine, form clutter ridge steering vector basis matrix ψ.

The space-time two-dimensional steering vector of clutter ridge steering vector basis matrix ψ corresponding to num the empty time frequency point at clutter ridge curve location place is formed, space-time two-dimensional steering vector size be MN × 1, the matrix of known ψ to be size be MN × num.This clutter ridge steering vector basis matrix can be determined by systematic parameter completely, without the need to training sample, and does not change with the change of range unit to be detected, thus is convenient to practical application.

Step 4, considers array element error condition, constructs new basis matrix ψ '.

Step 4 specifically comprises following sub-step:

4a) according to the impact of error pair array antenna, obtain array element error vector E _s:

Wherein, ⊙ represents the long-pending operation of Hadamard; 1 _nrepresent the column vector that N is capable, 1 _nin element be 1 entirely; ρ represents array element range error vector, and ρ is the column vector that N is capable, represent array element phase error vector, it is the column vector that N is capable; ρ=[ρ ₁ρ ₂ρ _n] ^t, ρ _nrepresent range error during the n-th array element Received signal strength in the receiving array of airborne early warn ing radar; represent phase error during the n-th array element Received signal strength in the receiving array of airborne early warn ing radar, n=1,2 ... N.

4b) according to array element error model, construct new clutter ridge steering vector basis matrix ψ ':

${\mathrm{\ψ}}^{\′}=\mathrm{diag}(E_s\⊗1_M)\mathrm{\ψ}$

Wherein, represent Kronecker product operation, 1 _mrepresent the column vector that M is capable, 1 _min element be 1 entirely, represent with vector element be the elements in a main diagonal form diagonal matrix (vector element by its at vector order be arranged on the principal diagonal of diagonal matrix).

Step 5, new clutter ridge steering vector basis matrix ψ ' is utilized under least square constraint, to carry out iterative fitting to the data of l range unit to be detected, obtain the remaining data after the matching of l range unit to be detected, l=1,2, ... L, L are the range unit number to be detected of airborne early warn ing radar.

In steps of 5, the data utilizing new clutter ridge steering vector basis matrix ψ ' to treat detecting distance unit carry out iterative fitting under least square constraint, carry out Combined estimator to array element error and fitting coefficient, the data of this range unit to be detected take from the space-time two-dimensional echo data received in step 1 successively; Fitting formula as shown in the formula,

$\begin{array}{c}{y}_l=\underset{{\mathrm{\α}}_l}{\min }{\left|\right|x_l-{\mathrm{\ψ}}^{\′}{\mathrm{\α}}_l\left|\right|}_2\\ =\underset{{\mathrm{\α}}_l}{\min }{\left|\right|x_l-\mathrm{diag}(E_s\⊗1_M)\mathrm{\ψ}{\mathrm{\α}}_l\left|\right|}_2\end{array}$

Wherein, || || ₂represent 2 norms asking vector or matrix, x _lfor the data of l in space-time two-dimensional echo data range unit to be detected, x _lit is the column vector that MN is capable; α _lfor l to be solved range unit to be detected to the fitting coefficient vector under basis matrix, α _lit is the column vector that num is capable; y _lbe the error of fitting norm of l range unit to be detected, l represents range unit sequence number to be detected, l=1,2 ... L, L are the range unit number to be detected of airborne early warn ing radar.

Step 5 specifically comprises following sub-step:

5a) set up the following 2 norm optimization models about fitting coefficient vector:

$\underset{{\mathrm{\α}}_l}{\min }{\left|\right|x_l-\mathrm{diag}(E_s\⊗1_M)\mathrm{\ψ}{\mathrm{\α}}_l\left|\right|}_2$

Wherein, || || ₂represent 2 norms asking vector or matrix, x _lfor the data of l in space-time two-dimensional echo data range unit to be detected, x _lit is the column vector that MN is capable; α _lfor l to be solved range unit to be detected to the fitting coefficient vector under basis matrix, α _lit is the column vector that num is capable; 1 _mrepresent the column vector that M is capable, 1 _min element be 1 entirely, represent Kronecker product operation.

Carry out parameter initialization, make array element error vector E _s=1 _n, 1 _nrepresent the column vector that N is capable, 1 _nin element be 1 entirely; Make data norm z _l=|| x _l|| ₂, make iteration value of delta _l=σ _l, σ _lfor the noise level of setting.

5b) by E _s=1 _nsubstitute into above-mentioned about in 2 norm optimization models of fitting coefficient vector, draw l range unit to be detected to the fitting coefficient vector α under basis matrix _l.

5c) according to sub-step 5b) the fitting coefficient vector α that draws _l, set up the following Optimized model about array element error vector:

$\underset{E_s}{\min }{\left|\right|x_l-\mathrm{diag}(E_s\⊗1_M)\mathrm{\ψ}{\mathrm{\α}}_l\left|\right|}_2$

Solve the above-mentioned Optimized model about array element error vector, draw and obtain the array element error vector E after upgrading _s.

5d) make iteration difference ${\mathrm{\δ}}_l=z_l-{\left|\right|x_l-\mathrm{diag}(E_s\⊗1_M)\mathrm{\ψ}{a}_l\left|\right|}_2,$ Make data norm z _lfor:

${\left|\right|x_l-\mathrm{diag}(E_s\⊗1_M)\mathrm{\ψ}{a}_l\left|\right|}_2.$

If 5e) z _lbe greater than σ _land δ _lbe greater than 0.01 σ _l, then sub-step 5b is back to); Otherwise, skip to sub-step 5f).

5f) by current E _svalue substitute in the computing formula of new clutter ridge steering vector basis matrix ψ ', draw the clutter ridge steering vector basis matrix ψ ' after matching, order matrix ψ ' ^*=ψ '; Obtain current E _sthe fitting coefficient vector α corresponding to value _l, by current E _sthe fitting coefficient vector α corresponding to value _lbe designated as the basis matrix optimal fitting coefficient of l range unit to be detected

According to matrix ψ ' ^*with the basis matrix optimal fitting coefficient of l range unit to be detected ask for the remaining data after the matching of l range unit to be detected, the remaining data after the matching of l range unit to be detected is x _l-ψ ' ^*α _l ^*; x _lbe the data of l range unit to be detected.

In steps of 5, the size of new basis matrix ψ ' is MN × num, and having MN>num, incomplete in the complex data vector space that therefore ψ ' ties up in MN × 1, all column vectors of namely new basis matrix ψ ' can not all sizes of perfect representation be the complex vector of MN × 1.Above-mentioned basis matrix is bipartite, and a part is clutter ridge steering vector basis matrix, and it is made up of steering vector during clutter ridge curve location place empty, and a part is the impact of the array element error introduced in iterative process in addition.Therefore the above-mentioned least-squares iteration matching clutter component can treated in the data of detecting distance unit carries out complete matching expression, and can not fully represent the echo signal not on clutter ridge.When there is echo signal in range unit data to be detected, error norm y _llarger output will be had, namely have larger error of fitting to exist.The main clutter component of this remaining data is by basis matrix ψ ' and factor alpha _labundant matching represents, thus makes clutter component in remaining data, obtain effective filtering.

Step 6, carries out CA-CFAR detection to the remaining data after the matching of l range unit to be detected, draws object detection results.

To the remaining data after the matching of l range unit to be detected carry out CA-CFAR detection (Cell-averaging constant false alarm rate, CA-CFAR), by the remaining data after matching compare with the mean value of the remaining data after the matching of the data of the reference distance unit of the surrounding of l range unit to be detected, according to the remaining data after the matching of l range unit to be detected judge whether to there is target with toaverage ratio size, finally will there is target or there is not target output.

Below in conjunction with emulation experiment, effect of the present invention is described further.

1) simulated conditions:

Emulation experiment of the present invention is carried out under MATLAB 7.11 software.In emulation experiment of the present invention, set airborne early warn ing radar is operated in positive side-looking mode, and the receiving array of airborne early warn ing radar is the evenly distributed linear array of 10 array element, and array element distance is half wavelength.In emulation experiment of the present invention, the echo data of use is that the clutter model simulation proposed according to Lincoln laboratory J.Ward produces, and detailed systematic parameter is with reference to following table.

2) simulation result compares

With reference to Fig. 2, it is pulse Doppler (PD) the result schematic diagram of space-time two-dimensional echo data original in emulation experiment.In Fig. 2, horizontal ordinate represents Doppler's passage, and ordinate represents range unit (range gate), and different gray scales represents the different amplitudes of echo data.As can be seen from Figure 2, in the space-time two-dimensional echo data emulated, clutter energy distribution is on a range-doppler plot not only relevant with antenna radiation pattern, and is subject to the modulation of distance change, can reflect real data situation more truly.

With reference to Fig. 3, it is pulse Doppler (PD) the result schematic diagram of fitting data drawn by the present invention in emulation experiment.In Fig. 3, horizontal ordinate represents Doppler's passage, and ordinate represents range unit (range gate), and different gray scales represents the different amplitudes of echo data.Herein, the fitting data that the present invention draws refers to the data ψ ' that step 5 draws ^*α _l ^*(fitting data of l range unit to be detected).As can be seen from Figure 3, the space-time two-dimensional echo data in fitting data and Fig. 2 is closely similar, substantially can reflect the clutter component in space-time two-dimensional echo data completely.Namely the present invention can carry out fine matching to echo data non-homogeneous in simulating scenes, thus fully can suppress it.

With reference to Fig. 4 a, it is the comparison diagram of the array element range error that estimated through iteration optimization by the present invention in emulation experiment and true array element range error.In Fig. 4 a, transverse axis represents array element sequence number, and the longitudinal axis represents array element range error, and asterisk represents true array element range error, and circle represents the array element range error estimated through iteration optimization by the present invention.With reference to Fig. 4 b, for the comparison diagram of the array element phase error that estimated through iteration optimization by the present invention in emulation experiment and true array element phase error, in Fig. 4 b, transverse axis represents array element sequence number, the longitudinal axis represents array element phase error, unit is degree, and asterisk represents true array element phase error, and circle represents the array element phase error estimated through iteration optimization by the present invention.The amplitude of the array element error as can be seen from Figure 4 estimated and phase place almost overlap completely with actual array element error, can verify the robustness of the present invention to array element error.

With reference to Fig. 5, be respectively by the target detection probability curve map that the present invention, GIP-EFA (generalized innerproduct-extended factored approach) algorithm and GIP-JDL (generalized innerproduct-joint domain localized) algorithm draw in emulation experiment.In Fig. 5, transverse axis represents signal to noise ratio (S/N ratio), and unit is dB, and the longitudinal axis represents target detection probability, and algorithm represents the present invention herein.As can be seen from Figure 5, GIP-EFA algorithm is more consistent with the detection perform of GIP-JDL algorithm, and target detection probability of the present invention is higher than other two kinds of algorithms, and as when detection probability is 0.8, the present invention has the performance advantage more than 3dB.

In sum, existing space-time adaptive processing method (STAP) carries out usually in high-dimensional data space, which increase computational complexity and required training sample number, the present invention is in order to obtain better performance, the present invention is mainly used in solving in higher dimensional space, the problem that space time processing calculated amount is huge and the requirement of training sample number is harsh, use angle Doppler plane of the present invention, in clutter ridge curve location place corresponding empty time steering vector represent clutter component in echo data, thus improve the clutter recognition performance of Adaptive Signal Processing, improve the detection probability of moving target.

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

Claims (5)

1. an airborne early warn ing radar object detection method, is characterized in that, comprises the following steps:

Step 1, utilizes airborne early warn ing radar antenna to receive space-time two-dimensional echo data;

Step 2, determines the normalization space-time two-dimensional frequency corresponding to clutter ridge curve location that space-time two-dimensional echo data is correlated with;

Step 3, the normalization space-time two-dimensional frequency structure clutter ridge steering vector basis matrix ψ corresponding to clutter ridge curve location;

Step 4, structure array element error vector E _s, wherein, ⊙ represents that Hadamard amasss; 1 _nrepresent the column vector that N is capable, 1 _nin element be 1 entirely; ρ represents array element range error vector, represent array element phase error vector; Construct new clutter ridge steering vector basis matrix ψ ':

${\mathrm{\ψ}}^{\′}=\mathrm{diag}(E_s\⊗1_M)\mathrm{\ψ}$

Wherein, represent Kronecker product, 1 _mrepresent the column vector that M is capable, 1 _min element be 1 entirely, represent with vector element be the diagonal matrix that the elements in a main diagonal is formed;

Step 5, new clutter ridge steering vector basis matrix ψ ' is utilized under least square constraint, to carry out iterative fitting to the data of l range unit to be detected, obtain the remaining data after the matching of l range unit to be detected, l=1,2, ... L, L are the range unit number to be detected of airborne early warn ing radar;

Step 6, carries out CA-CFAR detection to the remaining data after the matching of l range unit to be detected, draws object detection results.

2. a kind of airborne early warn ing radar object detection method as claimed in claim 1, it is characterized in that, described step 1 specifically comprises following sub-step:

1a) set under airborne early warn ing radar is operated in positive side-looking mode;

1b) airborne early warn ing radar is utilized outwards to transmit; The receiving array of airborne early warn ing radar is the even linear array of N number of array element composition, within the coherent accumulation time of M pulse, utilizes the receiving array of airborne early warn ing radar to receive the space-time two-dimensional echo data of L range unit of ground return.

3. a kind of airborne early warn ing radar object detection method as claimed in claim 1, it is characterized in that, described step 2 specifically comprises following sub-step:

2a) on interval [0,1], carry out uniform discrete, obtain multiple discrete point, the value of each discrete point is in interval [0,1], and discrete counting is expressed as num, arranges normalization spatial frequency normalization spatial frequency for the value of each discrete point after discretize;

2b) basis determine corresponding normalization Doppler frequency β=2vT _r/ d, wherein, v is the speed of carrier aircraft, T _rbe the pulse repetition time that airborne early warn ing radar transmits, d is the antenna spacing of the receiving array of airborne early warn ing radar;

2c) by normalization Doppler frequency with normalization spatial frequency be defined as the normalization space-time two-dimensional frequency corresponding to clutter ridge curve location.

4. a kind of airborne early warn ing radar object detection method as claimed in claim 1, it is characterized in that, described step 3 specifically comprises following sub-step:

Normalization space-time two-dimensional frequency 3a) corresponding to clutter ridge curve location, obtains corresponding time domain steering vector and spatial domain steering vector

$\left\{\begin{array}{c}{s}_t\left({\stackrel{\‾}{f}}_d\right)={\left[\begin{array}{cccc}1& e^{j2\mathrm{\π}{\stackrel{\‾}{f}}_d}& ...& e^{j2\mathrm{\π}(M-1){\stackrel{\‾}{f}}_d}\end{array}\right]}^T\\ s_s\left({\stackrel{\‾}{f}}_s\right)={\left[\begin{array}{cccc}1& e^{j2\mathrm{\π}{\stackrel{\‾}{f}}_d}& ...& e^{j2\mathrm{\π}(N-1){\stackrel{\‾}{f}}_s}\end{array}\right]}^{\text{T}}\end{array}\right.$

Wherein, representing time domain steering vector, is the column vector that M is capable; representing spatial domain steering vector, is the column vector that N is capable; for normalization Doppler frequency, for normalization spatial frequency, () ^trepresent matrix transpose operation; N represents the array number of airborne early warn ing radar receiving array, and M represents the accumulation umber of pulse of airborne early warn ing radar Received signal strength;

3b) according to the coupled relation of space-time two-dimensional echo data when sky on bidimensional, ask for the space-time two-dimensional steering vector of corresponding empty time frequency point wherein, represent Kronecker product;

3c) by num discrete space-time two-dimensional steering vector of counting corresponding combine, form clutter ridge steering vector basis matrix ψ, the matrix of ψ to be size be MN × num.

5. a kind of airborne early warn ing radar object detection method as claimed in claim 1, it is characterized in that, described step 5 specifically comprises following sub-step:

5a) set up the following 2 norm optimization models about fitting coefficient vector:

$\underset{{\mathrm{\α}}_l}{\min }{\left|\right|x_l-\mathrm{diag}(E_s\⊗1_M)\mathrm{\ψ}{\mathrm{\α}}_l\left|\right|}_2$

Wherein, || || ₂represent 2 norms asking vector or matrix, x _lfor the data of l in space-time two-dimensional echo data range unit to be detected, x _lit is the column vector that MN is capable; α _lfor l to be solved range unit to be detected to the fitting coefficient vector under basis matrix, α _lit is the column vector that num is capable; 1 _mrepresent the column vector that M is capable, 1 _min element be 1 entirely, represent Kronecker product operation;

Make array element error vector E _s=1 _n, 1 _nrepresent the column vector that N is capable, 1 _nin element be 1 entirely;

5b) by E _s=1 _nsubstitute into above-mentioned about in 2 norm optimization models of fitting coefficient vector, draw l range unit to be detected to the fitting coefficient vector α under basis matrix _l;

5c) according to sub-step 5b) the fitting coefficient vector α that draws _l, set up the following Optimized model about array element error vector:

$\underset{E_s}{\min }{\left|\right|x_l-\mathrm{diag}(E_s\⊗1_M)\mathrm{\ψ}{\mathrm{\α}}_l\left|\right|}_2$

Solve the above-mentioned Optimized model about array element error vector, draw and obtain the array element error vector E after upgrading _s;

5d) make iteration difference ${\mathrm{\δ}}_l=z_l-{\left|\right|x_l-\mathrm{diag}(E_s\⊗1_M)\mathrm{\ψ}{a}_l\left|\right|}_2,$ Make data norm z _lfor:

${\left|\right|x_l-\mathrm{diag}(E_s\⊗1_M)\ψa_l\left|\right|}_2;$

If 5e) z _lbe greater than σ _land δ _lbe greater than 0.01 σ _l, then sub-step 5b is back to), σ _lfor the noise level of setting; Otherwise, skip to sub-step 5f);

5f) by current E _svalue substitute in the computing formula of new clutter ridge steering vector basis matrix ψ ', draw the clutter ridge steering vector basis matrix ψ ' after matching, order matrix ψ ' ^*=ψ '; Obtain current E _sthe fitting coefficient vector α corresponding to value _l, by current E _sthe fitting coefficient vector α corresponding to value _lbe designated as the basis matrix optimal fitting coefficient of l range unit to be detected draw the remaining data x after the matching of l range unit to be detected _l-ψ ' ^*α _l ^*.