CN105223557A - Based on the airborne early warn ing radar clutter suppression method of accessory channel - Google Patents

Abstract

The invention discloses a kind of airborne early warn ing radar clutter suppression method based on accessory channel, its thinking is: obtain the three-dimensional echo data that airborne early warn ing radar receives, and obtain the clutter ridge of described three-dimensional echo data accordingly, and then obtain two-dimentional echo data; Again according to the clutter ridge of three-dimensional echo data, obtain num accessory channel and num search passage respectively, obtain spatial domain frequency vector corresponding to num accessory channel and temporal frequency vector corresponding to num accessory channel, and then obtain dimensionality reduction matrix corresponding to num accessory channel and matrix corresponding to num search passage successively, and the dimensionality reduction matrix of the accessory channel that is optimized accordingly and the transformation matrix based on accessory channel; Respectively dimension-reduction treatment is carried out to the matrix that described two-dimentional echo data and num search for passage corresponding, obtain the search passage steering vector after the echo data after dimensionality reduction and dimensionality reduction, and then obtain the filtered vector after dimensionality reduction, and accordingly clutter recognition process is carried out to the matrix of num search passage, obtain range Doppler figure.

Description

Based on the airborne early warn ing radar clutter suppression method of accessory channel

Technical field

The invention belongs to radar clutter suppression technology field, in particular to a kind of airborne early warn ing radar clutter suppression method based on accessory channel, be applicable to the decline problem solving airborne early warn ing radar clutter recognition performance in non-homogeneous clutter environment, improve its clutter recognition performance.

Background technology

Airborne early warn ing radar can be deployed in required place flexibly, rapidly with it and in widespread attention, its main task is the detection of a target in clutter background, and tracking is positioned to the target detected, and carry out effectively suppressing being the core improving airborne early warn ing radar serviceability to clutter.Therefore, before the target that locating and tracking detects, first need the clutter that exists in clutter reduction background or interference.If time under airborne early warn ing radar depending on being radiated at more smooth low scattering region, clutter or the interference of generation can be very weak, use conventional method process.But, the clutter spectrum generated due to the motion of airborne early warn ing radar is in the main lobe broadening of Doppler frequency domain and sidelobe clutter diffusion, make to present very strong empty time coupled characteristic, the clutter therefore needing to adopt space-time adaptive signal transacting (STAP) technology to suppress to produce or interference.Space-time adaptive process (STAP) technology can make full use of spatial information (si) and time-domain information, and can effective clutter reduction, but covariance matrix when almost cannot obtain abundant independent same distribution (independentandidenticallydistributed, IID) training sample as a rule to estimate sky.Even if obtain abundant independent same distribution number of training, also there is the difficulty that calculated amount and precision aspect are difficult to realize in the computing of inverting for high level matrix.

In the eighties, doctor R.Klemm of Germany has carried out ground-breaking theoretical research to space-time adaptive signal transacting (STAP) technology, he is by carrying out intensive analysis to noise performance, when finding empty, the number of the large eigenwert of covariance matrix is no more than N+M-1, wherein N is the array number of airborne early warn ing radar, M is the umber of pulse that airborne early warn ing radar is launched in a coherent processing inteval, illustrate that the full space-time adaptive signal transacting (STAP) being used for clutter reduction exists the possibility of dimension-reduction treatment really, on this basis, he proposes accessory channel method (AuxiliaryChannelReceiver-ACR), dimension after dimension-reduction treatment is by NM to N+M-1.Research shows this kind of dimension-reduction treatment in performance close to optimum full space time processing effect, but also there is following two problems in actual applications: first, dimension-reduction treatment in performance close to optimum full space time processing effect obtaining without when amplitude phase error, if consider spatial domain error, the clutter spectrum that airborne early warn ing radar produces can along spatial domain Directional Extension, its clutter dimension is obviously increased, and handling property obviously declines; The second, when the array number N of airborne early warn ing radar is larger, required processor dimension is also larger.Therefore, accessory channel method still needs to be optimized in actual applications or improves, thus clutter recognition performance is improved.

Summary of the invention

For above prior art Problems existing, the object of the invention is to propose a kind of airborne early warn ing radar clutter suppression method based on accessory channel, the method can solve traditional accessory channel method to the tolerance of spatial domain error difference and array number more time be difficult to the problem applied, and carry out dimension-reduction treatment by the utilization factor improving clutter ridge, thus the independent same distribution number of training needed for reducing, also can form recess at clutter ridge place simultaneously, improve clutter recognition effect.

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

Based on an airborne early warn ing radar clutter suppression method for accessory channel, it is characterized in that, comprise the following steps:

Step 1, obtains the three-dimensional echo data X that airborne early warn ing radar receives _{n × M × L}, and obtain the three-dimensional echo data X that airborne early warn ing radar receives accordingly _{n × M × L}the main beam spatial domain frequency θ that the clutter ridge formed and airborne early warn ing radar are launched _s, the three-dimensional echo data X then airborne early warn ing radar received _{n × M × L}rearrange by the mode of row, obtain the two-dimentional echo data X that airborne early warn ing radar receives _{nM × L}; Wherein, N is the element number of array of airborne early warn ing radar, and M is the umber of pulse that airborne early warn ing radar is launched in a coherent processing inteval, and L is the three-dimensional echo data X that airborne early warn ing radar receives _{n × M × L}range gate number;

Step 2, according to the three-dimensional echo data X that airborne early warn ing radar receives _{n × M × L}the clutter ridge formed, obtain num accessory channel and num search passage respectively, and then obtain the spatial domain frequency vector θ be made up of num accessory channel each self-corresponding spatial domain frequency, the temporal frequency vector be made up of num each self-corresponding temporal frequency of accessory channel respectively and num each self-corresponding time domain steering vector of search passage, then calculate the dimensionality reduction matrix T that num accessory channel is corresponding _b;

Step 3, according to the main beam spatial domain frequency θ that airborne early warn ing radar is launched _s, calculate the main beam spatial domain steering vector G that airborne early warn ing radar is launched _s, then according to num each self-corresponding time domain steering vector of search passage, appoint the time domain steering vector F getting a kth search passage _k, then according to the main beam spatial domain steering vector G that airborne early warn ing radar is launched _swith the time domain steering vector F of a kth search passage _k, calculate the search channel column vector S that a kth search passage is corresponding _k, and then obtain search access matrix S corresponding to num search passage; Wherein, k ∈ 1,2 ... num};

Step 4, searches for dimensionality reduction matrix T corresponding to search access matrix S and num accessory channel corresponding to passage according to num _b, calculate the optimization dimensionality reduction matrix that num_s × num_t accessory channel is corresponding and then the transformation matrix T calculated based on accessory channel; Wherein, k ∈ 1,2 ... num}, num_s are described optimization dimensionality reduction matrix the spatial frequency number comprised, num_t is described optimization dimensionality reduction matrix the Doppler frequency number comprised, num_s × num_t<<num;

Step 5, according to the transformation matrix T based on accessory channel, to the two-dimentional echo data X that airborne early warn ing radar receives _{nM × L}carry out dimension-reduction treatment respectively with num the search access matrix S searching for passage corresponding, obtain the echo data X after dimensionality reduction respectively _twith the search passage steering vector S after dimensionality reduction _t;

Step 6, according to the echo data X after dimensionality reduction _twith the search passage steering vector S after dimensionality reduction _t, calculate the filtering weight vector after dimensionality reduction;

Step 7, according to the filtering weight vector W after dimensionality reduction _t, to the echo data X after dimensionality reduction _tcarry out clutter recognition process, obtain range Doppler figure.

Compared with prior art, advantage of the present invention and improvement are:

First, the present invention utilizes clutter ridge information, different dimensionality reduction matrixes can be formed to different search passages, and by with search passage, there is identical Doppler frequency, the part accessory channel that forms along clutter ridge offsets the clutter component searching for passage, effectively improve the deficiency of original auxiliary channel algorithm to spatial domain error tolerance difference, improve practical application.

Second, the present invention utilizes the dimensionality reduction transition matrix based on accessory channel method, while doppler filtering process is carried out to the echo data of airborne early warn ing radar reception, also dimension-reduction treatment is carried out, effectively improve the clutter covariance matrix that original auxiliary channel algorithm causes because available independent same distribution number of training is not enough and estimate inaccurate problem, and the excessive problem increased with equipment cost of calculated amount that the degree of freedom of airborne early warn ing radar causes greatly, thus the sample number that can reduce required for estimate covariance matrix, make the present invention can while number of training deficiency, also clutter recognition performance can not be reduced.

Accompanying drawing explanation

Below in conjunction with the drawings and specific embodiments, the present invention is described in further detail.

Fig. 1 is the schematic flow sheet of a kind of airborne early warn ing radar clutter suppression method based on accessory channel of the present invention;

The range Doppler figure of Fig. 2 (a) for obtaining after use pulse Doppler (PD) algorithm process,

The range Doppler figure of Fig. 2 (b) for obtaining after original accessory channel method (ACR) process of use;

Fig. 3 is the range Doppler figure obtained after using the inventive method process;

The overall situation two dimension response diagram that Fig. 4 (a) obtains for using the inventive method,

The partial enlargement two dimension response diagram that Fig. 4 (b) obtains for using the inventive method;

Fig. 5 is the clutter afterpower comparison diagram obtained after using pulse Doppler (PD) algorithm, original accessory channel method (ACR) and the inventive method process respectively.

Embodiment

With reference to Fig. 1, be the schematic flow sheet of a kind of airborne early warn ing radar clutter suppression method based on accessory channel of the present invention, this kind, based on the airborne early warn ing radar clutter suppression method of accessory channel, comprises the following steps:

Step 1, obtains the three-dimensional echo data X that airborne early warn ing radar receives _{n × M × L}, and obtain the three-dimensional echo data X that airborne early warn ing radar receives accordingly _{n × M × L}the main beam spatial domain frequency θ that the clutter ridge formed and airborne early warn ing radar are launched _s, the three-dimensional echo data X then airborne early warn ing radar received _{n × M × L}rearrange by the mode of row, obtain the two-dimentional echo data X that airborne early warn ing radar receives _{nM × L}; Wherein, N is the element number of array of airborne early warn ing radar, and M is the umber of pulse that airborne early warn ing radar is launched in a coherent processing inteval, and L is the three-dimensional echo data X that airborne early warn ing radar receives _{n × M × L}range gate number.

Particularly, airborne early warn ing radar chooses positive side battle array airborne early warn ing radar, and positive side battle array airborne early warn ing radar antenna comprises N number of array element, and described N number of array element receives the three-dimensional echo data X of ground scatter body reflection _{n × M × L}, as the three-dimensional echo data X that acquisition airborne early warn ing radar receives _{n × M × L}; Wherein, N is the element number of array of airborne early warn ing radar, and M is the umber of pulse that airborne early warn ing radar is launched in a coherent processing inteval, and L is the three-dimensional echo data X that airborne early warn ing radar receives _{n × M × L}range gate number.

Wherein, because airborne early warn ing radar adopts positive side battle array airborne early warn ing radar, and the bay of positive side battle array airborne early warn ing radar is axially consistent with the heading of carrier aircraft, positive side battle array airborne early warn ing radar irradiates the main beam direction of ground scatter body and the axial cone angle cosine value formed of bay of positive side battle array airborne early warn ing radar, a kind of linear relationship with the Doppler frequency of ground scatter body echo, and described ground scatter body echo is exactly clutter, this clutter is in described cone angle cosine value and described Doppler frequency space, clutter distribution is straight line, using this straight line as clutter ridge, and obtain the main beam spatial domain frequency θ that airborne early warn ing radar is launched _sbe 0, the three-dimensional echo data X simultaneously airborne early warn ing radar received _{n × M × L}rearrange by the mode of row, obtain the two-dimentional echo data X that airborne early warn ing radar receives _{nM × L}.

Step 2, according to the three-dimensional echo data X that airborne early warn ing radar receives _{n × M × L}the clutter ridge formed, obtain num accessory channel and num search passage respectively, and then obtain the spatial domain frequency vector θ be made up of num accessory channel each self-corresponding spatial domain frequency, the temporal frequency vector be made up of num each self-corresponding temporal frequency of accessory channel respectively and num each self-corresponding time domain steering vector of search passage, then calculate the dimensionality reduction matrix T that num accessory channel is corresponding _b.

Particularly, three-dimensional echo data X airborne early warn ing radar received _{n × M × L}element number of array and airborne early warn ing radar launch in a coherent processing inteval umber of pulse composition space, as the three-dimensional echo data X that airborne early warn ing radar receives _{n × M × L}the two-dimensional space of array element-pulse domain, and receive three-dimensional echo data X according in the two-dimensional space of described array element-pulse domain along airborne early warn ing radar _{n × M × L}the clutter ridge formed, obtain num accessory channel and num search passage respectively, detailed process is:

If the three-dimensional echo data X that airborne early warn ing radar receives _{n × M × L}the clutter ridge slope formed is β, the three-dimensional echo data X that airborne early warn ing radar receives _{n × M × L}the maximal value of clutter ridge in the frequency of spatial domain formed is θ _max, the three-dimensional echo data X that airborne early warn ing radar receives _{n × M × L}the minimum value of clutter ridge in the frequency of spatial domain formed is θ _min, the three-dimensional echo data X that airborne early warn ing radar receives _{n × M × L}the maximal value of clutter ridge in Doppler frequency formed is the three-dimensional echo data X that airborne early warn ing radar receives _{n × M × L}the minimum value of clutter ridge in Doppler frequency formed is its expression formula is respectively:

${\mathrm{\&theta;}}_{\max }=\frac{d}{\mathrm{\&lambda;}}$

${\mathrm{\&theta;}}_{\min }=-\frac{d}{\mathrm{\&lambda;}}$

Wherein, N is the element number of array of airborne early warn ing radar, and M is the umber of pulse that airborne early warn ing radar is launched in a coherent processing inteval, and L is the three-dimensional echo data X that airborne early warn ing radar receives _{n × M × L}range gate number, V is carrier aircraft flying speed, and λ is wavelength, f _rfor pulse repetition rate, d represents the adjacent array element interval of airborne early warn ing radar.

Then, to the three-dimensional echo data X that airborne early warn ing radar receives _{n × M × L}the three-dimensional echo data X that the clutter ridge spatial domain frequency formed and airborne early warn ing radar receive _{n × M × L}the clutter ridge Doppler frequency formed evenly divides respectively, obtains num accessory channel and num search passage, and num each self-corresponding time domain steering vector of search passage.

If the spatial domain frequency of i-th accessory channel is θ _i, then the temporal frequency of i-th accessory channel is its expression formula is respectively:

${\mathrm{\&theta;}}_i=\frac{({\mathrm{\&theta;}}_{\max }-{\mathrm{\&theta;}}_{\min })}{num}(i-1)+{\mathrm{\&theta;}}_{\min },i\&Element;\{1,2...num\}$

The spatial domain frequency vector θ be then made up of num accessory channel each self-corresponding spatial domain frequency, and the temporal frequency vector be made up of num each self-corresponding temporal frequency of accessory channel expression formula be respectively:

θ＝[θ ₁,θ ₂,…θ _num]

Again according to the spatial domain frequency vector θ be made up of num accessory channel each self-corresponding spatial domain frequency, and the temporal frequency vector be made up of num each self-corresponding temporal frequency of accessory channel calculate the spatial domain steering vector column vector G be made up of num accessory channel each self-corresponding spatial domain steering vector respectively _bwith the time domain steering vector column vector F be made up of num each self-corresponding time domain steering vector of accessory channel _b, its expression is respectively:

G _b＝[1；e ^j2πθ；…；e ^j(N-1)2πθ]

And then, calculate the dimensionality reduction matrix T that num accessory channel is corresponding _b, its expression formula is:

$T_b=G_b\&CircleTimes;F_b$

Wherein, N is the element number of array of airborne early warn ing radar, G _bfor the spatial domain steering vector column vector be made up of num accessory channel each self-corresponding spatial domain steering vector, F _bfor the time domain steering vector column vector be made up of num each self-corresponding time domain steering vector of accessory channel, T _bfor the dimensionality reduction matrix that num accessory channel is corresponding, for Kronecker amasss sign of operation.

Step 3, according to the main beam spatial domain frequency θ that airborne early warn ing radar is launched _s, calculate the main beam spatial domain steering vector G that airborne early warn ing radar is launched _s, then according to num each self-corresponding time domain steering vector of search passage, appoint the time domain steering vector F getting a kth search passage _k, then according to the main beam spatial domain steering vector G that airborne early warn ing radar is launched _swith the time domain steering vector F of a kth search passage _k, calculate the search channel column vector S that a kth search passage is corresponding _k, and then obtain search access matrix S corresponding to num search passage; Wherein, k ∈ 1,2 ... num}.

Particularly, the search channel column vector S that in num search passage, a kth search passage is corresponding _kthe main beam spatial domain steering vector G launched according to airborne early warn ing radar _swith the time domain steering vector F of a kth search passage _kcalculate, its expression formula is:

$S_k=G_s\&CircleTimes;F_k$

Wherein, G _sfor the main beam spatial domain steering vector that airborne early warn ing radar is launched, and the main beam spatial domain steering vector of num each self-corresponding airborne early warn ing radar transmitting of search passage is the same; F _kfor the time domain steering vector of a kth search passage, S _kfor the search channel column vector that a kth search passage is corresponding, for Kronecker amasss sign of operation.

The main beam spatial domain steering vector G that airborne early warn ing radar is launched _swith the time domain steering vector F of a kth search passage _kexpression formula be respectively:

$G_s=\&lsqb;1;e^{j2{\mathrm{\&pi;\&theta;}}_s};...;e^{j(N-1)2{\mathrm{\&pi;\&theta;}}_s}\&rsqb;$

θ _s＝0

Wherein, θ _sfor the main beam spatial domain frequency that airborne early warn ing radar is launched, for the temporal frequency of a kth search passage, N is the element number of array of airborne early warn ing radar, and M is the umber of pulse that airborne early warn ing radar is launched in a coherent processing inteval.

The search channel column vector S corresponding according to a kth search passage _k, obtain the search access matrix S that num search passage is corresponding.

Step 4, searches for dimensionality reduction matrix T corresponding to search access matrix S and num accessory channel corresponding to passage according to num _b, calculate the optimization dimensionality reduction matrix that num_s × num_t accessory channel is corresponding and then the transformation matrix T calculated based on accessory channel; Wherein, k ∈ 1,2 ... num}, num_s are described optimization dimensionality reduction matrix the spatial frequency number comprised, num_t is described optimization dimensionality reduction matrix the Doppler frequency number comprised, num_s ∈ 1,2 ... num}num_t ∈ 1,2 ... num}, num_s × num_t<<num.

Particularly, if in num accessory channel i-th (i ∈ 1,2 ... num}) Doppler frequency of accessory channel is then choose (i-(num_t-1)/2) in num accessory channel individual to each self-corresponding Doppler frequency of (i+ (num_t-1)/2) individual accessory channel, as the Doppler frequency optimizing accessory channel, choose (i-(num_s-1)/2) in num accessory channel individual to each self-corresponding spatial frequency of (i+ (num_s-1)/2) individual accessory channel, as the spatial frequency optimizing accessory channel, then optimize accessory channel and comprise num_s spatial frequency and num_t Doppler frequency, and the spatial domain frequency vector θ of the accessory channel that is optimized accordingly _s'with the temporal frequency vector optimizing accessory channel its expression formula is respectively:

θ _s'＝[θ _{i-(num_s-1)/2},θ _{i-[(num_s-1)/2]+1},…θ _i,…,θ _{i+[(num_s-1)/2]-1},θ _{i+(num_s-1)/2}]

Generally, the relation of num_s, num_t and num meets num_s × num_t<<num, and num_s and num_t gets odd number value respectively.

According to the spatial domain frequency vector θ optimizing accessory channel _s'with the temporal frequency vector optimizing accessory channel calculate the spatial domain steering vector G optimizing accessory channel respectively _b'with the time domain steering vector F optimizing accessory channel _b', its expression formula is respectively:

$G_{b^{\&prime;}}=\&lsqb;1;e^{j2{\mathrm{\&pi;\&theta;}}_{s^{\&prime;}}};...e^{j(N-1)2{\mathrm{\&pi;\&theta;}}_{s^{\&prime;}}}\&rsqb;$

And then calculate optimization dimensionality reduction matrix corresponding to num_s × num_t accessory channel its expression formula is:

$T_{b^{*}}=G_{b^{\&prime;}}\&CircleTimes;F_{b^{\&prime;}}$

Wherein, N is the element number of array of airborne early warn ing radar, and M is the umber of pulse that airborne early warn ing radar is launched in a coherent processing inteval.

Therefore, the transformation matrix T based on accessory channel searches for optimization dimensionality reduction matrix corresponding to search access matrix S and num_s*num_t accessory channel corresponding to passage according to num calculate, its expression formula is:

$T={\left[\begin{array}{cc}{S}^H& {T_{b^{*}}}^H\end{array}\right]}^H$

Wherein, S is the search access matrix that num search passage is corresponding, for the optimization dimensionality reduction matrix that num_s × num_t accessory channel is corresponding, it is the dimensionality reduction matrix T that num accessory channel is corresponding _bsubset, num_s optimizes the spatial frequency number that comprises of accessory channel, and num_t optimizes the Doppler frequency number that accessory channel comprises, and T is the transformation matrix based on accessory channel, and H represents conjugate transpose.

Step 5, according to the transformation matrix T based on accessory channel, to the two-dimentional echo data X that airborne early warn ing radar receives _{nM × L}carry out dimension-reduction treatment respectively with num the search access matrix S searching for passage corresponding, obtain the echo data X after dimensionality reduction respectively _twith the search passage steering vector S after dimensionality reduction _t.

Particularly, according to the transformation matrix T based on accessory channel, to the two-dimentional echo data X that airborne early warn ing radar receives _{nM × L}carry out dimension-reduction treatment respectively with num the search access matrix S searching for passage corresponding, obtain the echo data X after dimensionality reduction _twith the search passage steering vector S after dimensionality reduction _t, its expression formula is respectively:

X _T＝T ^HX _NM×L

S _T＝T ^HS _k

Wherein, X _tfor the echo data after dimensionality reduction, T is the transformation matrix based on accessory channel, X _{nM × L}for the two-dimentional echo data that airborne early warn ing radar receives, S _trepresent the search passage steering vector after dimensionality reduction, S is the search access matrix S that num search passage is corresponding, and H represents conjugate transpose.

Through dimension-reduction treatment back echo data X _tdimension by NM dimensionality reduction to ((num_s × num_t)+1), thus realize the dimension-reduction treatment of the two-dimentional echo data that airborne early warn ing radar receives.

Step 6, according to the echo data X after dimensionality reduction _twith the search passage steering vector S after dimensionality reduction _t, the filtering weight vector after utilizing maximum likelihood method to calculate dimensionality reduction.

Particularly, according to the echo data X after dimensionality reduction _twith the search passage steering vector S after dimensionality reduction _t, the echo data X after utilizing maximum likelihood method to estimate dimensionality reduction _tcovariance matrix, namely utilize the echo data X after dimensionality reduction _tas independent same distribution training sample, estimate the echo data X after obtaining dimensionality reduction by maximum likelihood method _tcovariance matrix R _t, its expression formula is:

$R_T=\frac{1}{L1}\underset{j=1}{\overset{L1}{\&Sigma;}}{X}_{Tj}{X_{Tj}}^H$

L1＝2×num_s×num_t

Wherein, R _trepresent the echo data X after dimensionality reduction _tcovariance matrix, L1 represents the independent same distribution number of samples required for echo data of pending search passage, X _tjrepresent the echo data after the dimensionality reduction of a jth call number, j represents the echo data X after dimensionality reduction _ta jth call number, H represents conjugate transpose.

The computing formula of self-adaptation weight vector W is:

$W=\frac{R_{Per}^{-1}{S}_{Per}}{{S_{Per}}^{\&prime;}{R}_{Per}^{-1}{S}_{Per}}$

Wherein, R _perfor the original covariance matrix using traditional accessory channel method to obtain, S _perthe initial search passage steering vector obtained for using traditional accessory channel method.

Then, by the echo data X after dimensionality reduction _tcovariance matrix R _treplace the original covariance matrix R using traditional accessory channel method to obtain _per, with the search passage steering vector S after dimensionality reduction _treplace the initial search passage steering vector S using traditional accessory channel method to obtain _per, and calculate the filtering weight vector W after dimensionality reduction _t, its expression formula is:

$W_T=\frac{R_T^{-1}{S}_T}{{S_T}^{\&prime;}{R}_T^{-1}{S}_T}$

Wherein, W _trepresent the filtering weight vector after dimensionality reduction, R _tfor the echo data X after dimensionality reduction _tcovariance matrix, S _tfor the search passage steering vector after dimensionality reduction.

Step 7, according to the filtering weight vector W after dimensionality reduction _t, and utilize space-time adaptive processing method to the echo data X after dimensionality reduction _tcarry out clutter recognition process, obtain range Doppler figure.

Particularly, a kth search passage forms the individual transformation matrix based on accessory channel method of kth, num each self-corresponding transformation matrix based on accessory channel method of search passage, form the individual different transformation matrix based on accessory channel method of num, thus form the individual different filtering weight vector of num, the filtering weight vector W namely after dimensionality reduction _t.Therefore, according to the filtering weight vector W after dimensionality reduction _t, and utilize space-time adaptive process (STAP) method to the echo data X after dimensionality reduction _tcarry out clutter recognition process, obtain the echo data Y after clutter recognition process, and the echo data Y after clutter recognition process is exported, obtain range Doppler figure.

The expression formula of the echo data Y after described clutter recognition process is:

Y＝W _T ^HX _T

Wherein, W _tfor the filtering weight vector after dimensionality reduction, X _tfor the echo data after dimensionality reduction, H represents conjugate transpose.

Below in conjunction with emulation experiment, further explanation is verified to effect of the present invention.

(1) echo data emulation and experiment condition:

Emulation experiment of the present invention is carried out under MATLAB7.11 software, in emulation experiment of the present invention, the linear array that the antenna of airborne early warn ing radar adopts 128 array element evenly distributed, the ratio of adjacent array element distance and wavelength d/ λ is 0.5, the antenna normal direction of main beam pointing airborne early warn ing radar, namely be 0 ° with front normal direction angle, namely positive side array antenna.In emulation experiment of the present invention, the three-dimensional echo data that the airborne early warn ing radar of use receives is that the clutter model simulation proposed according to Lincoln laboratory J.Ward produces, and adds white Gaussian noise, and concrete simulation parameter is as shown in table 1:

Table 1

Carrier aircraft height	8km
		Carrier aircraft speed	150m/s
Umber of pulse	128
		Wavelength	0.1m
Pulse repetition rate	8000Hz
		Antenna axial direction and carrier aircraft velocity angle	0°
Miscellaneous noise ratio	70dB
		Range gate number	500
Main beam pointing and the axial angle of bay	90°

(2) content is emulated

In order to the superiority improving accessory channel algorithm is described, Fig. 2 (a) and Fig. 2 (b) gives the result of other several algorithm, wherein Fig. 2 (a) is the range Doppler figure obtained after using existing pulse Doppler (PD) algorithm process, Fig. 2 (b) is the range Doppler figure obtained after using the process of original accessory channel method (ACR).

The transverse axis of Fig. 2 (a) and Fig. 2 (b) represents Doppler's channel position respectively, and the longitudinal axis represents range gate sequence number respectively; White portion in Fig. 2 (a) is the main clutter afterpower distribution obtained after using existing pulse Doppler (PD) algorithm process, color is use the sidelobe clutter afterpower distribution obtained after existing pulse Doppler (PD) algorithm process compared with shallow portion region, and black region be that the noise afterpower obtained after using existing pulse Doppler (PD) algorithm process distributes; White portion in Fig. 2 (b) is the clutter afterpower distribution after using the process of original accessory channel method (ACR), and black region is the noise afterpower distribution after existing algorithm ACR process.

Can find out that stronger residual spur occupies more distance-Doppler unit area from Fig. 2 (a), Radar Targets'Detection in direct impact described distance-doppler cells region, can find out that white portion and the more shallow region of color greatly reduce from Fig. 2 (b), show that the clutter component in echo data obtains suppression.

Fig. 3 is the range Doppler figure obtained after using the inventive method process.White portion in Fig. 3 and the more shallow region of color greatly reduce, show that the clutter component in the three-dimensional echo data that echo data airborne early warn ing radar receives obtains good suppression, and (2* (num+1)-3) of independent same distribution number of training required for original accessory channel method (ACR) is individual, and to reduce to (2* (num_s*num_t+1)-3) wanted required for the present invention individual, reduces required independent same distribution training sample number.

The overall situation two dimension response diagram that Fig. 4 (a) obtains for using the inventive method, Fig. 4 (b) is the partial enlargement two dimension response diagram using the inventive method to obtain; The transverse axis of Fig. 4 (a) and Fig. 4 (b) represents normalization Doppler frequency respectively, and the longitudinal axis represents normalization spatial frequency respectively.Can see that from Fig. 4 (b) the clutter ridge that the three-dimensional echo data using the inventive method can receive at airborne early warn ing radar is formed forms recess.

Fig. 5 is the clutter afterpower comparison diagram obtained after using existing algorithm PD, ACR and the inventive method process respectively; The transverse axis of Fig. 5 represents Doppler's passage, and the longitudinal axis represents clutter afterpower; The clutter afterpower that solid line in Fig. 5 obtains after representing use existing pulse Doppler (PD) algorithm process, the clutter afterpower of dotted line for obtaining after original accessory channel method (ACR) process of use, pecked line is the clutter afterpower obtained after using the inventive method process.

What the inventive method reduced relative to the main lobe gain of original auxiliary channel algorithm as can see from Figure 5 lacks, good to spatial domain error tolerance, and less independent same distribution number of training suppressed sidelobes clutter can be utilized, and main-lobe clutter district narrows, and obtains better clutter recognition effect.

This simulation result shows, the present invention can obtain clutter recognition effect better when reducing training sample and array element error.

In sum, Simulation experiments validate correctness of the present invention, validity and reliability.

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 (10)

1., based on an airborne early warn ing radar clutter suppression method for accessory channel, it is characterized in that, comprise the following steps:

Step 1, obtains the three-dimensional echo data X that airborne early warn ing radar receives _{n × M × L}, and obtain the three-dimensional echo data X that airborne early warn ing radar receives accordingly _{n × M × L}the main beam spatial domain frequency θ that the clutter ridge formed and airborne early warn ing radar are launched _s, the three-dimensional echo data X then airborne early warn ing radar received _{n × M × L}rearrange by the mode of row, obtain the two-dimentional echo data X that airborne early warn ing radar receives _{nM × L}; Wherein, N is the element number of array of airborne early warn ing radar, and M is the umber of pulse that airborne early warn ing radar is launched in a coherent processing inteval, and L is the three-dimensional echo data X that airborne early warn ing radar receives _{n × M × L}range gate number;

Step 2, according to the three-dimensional echo data X that airborne early warn ing radar receives _{n × M × L}the clutter ridge formed, obtain num accessory channel and num search passage respectively, and then obtain the spatial domain frequency vector θ be made up of num accessory channel each self-corresponding spatial domain frequency, the temporal frequency vector be made up of num each self-corresponding temporal frequency of accessory channel respectively and num each self-corresponding time domain steering vector of search passage, then calculate the dimensionality reduction matrix T that num accessory channel is corresponding _b;

Step 3, according to the main beam spatial domain frequency θ that airborne early warn ing radar is launched _s, calculate the main beam spatial domain steering vector G that airborne early warn ing radar is launched _s, then according to num each self-corresponding time domain steering vector of search passage, appoint the time domain steering vector F getting a kth search passage _k, then according to the main beam spatial domain steering vector G that airborne early warn ing radar is launched _swith the time domain steering vector F of a kth search passage _k, calculate the search channel column vector S that a kth search passage is corresponding _k, and then obtain search access matrix S corresponding to num search passage; Wherein, k ∈ 1,2 ... num};

Step 4, searches for dimensionality reduction matrix T corresponding to search access matrix S and num accessory channel corresponding to passage according to num _b, calculate the optimization dimensionality reduction matrix that num_s × num_t accessory channel is corresponding and then the transformation matrix T calculated based on accessory channel; Wherein, k ∈ 1,2 ... num}, num_s are described optimization dimensionality reduction matrix the spatial frequency number comprised, num_t is described optimization dimensionality reduction matrix the Doppler frequency number comprised, num_s × num_t<<num;

Step 5, according to the transformation matrix T based on accessory channel, to the two-dimentional echo data X that airborne early warn ing radar receives _{nM × L}carry out dimension-reduction treatment respectively with num the search access matrix S searching for passage corresponding, obtain the echo data X after dimensionality reduction respectively _twith the search passage steering vector S after dimensionality reduction _t;

Step 6, according to the echo data X after dimensionality reduction _twith the search passage steering vector S after dimensionality reduction _t, calculate the filtering weight vector after dimensionality reduction;

Step 7, according to the filtering weight vector W after dimensionality reduction _t, to the echo data X after dimensionality reduction _tcarry out clutter recognition process, obtain range Doppler figure.

2. a kind of airborne early warn ing radar clutter suppression method based on accessory channel as claimed in claim 1, it is characterized in that, in step 2, described be made up of num accessory channel each self-corresponding spatial domain frequency spatial domain frequency vector θ, be made up of num each self-corresponding temporal frequency of accessory channel temporal frequency vector its expression formula is respectively:

Wherein, i ∈ { 1,2 ... num}, θ _maxfor the three-dimensional echo data X that airborne early warn ing radar receives _{n × M × L}the maximal value of clutter ridge in the frequency of spatial domain formed, θ _minfor the three-dimensional echo data X that airborne early warn ing radar receives _{n × M × L}the minimum value of clutter ridge in the frequency of spatial domain formed, for the three-dimensional echo data X that airborne early warn ing radar receives _{n × M × L}the maximal value of clutter ridge in Doppler frequency formed, for the three-dimensional echo data X that airborne early warn ing radar receives _{n × M × L}the minimum value of clutter ridge in Doppler frequency formed, N is the element number of array of airborne early warn ing radar, and M is the umber of pulse that airborne early warn ing radar is launched in a coherent processing inteval, and L is the three-dimensional echo data X that airborne early warn ing radar receives _{n × M × L}range gate number.

3. a kind of airborne early warn ing radar clutter suppression method based on accessory channel as claimed in claim 1, is characterized in that, in step 2, described in obtain the dimensionality reduction matrix T that is made up of num accessory channel _b, its process is:

According to the spatial domain frequency vector θ be made up of num accessory channel each self-corresponding spatial domain frequency, and the temporal frequency vector be made up of num each self-corresponding temporal frequency of accessory channel calculate the spatial domain steering vector column vector G be made up of num accessory channel each self-corresponding spatial domain steering vector respectively _bwith the time domain steering vector column vector F be made up of num each self-corresponding time domain steering vector of accessory channel _b, its expression is respectively:

And then, calculate the dimensionality reduction matrix T that num accessory channel is corresponding _b, its expression formula is:

$T_b=G_b\&CircleTimes;F_b$

Wherein, N is the element number of array of airborne early warn ing radar, G _bfor the spatial domain steering vector column vector be made up of num accessory channel each self-corresponding spatial domain steering vector, F _bfor the time domain steering vector column vector be made up of num each self-corresponding time domain steering vector of accessory channel, T _bfor the dimensionality reduction matrix that num accessory channel is corresponding, for Kronecker amasss sign of operation.

4. a kind of airborne early warn ing radar clutter suppression method based on accessory channel as claimed in claim 1, is characterized in that, in step 3, the described search access matrix S be made up of num search passage, its process is:

The search channel column vector S that in num search passage, a kth search passage is corresponding _kthe main beam spatial domain steering vector G launched according to airborne early warn ing radar _swith the time domain steering vector F of a kth search passage _kcalculate, its expression formula is:

$S_k=G_s\&CircleTimes;F_k$

Wherein, G _sfor the main beam spatial domain steering vector that airborne early warn ing radar is launched, and the main beam spatial domain steering vector of num each self-corresponding airborne early warn ing radar transmitting of search passage is the same; F _kfor the time domain steering vector of a kth search passage, S _kfor the search channel column vector that a kth search passage is corresponding, for Kronecker amasss sign of operation;

The main beam spatial domain steering vector G that airborne early warn ing radar is launched _swith the time domain steering vector F of a kth search passage _kexpression formula be respectively:

Wherein, θ _sfor the main beam spatial domain frequency that airborne early warn ing radar is launched, for the temporal frequency of a kth search passage, N is the element number of array of airborne early warn ing radar, and M is the umber of pulse that airborne early warn ing radar is launched in a coherent processing inteval;

The search channel column vector S corresponding according to a kth search passage _k, obtain the search access matrix S that num search passage is corresponding.

5. a kind of airborne early warn ing radar clutter suppression method based on accessory channel as claimed in claim 1, is characterized in that, in step 4, described in obtain optimization dimensionality reduction matrix corresponding to num_s*num_t accessory channel its expression formula is:

$T_{b^{*}}=G_{b^{\&prime;}}\&CircleTimes;F_{b^{\&prime;}}$

Wherein,

num_s is the spatial frequency number that optimization accessory channel comprises, num_t is the Doppler frequency number that optimization accessory channel comprises, i ∈ { 1,2 ... num}, N is the element number of array of airborne early warn ing radar, and M is the umber of pulse that airborne early warn ing radar is launched in a coherent processing inteval.

6. a kind of airborne early warn ing radar clutter suppression method based on accessory channel as claimed in claim 1, is characterized in that, in step 4, the described transformation matrix T based on accessory channel, its expression formula is:

$T={\left[\begin{array}{cc}{S}^H& {T_{b*}}^H\end{array}\right]}^H$

Wherein, S is the search access matrix that num search passage is corresponding, for the optimization dimensionality reduction matrix that num_s × num_t accessory channel is corresponding, it is the dimensionality reduction matrix T that num accessory channel is corresponding _bsubset, num_s optimizes the spatial frequency number that comprises of accessory channel, and num_t optimizes the Doppler frequency number that accessory channel comprises, and T is the transformation matrix based on accessory channel, and H represents conjugate transpose.

7. as claim 1 or claim 5 or a kind of airborne early warn ing radar clutter suppression method based on accessory channel according to claim 6, it is characterized in that, in step 4, described num_s optimizes the spatial frequency number that comprises of accessory channel and described num_t to optimize the Doppler frequency number that accessory channel comprises, and its pass is:

num_s×num_t<<num

Further, num_s and num_t gets odd number value respectively; Wherein, num is search channel number, and num is also accessory channel number.

8. a kind of airborne early warn ing radar clutter suppression method based on accessory channel as claimed in claim 1, is characterized in that, in steps of 5, and the echo data X after described dimensionality reduction _twith the search passage steering vector S after dimensionality reduction _t, its expression formula is respectively:

X _T＝T ^HX _NM×L

S _T＝T ^HS

Wherein, X _tfor the echo data after dimensionality reduction, T is the transformation matrix based on accessory channel, X _{nM × L}for the two-dimentional echo data that airborne early warn ing radar receives, S _trepresent the search passage steering vector after dimensionality reduction, S is the search access matrix that num search passage is corresponding, and H represents conjugate transpose.

9. a kind of airborne early warn ing radar clutter suppression method based on accessory channel as claimed in claim 1, is characterized in that, in step 6, described in obtain the filtering weight vector W after dimensionality reduction _tcomprise the individual different filtering weight vector of num, its feature is: a kth search passage forms the individual transformation matrix based on accessory channel method of kth, num each self-corresponding transformation matrix based on accessory channel method of search passage, form the individual different transformation matrix based on accessory channel method of num, thus form the individual different filtering weight vector of num.

10. a kind of airborne early warn ing radar clutter suppression method based on accessory channel as claimed in claim 1, is characterized in that, in step 7, described in obtain range Doppler figure, its process is:

According to the filtering weight vector W after dimensionality reduction _t, and utilize space-time adaptive treatment S TAP method to the echo data X after dimensionality reduction _tcarry out clutter recognition process, obtain the echo data Y after clutter recognition process, and the echo data Y after clutter recognition process is exported, obtain range Doppler figure; Echo data Y after described clutter recognition process, its expression formula is:

Y＝W _T ^HX _T

Wherein, W _tfor the filtering weight vector after dimensionality reduction, X _tfor the echo data after dimensionality reduction, H represents conjugate transpose.