CN103698753A - Passive passage correcting method of small-size array - Google Patents

Abstract

The invention relates to a passive passage correcting method of a small-size array. According to the method, under certain signal-to-noise requirements, the stable and reliable single information source echo is determined according to the size of a characteristic value after the strong echo characteristic decomposition; the single information source echo snapshot data is used for estimating the passage amplitude gain factor; the single information source ocean echo snapshot data subjected to the amplitude calibration and the existing array information are utilized for estimating the passage phase error factor, and the phase correction is realized. The passive passage correcting method has the advantages that the influence such as multipath effect and boat echo interference in the echo can be avoided, and the requirement on the quality of the ocean echo data is lower, so the better applicability to the radar work environment change can be realized, and the long-time continuous work can be realized; the high-intensity strong single information source characteristic ocean echo is utilized, and the correction results have good precision and robustness; the calculation quantity is small, the real-time performance is high, and the long-time stable work can be realized; the application flexibility of a radar is greatly improved; the detection performance is improved, and meanwhile, the development cost and the maintenance cost of the radar are greatly reduced.

Description

A kind of method for correcting passive channels of small array

Technical field

The present invention relates to a kind of method for correcting passive channels of small array, is to utilize single information source marine echo data that high-frequency ground wave radar receives planar array to be carried out to the method for passive channel correction.

Background technology

High-frequency ground wave radar (HF Ground Wave Radar) utilizes vertical polarization frequency electromagnetic waves to decay little feature and ocean surface to the single order of electric wave, second order dispersion mechanism at conduction ocean surface diffraction propagation, realizes drive marine mathematic(al) parameter and the marine moving targets such as naval vessel, aircraft such as over-the-horizon detection wind field, Lang Chang, flow field.

Due to receiving antenna in Radar Receiver System, the gain inconsistency of AFE (analog front end), the factors such as the impact of periphery electromagnetic environment, the magnitude-phase characteristics in actual reception echo between each passage there are differences, and is referred to as passage amplitude phase error.Passage amplitude phase error does not conform to the actual conditions the battle array Flow model that beamforming algorithm and orientation algorithm for estimating adopt, and very little error also can make the performance degradation of this class algorithm, is one of key issue affecting high-frequency ground wave radar detection performance.

Existing bearing calibration mainly contains two classes: active correction method and passive bearing calibration.Active correction method is passed through at array

The place ahead arranges accurately known auxiliary information source of orientation, the correction of the auxiliary information source echo receiving by radar and its known orientation information acquisition passage amplitude phase error.In passive bearing calibration, by estimation being combined to amplitude phase error in the orientation of space information source according to certain majorized function, do not need the known auxiliary source of azimuth information.Correlation technique elaborates in < < Estimation of Spatial Spectrum theory and algorithm > > (publishing house of Tsing-Hua University 2004).

Sea state high-frequency ground wave radar array operated by rotary motion is on bank, as adopted active correction method, need on the island in array the place ahead or ship, auxiliary source be set, with high costs, arranges with maintenance difficulties very greatly, and is difficult to guarantee real-time stabilization work.Passive bearing calibration generally need to be carried out repeatedly complicated interative computation, and calculated amount is larger, not necessarily guarantees requirement of real-time.And when initial amplitude phase error information is not enough, optimizing algorithm very easily converges to local optimum, is difficult to practicality.

Wuhan University's wave propagation laboratory once considered to use marine known natural or artificial object to the reflected signal of radar wave as correction signal.Thereby from echo, extract according to its known azran information the mismatching that correction signal is estimated each passage.Its concrete implementation detail can referring to Chinese invention patent " a kind of method of utilizing marine echo to carry out array channel calibration " (patent No.: CN03128238.5), " a kind of method of utilizing Ionospheric Echo to carry out higher-frequency radar aerial array channel correcting " (patent No.: CN200610018271.5) and " a kind of multichannel HF skywave radar receiving channels calibration system " (patent No.: CN201120517053.2).Utilize the fixation reflex things such as known island, beacon and drilling platform or the ionosphere surveyed in marine site directly to launch echo, do not need to consider setting and the maintenance issues of auxiliary source, can realize online real time correction, there is certain practical value.But this two actual be special active correction method, range of application and actual effect are limited, are not suitable for the exposed waters without fixation reflex thing, and are still subject to noise and ship echo, the impact of the unfavorable factors such as multipath effect.

Chinese invention patent " a kind of passive bearing calibration based on the non-linear antenna array " (patent No.: CN200610071360.6)

Aerial array is set to the non-rectilinear form that contains the even group of translation invariant array element, by translation invariant battle array, is detected and is obtained a large amount of single information source echoes, is obtained the amplitude phase error coefficient of passage, and then realize channel correcting by these echoes.By structure maximum likelihood cost function, adopt the initial value iterative method of low dimension local optimum to obtain global optimum.But the method need to be arranged to array to meet translation invariant form, formation design is upper restricted.When echo quality is lower slightly, algorithm easily converges to local optimum, and precision is not high, and this makes its result real-time on proofreading and correct poor.

Summary of the invention

Above-mentioned technical matters of the present invention is mainly solved by following technical proposals:

A method for correcting passive channels for small array, is characterized in that, comprises the following steps:

Step 1, the second-order statistics based on Array Model, the eigenwert of decomposing according to strong echo character size is determined reliable and stable single information source echo; Concrete grammar is: by strong signal to noise ratio (S/N ratio) echo structure array received autocorrelation matrix, carry out by eigenwert S relation, determining single information source echo after feature decomposition, work as S ₁/ S ₂>=S _i/ S _i+1i=2 ... during N, be defined as single information source echo.

Described array received autocorrelation matrix is in the four-dimensional echo data that conversion obtains through secondary FFT, distance element and the array snap autocorrelation matrix that in Doppler frequency, independently strong signal to noise ratio (S/N ratio) marine echo is corresponding;

Step 2, utilize the fast beat of data of each passage list information source echo energy and ratio estimating channel amplitude gain coefficient, and according to the channel amplitude gain coefficient of estimating, the fast beat of data of each passage list information source echo is carried out to amplitude self-correcting; The fast beat of data of described each passage list information source echo refers to the fast beat of data in respective channel in the array received autocorrelation matrix of step 1 time all single information source echoes.

It should be noted that: it is passage that array has one of two dimension from phase matrix, another is snap.Each single information source echo is not always the case, and is exactly " the fast beat of data of each passage list information source echo " by all single information source snap aggregation of data in corresponding passage dimension together.

Step 3, utilizes in step 2 the fast beat of data of single information source marine echo after amplitude calibration and known array information by by the major function of asking of linearity

pSO global optimization search estimated initial phases error; Wherein, θ=[θ ₁θ ₂θ _m] ^tfor the angle of arrival of all information sources, Φ=[Φ ₁Φ ₂Φ _n] ^tfor phase error coefficient,

for the steering vector matrix of all information sources,

for each information source power of estimating, λ is wavelength, (x _n, y _n) be the coordinate of n antenna;

Step 4, is searched for by the sue for peace iteration optimization of MUSIC cost function of the initial phase error coefficient of estimating in step 3, realizes phase place high-precision correction; Specifically: utilize in step 2 the fast beat of data of single information source echo of amplitude correction, the initial phase error that the global optimization search obtaining in known array position information and step 3 obtains, the optimization phase error coefficient carrying out based on summation MUSIC function estimates, optimized results is based on formula:

by this optimization, realize phase place self-correcting, wherein U (θ _i) be the noise vector space of single information source i; Wherein, θ _ibe the angle of arrival of i information source, a (θ _i) be i the steering vector that information source is corresponding, Φ=[Φ ₁Φ ₂Φ _n] ^tfor phase error coefficient,

for complex matrix corresponding to phase error to be estimated.

At the method for correcting passive channels of above-mentioned a kind of small array, in described step 2, utilize the fast beat of data of single information source echo receive to carry out the estimation of channel amplitude gain coefficient, the amplitude gain coefficient of each array element directly by

estimate, realize amplitude self-correcting.Does (it is just passable that amplitude self-correcting only provide this formula? if so, need to supplement the parameter-definition of this formula)

At the method for correcting passive channels of above-mentioned a kind of small array, the concrete grammar of described step 3 comprises following sub-step:

The phase error to be estimated of step 3.1, each particle of random initializtion and angle of arrival value P _i=[θ, Φ] and initial velocity

iterations k=0, i is population;

Step 3.2, by

calculate each particle cost function value F _i;

Step 3.3, by current particle information, obtained the traversal optimal value O of each particle _iwith particle optimum solution

global optimum O _dwith globally optimal solution

Step 3.4, calculate each particle traversal search speed

each particle is carried out to the renewal of solve for parameter, based on following formula:

$v_i^{k+1}={\mathrm{wv}}_i^k+c_1{r}_1(g_i^k-P_i^k)+c_2{r}_2(g_d^k-P_i^k);$

$P_i^{k+1}=P_i^k+v_i^{k+1};$

Wherein, w is primary particle speed weight, c ₁, c ₂be respectively Local Search speed and global search speed weight, r ₁, r ₂it is the random number between 0 to 1;

In particle position traversal, for guaranteeing the ergodicity of global search, the size of restriction particle rapidity is no more than the threshold value of setting;

If κ meets for the particle rapidity that is greater than this value, adopt following formula method restriction particle rapidity: $v_i^{k+1}=\frac{\mathrm{\&kappa;}}{\left|v_i^{k+1}\right|}{v}_i^{k+1}$

Step 3.5, k=k+1, if k<200 goes to step 2, otherwise with globally optimal solution g _dfor output, finish traversal.

At the method for correcting passive channels of above-mentioned a kind of small array, in described step 4, based on summation MUSIC cost function

optimization phase error estimation method adopts iteration convergence method, and its step is as follows:

Step 4.1 first, is carried out the estimation of DOA under current phase error, based on following formula:

$P_m\left(\mathrm{\&theta;}\right)=\frac{1}{{\left|\right|U^H\left({\mathrm{\&theta;}}_i\right)\mathrm{Ca}\left(\mathrm{\&theta;}\right)\left|\right|}^2};$

Wherein, U (θ _i) be the noise vector space of single information source i, θ _ibe the angle of arrival of i information source, a (θ _i) be i the steering vector that information source is corresponding, Φ=[Φ ₁Φ ₂Φ _n] ^tfor phase error coefficient,

for complex matrix corresponding to phase error to be estimated; The angle of arrival of each independent source is obtained by its MUSIC function spectrum peak

Step 4.2, the DOA in obtaining step 4.1 estimates the estimation of laggard line phase error Φ, adopts the single order form of gauss-newton method to carry out local optimum search, based on following formula:

Wherein,

B=[B(1) ^T?B(2) ^T?…?B(M) ^T] ^T

B(i)=-U(θ _i) ^Hdiag(jΦa(θ _i))

Z=[Z(1) ^T?Z(2) ^T?…?Z(M) ^T] ^T；

Z(i)=U(θ _i) ^HΦa(θ _i)

Wherein, Re is for getting real part computing, U (θ _i) be the noise vector space of single information source m, θ _ibe the angle of arrival of i information source, a (θ _i) be i the steering vector that information source is corresponding, Φ=[Φ ₁Φ ₂Φ _n] ^tfor phase error coefficient; By the gradient direction of determining with above formula, obtain current optimum angle estimated value Φ '=Φ+μ r (Φ), wherein step size mu is determined by linear search;

Step 4.3, in step 4.1, under 4.2 alternative manner, upgrades current cost function value, based on following formula:

$F\left({\mathrm{\&Phi;}}^{\&prime;}\right)=\underset{i}{\mathrm{\&Sigma;}}{\left|\right|a^H\left({\mathrm{\&theta;}}_i\right)C^{H}U\left({\mathrm{\&theta;}}_i\right)\left|\right|}^2$

Wherein, U (θ _i) be the noise vector space of single information source m, θ _ibe the angle of arrival of i information source, a (θ _i) be i the steering vector that information source is corresponding,

for complex matrix corresponding to phase error to be estimated)

When cost function value restrains, iterative search finishes.

Method for correcting passive channels at above-mentioned a kind of small array, also comprise that one is calibrated the strong echo of single information source by the channel amplitude gain coefficient of estimating and phase error estimation and phase error value, by DOA spectrum snr value size after calibration, certainly adjudicate the step of the reliability of phase alignment, concrete grammar is: after having obtained phase error estimation and phase error value, single information source echo data is carried out to passage amplitude and phase correction, and calculate the MUSIC spectrum signal to noise ratio (S/N ratio) of each information source, by the reliability of its snr value size judgement calibration of amplitude and phase.Be specially: the information source number that MUSIC spectrum snr value is greater than 8dB accounts for 70% when above, judges that calibration is effective; Otherwise think that this result can not be used when time alignment error.

Therefore, tool of the present invention has the following advantages: not needing the existence of auxiliary source completely, is the real method for correcting passive channels of a class; Each high strength echo information source number accuracy of judgement is reliable, affected by abnormal electrical magnetic environment little; To the quality of data of marine echo, require lower, minimumly can, under the reliable high strength marine echo of 10 left and right, obtain high-precision amplitude phase error estimated value, and algorithm be sane, this point makes it more can adapt to the change of radar operating environment, for a long time free of discontinuities work; The present invention, gets at sieve the method based on Array Model adopting aspect single information source marine echo and can obtain reliable and stable single information source echo without constraint the design of planar array; The present invention has first adopted the fast algorithm of lower accuracy to obtain and has estimated phase error in the method for phase error estimation and phase error, re-uses the phase error that High Convergent Precision algorithm obtains degree of precision.Algorithm structure is succinct, and calculated amount is little, real-time, and correction result is accurately sane.

Accompanying drawing explanation

Fig. 1 is high-frequency ground wave radar fundamental diagram.

Fig. 2 is high-frequency ground wave radar array received signal model figure.

Fig. 3 is phase error estimation algorithm process flow diagram.

Fig. 4 is the MUSIC Estimation of Spatial Spectrum curve map that adopts method calibration of the present invention front and back.

Embodiment

Below by embodiment, and by reference to the accompanying drawings, technical scheme of the present invention is described in further detail.In figure, wherein 1 for radio frequency amplifies, and 2 is local oscillation signal, and 3 is receiving front-end, and 4 is frequency mixer, and 5 is digital receiver, and 6 for separating distance element FFT, and 7 is the sliding window FFT of secondary.

Embodiment:

Below in conjunction with accompanying drawing and embodiment, the present invention is done to more detailed explanation.

Step 1, the second-order statistics based on Array Model, the eigenwert of decomposing according to strong echo character size is determined reliable and stable single information source echo; Concrete grammar is: by strong signal to noise ratio (S/N ratio) echo structure array received autocorrelation matrix, carry out by eigenwert S relation, determining single information source echo after feature decomposition, work as S ₁/ S ₂>=S _i/ S _i+1i=2 ... during N, be defined as single information source echo.

1. signal model.

High-frequency ground wave radar adopts linear frequency modulation to interrupt continuous wave (FMICW) system.Under this waveform system, marine echo enters after receiver, through lower mixing, and digital received, the distance element that can realize marine echo data with FFT is separated, then can obtain distance-Doppler's two dimension echo spectrum (as shown in Figure 1) by secondary FFT.

Necessarily in short-term in the marine echo of coherent accumulation, the sliding window FFT by certain overlapping step-length can obtain in distance-Doppler frequency separated, the fast beat of data of a large amount of and sane marine echo.As shown in Figure 2, the array element i position coordinates of establishing planar array is (x _i, y _i) (i=1 ... N), the i that receives on array when t snap single information source echo can be with being described as drag:

X _t=GC（AS _t+N _t）

G=diag(g ₁?g ₂?…?g _N)

$C=\mathrm{diag}\left(\begin{array}{cccc}{e}^{{\mathrm{j\&Phi;}}_1}& e^{{\mathrm{j\&Phi;}}_2}& ...& e^{j{\mathrm{\&Phi;}}_N}\end{array}\right)$

$A={\left[\begin{array}{cccc}{e}^{j\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}(x_1\sin {\mathrm{\&theta;}}_i+{\mathrm{<figure-callout\; id="1"\; label="y"\; filenames="131219143925.png,131219143928.png"\; state="\{\{state\}\}">y</figure-callout>}}_1{\cos \mathrm{\&theta;}}_i)}& e^{j\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}(x_2{\sin \mathrm{\&theta;}}_i+y_2{\cos \mathrm{\&theta;}}_i)}& ...& e^{j\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}(x_N{\sin \mathrm{\&theta;}}_i+y_N{\cos \mathrm{\&theta;}}_i)}\end{array}\right]}^T$

G is amplitude gain matrix of coefficients, and C is phase error matrix, and diag represents vectorial diagonalization computing, and A is the steering vector of echo i correspondence on array, S _tfor the original echo that array received arrives, N _tfor with the incoherent additive noise of signal.

2. the single information source reflection pickup based on Estimation of Spatial Spectrum theory.

The known array received signal of second-order statistics by Array Model has following form through feature decomposition

$R_x=X_t{X}_t^H={\mathrm{USU}}^H$

S=diag (λ wherein ₁λ ₂λ _n), λ ₁> λ ₂> ... λ _m> λ _m+1=λ _m+2=...=λ _n, M is irrelevant information source number, and N is element number of array, known through feature decomposition by above formula, and signal subspace and noise subspace are determined, λ _i, i=1 ... M is each irrelevant information source power, λ _j(j=M+1 ... N) be noise power.Definition: for each the separated marine echo in frequency of distance spectrum, all can obtain this sequential value.Guaranteeing under certain state of signal-to-noise, for single information source data, max[Y _i]=Y ₁, and its value will be larger, reflect the now echo quality condition of single information source, and this value is larger, shows that this echo is reliable and stable in the time at coherent accumulation.And under data from multiple sources, due to the power ratio of the power ratio between each information source much smaller than signal and noise, max[Y now _i] ≠ Y ₁, can intuitively determine fast single information source echoed signal thus.

Step 2, utilize the fast beat of data of each passage list information source echo energy and ratio estimating channel amplitude gain coefficient, and according to the channel amplitude gain coefficient of estimating, the fast beat of data of each passage list information source echo is carried out to amplitude self-correcting; The fast beat of data of described each passage list information source echo refers to the fast beat of data in respective channel in the array received autocorrelation matrix of step 1 time all single information source echoes.

1. channel amplitude is proofreaied and correct.

The amplitude gain coefficient of each array element can be directly by

${\hat{g}}_i=\sqrt{\underset{l}{\mathrm{\&Sigma;}}\underset{t}{\mathrm{\&Sigma;}}{\left|X_i(t,l)\right|}^2/\underset{l}{\mathrm{\&Sigma;}}\underset{t}{\mathrm{\&Sigma;}}{\left|X_1(t,l)\right|}^2}$

Calculate, take each energy taking soon of single information source echo that each passage receives and be with reference to calculating amplitude gain coefficient.

2. channel phases is proofreaied and correct.

When array is after amplitude correction, the single information source echo model receiving on each array is reduced to

X _t=C（AS _t+N _t）

Adopt PSO algorithm under high-dimensional to carry out phase error initial value and estimate, by the sue for peace optimal estimation of MUSIC of this estimation results, obtain phase error estimation and phase error accurately.Global optimization obtains effective phase error and estimates real-time and the reliability that can guarantee passive correction with degree of precision phase error estimation algorithm combination technology.

Step 3, utilizes in step 2 the fast beat of data of single information source marine echo after amplitude calibration and known array information by by the major function of asking of linearity

pSO global optimization search estimated initial phases error; Wherein, θ=[θ ₁θ ₂θ _m] ^tfor the angle of arrival of all information sources, Φ=[Φ ₁Φ ₂Φ _n] ^tfor phase error coefficient,

for the steering vector matrix of all information sources,

for each information source power of estimating, λ is wavelength, (x _n, y _n) be the coordinate of n antenna;

1. the global search phase error pre-estimating technology based on particle cluster algorithm

In this step, to estimating the phase error of acquisition, do not do strict accuracy requirement, therefore on calculated amount and algorithm complex, can select comparatively simple and quick maximum likelihood method cost function

only need to obtain phase error as a result of, use 100 particle left and right in 200 global optimization number of times, to carry out overall PSO optimum search.Although population now does not enough travel through whole search field of definition, under the traversal mode of global optimization, can effectively avoid being absorbed in local optimum.

Basic step is as shown in Figure 3:

1) phase error to be estimated of each particle of random initializtion and angle of arrival value P _i=[θ, Φ] and initial velocity i is population;

2) by

calculate each particle cost function value F _i;

3) by current particle information, obtained the traversal optimal value O of each particle _iwith particle optimum solution

global optimum O _dwith globally optimal solution

4) calculate each particle traversal search speed

each particle is carried out to the renewal of solve for parameter,

$v_i^{k+1}={\mathrm{wv}}_i^k+c_1{r}_1(g_i^k-P_i^k)+c_2{r}_2(g_d^k-P_i^k)$

$P_i^{k+1}=P_i^k+v_i^{k+1}$

Wherein, w is primary particle speed weight, c ₁, c ₂be respectively Local Search speed and global search speed weight, r ₁, r ₂it is the random number between 0 to 1.

In particle position traversal, for guaranteeing the ergodicity of global search, the size of restriction particle rapidity is no more than certain threshold value.

?

for the particle rapidity that is greater than this value, adopt following formula method restriction particle rapidity: $v_i^{k+1}=\frac{\mathrm{\&kappa;}}{\left|v_i^{k+1}\right|}{v}_i^{k+1}.$

5) k=k+1, if k<200 goes to step 2, otherwise with globally optimal solution g _dfor output, finish traversal.

At above PSO algorithm, carry out in fast search calculating, the information only having used between each particle and particle travels through calculating, needn't carry out the differentiate of the multidimensional function of gradient class optimizing algorithm, and the lengthy and tedious calculating from local optimum to global optimum's strategy, algorithm calculated amount is reduced greatly.Just because of the gradient information of not considering cost function, make its globally optimal solution cannot guarantee strict convergence, but its result can meet the requirement to this step in the present invention.From the check of computer simulation emulation and measured data, all confirmed that phase error coefficient that the method obtains and actual value gap are in 20 degree, can be follow-up degree of precision and solve effective initial value is provided.

Although the present invention provides, be to take the global optimization searching method that PSO algorithm is example, but from concerning the demand of this step, adopt other quick global optimizing algorithm, as simulated annealing, evolutionary computation, Chaos Search, the methods such as random sampling, strictly not limiting under the design of global solution precision, all can be used for estimating of phase error.This point makes the present invention can have great dirigibility in actual applications.

Step 4, is searched for by the sue for peace iteration optimization of MUSIC cost function of the initial phase error coefficient of estimating in step 3, realizes phase place high-precision correction; Specifically: utilize in step 2 the fast beat of data of single information source echo of amplitude correction, the initial phase error that the global optimization search obtaining in known array position information and step 3 obtains, the optimization phase error coefficient carrying out based on summation MUSIC function estimates, optimized results is based on formula:

by this optimization, realize phase place self-correcting, wherein U (θ _i) be the noise vector space of single information source i; Wherein, θ _ibe the angle of arrival of i information source, a (θ _i) be i the steering vector that information source is corresponding, Φ=[Φ ₁Φ ₂Φ _n] ^tfor phase error coefficient,

for complex matrix corresponding to phase error to be estimated.

Based on summation MUSIC phase error estimation algorithm

Another core algorithm of the present invention-based on summation MUSIC phase error estimation algorithm, can by the initial phase estimation results obtaining before, be the good accuracy solution of information acquisition phase error coefficient.Its actual algorithm pattern that solves middle employing iteration convergence carries out phase error coefficient and estimates that the Joint iteration of estimating with the angle of arrival solves.

Algorithm steps is as shown in Figure 3:

1) calculate the angle of arrival and cost function value;

Under priori phase error, carry out the estimation of DOA.

$P_m\left(\mathrm{\&theta;}\right)=\frac{1}{{\left|\right|U\left(m\right)\mathrm{\&Phi;a}\left(\mathrm{\&theta;}\right)\left|\right|}^2}$

The DOA of each independent source is obtained by its MUSIC function spectrum peak

2) local optimum of phase error solves, and adopts gauss-newton method, obtains the optimal estimation solution of phase error;

Have:

Wherein,

B=[B(1) ^T?B(2) ^T?…?B(M) ^T] ^T

B(m)=-U(m) ^Hdiag(jΦa(θ _m))

Z=[Z(1) ^T?Z(2) ^T?…?Z(M) ^T] ^T

Z(m)=U(m) ^HΦa(θ _m)

The gradient direction of being determined by above formula obtains current optimal estimation value

wherein step size mu is determined by linear search.

3) angle of arrival calculates with cost function value and upgrades, and the angle of arrival adopts the method in step 1) to realize, and cost function is calculated by following formula:

$F\left({\mathrm{\&Phi;}}^{\&prime;}\right)=\underset{i}{\mathrm{\&Sigma;}}{\left|\right|a^H\left({\mathrm{\&theta;}}_i\right)C^{H}U\left({\mathrm{\&theta;}}_i\right)\left|\right|}^2$

4) as cost function value, convergence finishes, otherwise goes to step 2;

Step 5, calibration steps: by the channel amplitude gain coefficient of estimating and phase error estimation and phase error value, the strong echo of single information source is calibrated, certainly adjudicated the reliability of phase alignment by DOA spectrum snr value size after calibration.

In above algorithmic procedure, carry out the iteration of phase error and the angle of arrival and upgrade calculating until cost function value convergence is minimum.Due to the orthogonal property of MUSIC function, well estimating under acquisition of initial phase error, this algorithm will strictly be restrained in globally optimal solution position as shown in Figure 4, and precision is also apparently higher than the optimum solution under linear class cost function.

One) passage amplitude and phase correction judgement.

Passive calibration methods, when practical engineering application, itself can guarantee that algorithm convergence drops to global minimum position, but the confirmation that whether has reached unique this point of Global optimal solution cannot be undertaken by real-time software, and its process of argumentation will expend a large amount of computational resources.By certain calibration thresholding index, can adjudicate fast the validity of calibration result.This point is that this algorithm can sound and stable operation and avoid the committed step of alignment error.

After having obtained channel amplitude gain coefficient and channel phases error estimate, single information source echo data is carried out to channel correcting,, in raw data, the data of each passage, divided by corresponding amplitude gain value, deduct phase error estimation and phase error value in phase place.

${\tilde{X}}_i=\frac{e^{-j{\mathrm{\&phi;}}_i}}{g_i}{X}_i$

The DOA that is carried out single information source echo by the data after amplitude and phase correction estimates, the MUSIC function snr value that the DOA of wherein usining estimates is as output:

$\mathrm{SNR}\left({\mathrm{\&theta;}}_i\right)=10\log \left(\frac{\max \left[P\left(\mathrm{\&theta;}\right)\right]}{\min \left[P\left(\mathrm{\&theta;}\right)\right]}\right)$

In the MUSIC of all single information source echoes snr value, the value number that is greater than 8dB accounts for 70% when above of information source sum, judges that this calibration result is effectively reliable, and thinks that its calibration result is insincere during less than 70%.

Specific embodiment described herein is only to the explanation for example of the present invention's spirit.Those skilled in the art can make various modifications or supplement or adopt similar mode to substitute described specific embodiment, but can't depart from spirit of the present invention or surmount the defined scope of appended claims.

Claims (5)

1. a method for correcting passive channels for small array, is characterized in that, comprises the following steps:

Step 1, the second-order statistics based on Array Model, the eigenwert of decomposing according to strong echo character size is determined reliable and stable single information source echo; Concrete grammar is: by strong signal to noise ratio (S/N ratio) echo structure array received autocorrelation matrix, carry out by eigenwert S relation, determining single information source echo after feature decomposition, work as S ₁/ S ₂>=S _i/ S _i+1i=2 ... during N, be defined as single information source echo;

Described array received autocorrelation matrix is in the four-dimensional echo data that conversion obtains through secondary FFT, distance element and the array snap autocorrelation matrix that in Doppler frequency, independently strong signal to noise ratio (S/N ratio) marine echo is corresponding;

Step 2, utilize the fast beat of data of each passage list information source echo energy and ratio estimating channel amplitude gain coefficient, and according to the channel amplitude gain coefficient of estimating, the fast beat of data of each passage list information source echo is carried out to amplitude self-correcting; The fast beat of data of described each passage list information source echo refers to the fast beat of data in respective channel in the array received autocorrelation matrix of step 1 time all single information source echoes;

Step 3, utilizes in step 2 the fast beat of data of single information source marine echo after amplitude calibration and known array information by by the major function of asking of linearity

pSO global optimization search estimated initial phases error; Wherein, θ=[θ ₁θ ₂θ _m] ^tfor the angle of arrival of all information sources, Φ=[Φ ₁Φ ₂Φ _n] ^tfor phase error coefficient,

for the steering vector matrix of all information sources,

for each information source power of estimating, λ is wavelength, (x _n, y _n) be the coordinate of n antenna;

Step 4, is searched for by the sue for peace iteration optimization of MUSIC cost function of the initial phase error coefficient of estimating in step 3, realizes phase place high-precision correction; Specifically: utilize in step 2 the fast beat of data of single information source echo of amplitude correction, the initial phase error that the global optimization search obtaining in known array position information and step 3 obtains, the optimization phase error coefficient carrying out based on summation MUSIC function estimates, optimized results is based on formula: by this optimization, realize phase place self-correcting, wherein U (θ _i) be the noise vector space of single information source i; Wherein, θ _ibe the angle of arrival of i information source, a (θ _i) be i the steering vector that information source is corresponding, Φ=[Φ ₁Φ ₂Φ _n] ^tfor phase error coefficient, for complex matrix corresponding to phase error to be estimated.

2. the method for correcting passive channels of a kind of small array according to claim 1, is characterized in that, in described step 2, utilizes the fast beat of data of single information source echo receive to carry out the estimation of channel amplitude gain coefficient, the amplitude gain coefficient of each array element directly by

estimate, realize amplitude self-correcting.

3. according to claim 1, it is characterized in that, the concrete grammar of described step 3 comprises following sub-step:

The phase error to be estimated of step 3.1, each particle of random initializtion and angle of arrival value P _i=[θ, Φ] and initial velocity

iterations k=0, i is population;

Step 3.2, by calculate each particle cost function value F _i;

Step 3.3, by current particle information, obtained the traversal optimal value O of each particle _iwith particle optimum solution

global optimum O _dwith globally optimal solution

Step 3.4, calculate each particle traversal search speed

each particle is carried out to the renewal of solve for parameter, based on following formula:

$v_i^{k+1}={\mathrm{wv}}_i^k+c_1{r}_1(g_i^k-P_i^k)+c_2{r}_2(g_d^k-P_i^k);$

$P_i^{k+1}=P_i^k+v_i^{k+1};$

Wherein, w is primary particle speed weight, c ₁, c ₂be respectively Local Search speed and global search speed weight, r ₁, r ₂it is the random number between 0 to 1;

In particle position traversal, for guaranteeing the ergodicity of global search, the size of restriction particle rapidity is no more than the threshold value of setting;

If κ meets

for the particle rapidity that is greater than this value, adopt following formula method restriction particle rapidity: $v_i^{k+1}=\frac{\mathrm{\&kappa;}}{\left|v_i^{k+1}\right|}{v}_i^{k+1}$

Step 3.5, k=k+1, if k<200 goes to step 2, otherwise with globally optimal solution g _dfor output, finish traversal.

4. the method for correcting passive channels of a kind of small array according to claim 1, is characterized in that, in described step 4, based on summation MUSIC cost function

optimization phase error estimation method adopts iteration convergence method, and its step is as follows:

Step 4.1 first, is carried out the estimation of DOA under current phase error, based on following formula:

$P_m\left(\mathrm{\&theta;}\right)=\frac{1}{{\left|\right|U^H\left({\mathrm{\&theta;}}_i\right)\mathrm{Ca}\left(\mathrm{\&theta;}\right)\left|\right|}^2};$

Wherein, U (θ _i) be the noise vector space of single information source i, θ _ibe the angle of arrival of i information source, a (θ _i) be i the steering vector that information source is corresponding, Φ=[Φ ₁Φ ₂Φ _n] ^tfor phase error coefficient, for complex matrix corresponding to phase error to be estimated; The angle of arrival of each independent source is obtained by its MUSIC function spectrum peak

Step 4.2, the DOA in obtaining step 4.1 estimates the estimation of laggard line phase error Φ, adopts the single order form of gauss-newton method to carry out local optimum search, based on following formula:

Wherein,

B=[B(1) ^T?B(2) ^T?…?B(M) ^T] ^T

B(i)=-U(θ _i) ^Hdiag(jΦa(θ _i))

Z=[Z(1) ^T?Z(2) ^T?…?Z(M) ^T] ^T；

Z(i)=U(θ _i) ^HΦa(θ _i)

Wherein, Re is for getting real part computing, U (θ _i) be the noise vector space of single information source m, θ _ibe the angle of arrival of i information source, a (θ _i) be i the steering vector that information source is corresponding, Φ=[Φ ₁Φ ₂Φ _n] ^tfor phase error coefficient; By the gradient direction of determining with above formula, obtain current optimum angle estimated value Φ '=Φ+μ r (Φ), wherein step size mu is determined by linear search;

Step 4.3, in step 4.1, under 4.2 alternative manner, upgrades current cost function value, based on following formula:

$F\left({\mathrm{\&Phi;}}^{\&prime;}\right)=\underset{i}{\mathrm{\&Sigma;}}{\left|\right|a^H\left({\mathrm{\&theta;}}_i\right)C^{H}U\left({\mathrm{\&theta;}}_i\right)\left|\right|}^2$

Wherein, U (θ _i) be the noise vector space of single information source m, θ _ibe the angle of arrival of i information source, a (θ _i) be i the steering vector that information source is corresponding,

for complex matrix corresponding to phase error to be estimated;

When cost function value restrains, iterative search finishes.

5. the method for correcting passive channels of a kind of small array according to claim 1, it is characterized in that, also comprise that one is calibrated the strong echo of single information source by the channel amplitude gain coefficient of estimating and phase error estimation and phase error value, by DOA spectrum snr value size after calibration, certainly adjudicate the step of the reliability of phase alignment, concrete grammar is: after having obtained phase error estimation and phase error value, single information source echo data is carried out to passage amplitude and phase correction, and calculate the MUSIC spectrum signal to noise ratio (S/N ratio) of each information source, by the reliability of its snr value size judgement calibration of amplitude and phase; Be specially: the information source number that MUSIC spectrum snr value is greater than 8dB accounts for 70% when above, judges that calibration is effective; Otherwise think that this result can not be used when time alignment error.

Images