CN1321331C - Method for array channel calibration by utilizing ocean echo wave - Google Patents

Abstract

The present invention relates to a method for using sea echo waves to carry out array passage correction. The method is characterized in that an antenna array comprising at least two translation invariant array element pairs is used for receiving sea echo waves; a single arrival direction frequency point is detected in the sea echo waves through the statistical analysis to sea echo wave two-dimensional spectra; the statistical average is carried out to the echo wave spectrum amplitude corresponding to the single arrival direction frequency point, and the amplitude characteristics of passages are estimated; amplitude correction can be realized; according to the echo wave spectrum phase positions and the known reflecting signal source echo wave information such as islands, lighthouses, drilling platforms, etc., the phase characteristics of the passages are estimated by an MUSIC(multiple signal classification) algorithm, statistical average is carried out to the multiple results, and phase correction can be realized.

Description

Method for correcting array channel by using ocean echo

Technical Field

The invention relates to a method for correcting array channels of a high-frequency ground wave radar by using ocean echoes.

Background

The high-frequency ground wave radar is a novel radar for detecting long-distance targets (ships, low-altitude airplanes, cruise missiles, ocean surfaces and the like) by utilizing diffraction of high-frequency (3-30 MHz) electromagnetic waves along the surface of the earth, has the outstanding advantages of long detection distance, anti-stealth, anti-radiation missiles, low-altitude defense resistance, capability of detecting ocean surface states and the like (compared with the conventional radar), and has great development potential.

The high-frequency ground wave radar adopts a phased array antenna, and utilizes digital beam forming and space spectrum estimation technologies to carry out beam scanning and DOA (direction of arrival) estimation on a marine target, so that the high-frequency ground wave radar can effectively detect the sea surface states of wind, waves, currents and the like and moving targets of airplanes, ships and the like.

Due to the inconsistency of hardware and the influence of multiple factors such as mutual coupling effect between antennas, the amplitude characteristics and the phase characteristics of all receiving channels forming the radar array are different in practice, so that the amplitude and the phase changes of echo signals passing through different channels are inconsistent. The inconsistency of the channel characteristics causes the error increase and even complete failure of beam scanning and DOA estimation, and is one of the key problems influencing the detection performance of the high-frequency ground wave radar. In order to ensure that the radar can work effectively, measures must be taken to limit the inconsistency between array channels within a certain range: on one hand, the consistency of each channel is ensured as much as possible during manufacturing through proper measures (such as component screening); on the other hand, the difference in channel characteristics can be further reduced by correction.

The existing array channel correction methods can be divided into passive correction and active correction.

In the passive correction method, a signal source with an accurately known direction is not needed, the amplitude and phase errors of each channel are calculated by directly utilizing the received measured data and some a priori knowledge (such as an array form), and then compensation correction is carried out. Some passive correction methods also enable joint estimation of signal direction of arrival and channel error. The method is described in detail in the book "estimation of spatial spectrum and application thereof" (1997 of university of Chinese science and technology publishers) edited by Liude tree, Luo Jing Qing, etc.

In the active correction method, a known signal source is placed in an open field far enough away from the array, signals are transmitted, the amplitude and the phase of output signals of each receiving channel are measured, and phase differences caused by the spatial position of the array are deducted, so that channel error information can be obtained. The correction method is simple in principle and good in effect, and is widely applied in practice. In high frequency ground wave radar, the signal source for correction is a transponder placed far in front of the array for amplifying the received radar signal and transmitting it back, and the array response of the response signal is compared with the ideal array response to obtain the estimation of the channel amplitude and phase error.

In the existing channel correction method, passive correction needs multiple complex iterative operations, the calculated amount is large, the real-time requirement cannot be met necessarily, and the passive correction is likely to converge on a local minimum value rather than a global minimum value, so that an error result is obtained. Although the active correction using a transponder has a simple principle and a good effect, the application in practical situations is limited: the transponder is difficult to place and maintain on the sea and difficult to work for a long time; it is difficult to eliminate the influence of multipath effect caused by islands, ships, and the like.

For better illustration of the present invention, the operation of the high frequency ground wave radar will be described. The High Frequency Ground wave radar adopts FMCW (Frequency-modulated continuous wave) system, and under the condition of co-station transmission and receiving, the FMICW (Frequency-modulated interrupted continuous wave) system is formed by interrupting the system to solve the problem of isolation of transmission and receiving, and the paper published by Rafaat Khan et al entitled "High Frequency Ground wave radar Target Detection and Tracking" (Target Detection and Tracking With a High Frequency group wave radar IEEE Journal of scientific Engineering, 1994, 19 (4): 540-548) has detailed description.

The radar signal generator generates an FMCW local oscillator signal, which may be expressed as

f₀For the radar signal carrier frequency, α is the sweep rate, T is the sweep period, and A and * are the signal amplitude and initial phase, respectively. The local oscillator signal becomes a transmitting signal after being interrupted by the gate control pulse

S_T(t)＝S(t)g(t) (2)

The gating pulse g (t) may be expressed as

$g\left(t\right)=\underset{p=0}{\overset{p-1}{\mathrm{\&Sigma;}}}\mathrm{rect}\left[\frac{t-\mathrm{pq}-\frac{{\mathrm{<figure-callout\; id="0"\; label="T"\; filenames="C0312823800111.png"\; state="\{\{state\}\}">T</figure-callout>}}_0}{2}}{T_0}\right]---\left(3\right)$

P is the number of gated pulses within the sweep period T, T₀And q are the pulse width and period, respectively.Representative width of T₀A rectangular pulse centered at the origin.

If the target moves at a radial velocity v (positive away from the radar) at a distance r, the time delay of the target's reflected signal received by the radar is

$\mathrm{\&tau;}=\frac{2(r+\mathrm{vt})}{c}---\left(4\right)$

Where c is the speed of light. The radar receives signals of

S_R(t)＝K_RS_T(t-τ) (5)

K_RIs the propagation attenuation factor. After the received signal and the local oscillator signal are mixed, the baseband signal is obtained by low-pass filtering demodulation

$S_I\left(t\right)=\mathrm{lowpass}\{S\left(t\right)\&CenterDot;S_R\left(t\right)\}$

$=A_I\cos \left(2\mathrm{\&pi;}(\mathrm{\&alpha;\&tau;t}-{\mathrm{<figure-callout\; id="0"\; label="f"\; filenames="C0312823800111.png"\; state="\{\{state\}\}">f</figure-callout>}}_0\mathrm{\&tau;}-\frac{{\mathrm{\&alpha;\&tau;}}^2}{2})\right)---\left(6\right)$

A₁Is the baseband signal amplitude. The low-pass filtering removes the pulse modulation and makes the baseband signal continuous wave, so the term gating pulse g (t) is not included in equation (6). After substituting equation (4) into equation (6), the expansion can be achieved by omitting a small amount of phase

$S_I\left(t\right)\&ap;A_I\cos \left(2\mathrm{\&pi;}(2\left(\frac{\mathrm{\&alpha;r}-f_{0}v}{c}\right)t+\frac{2\mathrm{\&alpha;v}}{c}{t}^2-\frac{2f_{0}r}{c}-\frac{2\mathrm{\&alpha;}{r}^2}{c^2})\right)=A_I\cos \left({\mathrm{\&phi;}}_{\mathrm{\&tau;}}\right)---\left(7\right)$

Instantaneous frequency of baseband signal is

$f_{\mathrm{\&tau;}}\left(t\right)=\frac{1}{2\mathrm{\&pi;}}\frac{d{\mathrm{\&phi;}}_{\mathrm{\&tau;}}}{\mathrm{dt}}=\frac{2\mathrm{\&alpha;r}}{c}-\frac{2f_{0}v}{c}+\frac{4\mathrm{\&alpha;vt}}{c}---\left(8\right)$

The first term is caused by the target distance, and the second and third terms are caused by the target radial velocity. In high-frequency radars

$\left|\frac{2\mathrm{\&alpha;r}}{c}\right|>>|-\frac{2f_{0}v}{c}+\frac{4\mathrm{\&alpha;vt}}{c}|,$ Thus is provided with $f_{\mathrm{\&tau;}}\left(t\right)\&ap;\frac{2\mathrm{\&alpha;r}}{c}.$

The above analysis shows that performing an FFT on the baseband signal after a/D conversion can obtain a discrete spectrum corresponding to the distance, this time the FFT is called distance conversion, and the obtained distance spectrum is

$R_I\left[m\right]=\mathrm{FFT}\left\{S_I\left(t\right)\right\}$

${=A}_I\&CenterDot;\mathrm{FFT}\left\{\cos \left(2\mathrm{\&pi;}(\mathrm{\&alpha;\&tau;t}-{\mathrm{<figure-callout\; id="0"\; label="f"\; filenames="C0312823800111.png"\; state="\{\{state\}\}">f</figure-callout>}}_0\mathrm{\&tau;}-\frac{{\mathrm{\&alpha;\&tau;}}^2}{2})\right)\right\}$

$=A_I\&CenterDot;R\left[m\right]---\left(9\right)$

Using the distance spectrum obtained in a sweep period as a line, i_maxThe distance spectrum obtained from one sweep period can form one_max×m_maxMatrix array

m_maxIs the farthest distance element number.

Now, the phase of each line in R is analyzed according to the change rule of the sweep period ordinal number (line ordinal number) l. During the first sweep period, the target distance is

r_l＝r+v(l-1)T (11)

The phase of the baseband signal of the ith sweep period is

${\mathrm{\&phi;}}_{\mathrm{l\&tau;}}=2\mathrm{\&pi;}(2\left(\frac{{\mathrm{\&alpha;r}}_l-f_{0}v}{c}\right)t+\frac{2\mathrm{\&alpha;v}}{c}{t}^2-\frac{2f_0{r}_l}{c}\text{-}\frac{{2\mathrm{\&alpha;r}}_l^2}{c^2})---\left(12\right)$

Within 100 sweep periods, i.e. /)_maxWhen the phase difference is less than or equal to 100, some small phase terms are omitted, and the phase difference of the baseband signals of two continuous sweep periods is

$\mathrm{\&Delta;\&phi;}\&ap;2\mathrm{\&pi;}(-\frac{{2f}_{0}v}{c})T---\left(13\right)$

According to this approximation, the l-th row differs from the 1 st row in R by only one phase factor e^{-j2π(l-1)-(2f0v/c)}Can be approximately expressed as

The FFT performed once more for each column of equation (14) can obtain a doppler spectrum corresponding to the velocity, and this FFT is called doppler transform. Therefore, after sampling a plurality of frequency sweep period baseband signals, performing FFT processing twice to obtain a discrete two-dimensional echo spectrum

Z(m，n)＝FFT{FFT{S_I(t)}} (15)

Where m is the discrete frequency in the distance dimension and n is the discrete frequency in the velocity (doppler frequency) dimension. The frequency of the peak of the target echo in the distance dimension is f _τ2 α r/c, the frequency of peaks in the velocity dimension is f_v＝-2f₀v/c, carrying out peak detection on the two-dimensional echo spectrum to obtain the target distance and speed.

From the above discussion of the working principle of the high-frequency ground wave radar, it can be known that the two FFTs actually separate target echo signals with different distances and speeds, so that the target echo signals correspond to different frequency points in the two-dimensional echo spectrum. The high-frequency ground wave radar receives a large amount of ocean echo signals with strong energy, the ocean echo signals are dispersed on a plurality of frequency points of a two-dimensional echo spectrum, and a part of the ocean echo signals are only in a single arrival direction. Through the statistical analysis of the two-dimensional echo spectrum output of the specific array, the frequency points in the single arrival direction can be detected, then the amplitude-phase characteristics of each channel are estimated according to the known echo information of the reflection source, and the statistical averaging is carried out to improve the precision.

Disclosure of Invention

Aiming at the defects of the existing method, the invention aims to provide a real-time, accurate, cheap, more stable and reliable array channel correction method by utilizing ocean echo information received by a high-frequency ground wave radar so as to reduce channel amplitude-phase errors and improve the performance of a radar system.

In order to achieve the above object, the array correction method adopted by the present invention is: receiving ocean echoes by an antenna array comprising at least two array element pairs with unchanged translation, and detecting single arrival direction frequency points in the ocean echoes by statistical analysis of a two-dimensional spectrum of the ocean echoes; carrying out statistical averaging on the echo spectrum amplitude corresponding to the frequency points in the single arrival direction, estimating the amplitude characteristic of each channel, and realizing amplitude correction; according to the echo spectrum phase corresponding to the frequency point in the single arrival direction and the echo information of the known reflection signal source, such as an island, a lighthouse, a drilling platform and the like, the phase characteristics of each channel are estimated by adopting a MUSIC (multiple signal classification) algorithm, and a plurality of results are counted and averaged to realize phase correction.

The invention has the advantages that: complex iterative operation is not adopted, the calculated amount is not large, and the requirement of channel correction instantaneity can be met; a statistical method is adopted for a large number of echo signals, so that the accuracy of channel correction is improved; by utilizing a large amount of continuous ocean echoes and known reflection source information, the problems of placement and maintenance of the transponder are avoided, so that the channel correction is cheaper and can be stably and reliably carried out for a long time; the MUSIC algorithm capable of multi-source DOA estimation is adopted, and the influence of multipath effect is eliminated.

The present invention will be described in more detail with reference to the accompanying drawings and examples.

Drawings

FIG. 1 is a working principle diagram of a high-frequency ground wave radar

FIG. 2 is a schematic diagram of a two-dimensional echo spectrum of a high-frequency ground wave radar

FIG. 3 is a diagram of a specific array for detecting frequency points in a single arrival direction in a two-dimensional echo spectrum

FIG. 4 is a schematic view of a uniform linear array

Detailed Description

The key point of the invention is to detect the frequency point of a single arrival direction in a two-dimensional echo spectrum, which requires that an antenna array contains a specific array form, as shown in fig. 3.

Suppose that the high-frequency ground wave radar has M antenna unitsIs marked as (x)_i，y_i) I is 1, 2, … M. The specific array for detecting the single-arrival-direction echo spectrum frequency point is composed of array elements 1-4, wherein 1 and 2 form an array element pair A₁3 and 4 form an array element pair A₂。A₁And A₂With translational invariance between, i.e. A₁After translation can be connected with A₂Completely coincide with (x)₂，y₂)＝(x₁+d，y₁)，(x₄，y₄)＝(x₃+d，y₃). Let the channel complex gain of array element i be g_ie^Jφi Taking array element 1 as the origin of coordinates, the two-dimensional echo spectrum of the array is

$Z_i(m,n)=g_i{e}^{\mathrm{j\&phi;i}}[\underset{k=1}{\overset{k}{\mathrm{\&Sigma;}}}{S}_k(m,n){\mathrm{<figure-callout\; id="2"\; label="e\; j"\; filenames="C0312823800111.png"\; state="\{\{state\}\}">e</figure-callout>}}^j+N_i(m,n)]---\left(16\right)$

Wherein S is_k(m, n) is the k-th echo signal received by the array element 1Number (arrival direction is theta)_k) Two-dimensional echo spectrum of (1)_iAnd (m, n) is a noise component in the two-dimensional echo spectrum of the array element i, lambda is a signal wavelength, and K is the number of arrival directions of echo signals. Order to ${\mathrm{\&eta;}}_1=\frac{Z_2(m,n)Z_3(m,n)}{Z_1(m,n)Z_4(m,n)},$ When K is 1 and N_iWhen (m, n) is 0, there are ${\mathrm{\&eta;}}_1=\frac{g_2{g}_3}{g_1{g}_4}{e}^{J({\mathrm{\&phi;}}_2+{\mathrm{\&phi;}}_3-{\mathrm{\&phi;}}_1-{\mathrm{\&phi;}}_4)},$ This indicates that

For a certain frequency point (m, n) in the two-dimensional echo spectrum, eta in the ideal case of only one arrival direction and no noise component₁Is a fixed quantity related only to the channel complex gain. Noise is inevitable in real systems, corresponding eta₁Is distributed atNear this fixed amount. Simple analysis shows that the number K of the arrival directions is more than or equal to 2, and eta corresponding to the frequency points₁Is a variation quantity related to the echo intensity and the arrival direction, and is in a dispersed state on the complex plane, and eta corresponding to the frequency point with the number of arrival directions K equal to 1₁Then the focus is near a certain point on the complex plane. Corresponding eta of all frequency points exceeding a certain signal-to-noise ratio threshold in the two-dimensional echo spectrum₁Marked on complex plane, there is one and only one region where η appears₁Phenomenon of aggregation in which most of η₁The frequency point corresponding to the value has only one arrival direction.

Order to ${\mathrm{\&eta;}}_2=\frac{Z_2(m,n){\mathrm{<figure-callout\; id="1"\; label="Z"\; filenames="C0312823800111.png"\; state="\{\{state\}\}">Z</figure-callout>}}_1^{*}(m,n)}{Z_4(m,n)Z_3^{*}(m,n)},$ Eta. according to the same analysis as above₂The concentration region also appears on the complex plane, where most of η₂The frequency point corresponding to the value has only one arrival direction. Eta corresponding to frequency point due to single arrival direction₁And η₂Are all concentrated in respective concentration areas, and the eta corresponding to the frequency points of multiple arrival directions₁And η₂Is distributed dispersedly with little possibility of falling into a gathering zone, so that eta can be used₁And η₂Whether the frequency points fall into the aggregation area simultaneously is used as a criterion for detecting the frequency points in the single arrival direction.

From the above analysis, it can be known that the single arrival direction frequency point in the two-dimensional echo spectrum can be detected as long as the antenna array contains at least two array element pairs with unchanged translation. The arrays to which the channel correction method of the present invention is applied all contain this particular array format.

Assuming that the frequency point in the single arrival direction is (m ', n'), and K is 1, the two-dimensional echo spectrum output corresponding to the frequency point can be obtained by substituting equation (16)

$Z_i(m^{\&prime;},n^{\&prime;})=g_i{e}^{j{\mathrm{\&phi;}}_i}[S_1(m^{\&prime;},n^{\&prime;}){\mathrm{<figure-callout\; id="2"\; label="e\; j"\; filenames="C0312823800111.png"\; state="\{\{state\}\}">e</figure-callout>}}^j+N_i(m^{\&prime;},n^{\&prime;})]---\left(17\right)$

When N is present_iWhen (m ', n') is 0, there are

g_i＝|Z_i(m′，n′)|/|S₁(m′，n′)| (18)

Based on the receiving channel of array element 1, g₁＝1，|S₁(m′，n′)|＝|Z₁(m ', n') |, into the formula (18) to obtain

g_i＝|Z_i(m′，n′)|/|Z₁(m′，n′)| (19)

The actual two-dimensional echo spectrum is noisy and g is obtained from different single-arrival-direction frequency points_iThere will be some random fluctuations and statistical averaging can be performed to improve the estimation accuracy. Estimating channel amplitude gain g_iThen, the echo data of each channel is divided by g_iAmplitude correction can be achieved.

After amplitude correction, the channel amplitude gain g_iWhen N is 1, according to formula (17)_iWhen (m ', n') is 0, there is

$e^{J{\mathrm{\&phi;}}_i}=[Z_i(m^{\&prime;},n^{\&prime;})/S_1(m^{\&prime;},n^{\&prime;})]e^{-\mathrm{<figure-callout\; id="2"\; label="J"\; filenames="C0312823800111.png"\; state="\{\{state\}\}">J</figure-callout>}\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}\left(x_i\sin {\mathrm{\&theta;}}_1{y}_i\cos {\mathrm{\&theta;}}_1\right)}---\left(20\right)$

Using the receiving channel of the array element 1 as a reference and the position thereof as the origin of coordinates, i.e. $e^{j\mathrm{\&phi;}1}=1$ ，(x₁，y₁) When it is (0, 0), then there is S₁(m′，n′)＝Z₁(m ', n') into the formula (20) to obtain

$e^{J{\mathrm{\&phi;}}_i}=[Z_i(m^{\&prime;},n^{\&prime;})/Z_1(m^{\&prime;},n^{\&prime;})]e^{-\mathrm{<figure-callout\; id="2"\; label="J"\; filenames="C0312823800111.png"\; state="\{\{state\}\}">J</figure-callout>}\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}(X_i{\sin \mathrm{\&theta;}}_1+y_i\cos {\mathrm{\&theta;}}_1)}---\left(21\right)$

Writing formula (16) in matrix form

Z(m，n)＝GAS(m，n)+GN(m，n) (22)

Wherein Z (m, n) ═ Z₁(m，n)，Z₂(m，n)，…，Z_M(m，n，)]^T

$G=\mathrm{diag}(g_1{e}^{j{\mathrm{\&phi;}}_1},g_2{e}^{{\mathrm{j\&phi;}}_2},...,g_M{e}^{j{\mathrm{\&phi;}}_M})$

A＝[a(θ₁)，a(θ₂)，…，a(θ_K)]

$a\left(\mathrm{\&theta;}\right)=[{\mathrm{<figure-callout\; id="2"\; label="e\; J"\; filenames="C0312823800111.png"\; state="\{\{state\}\}">e</figure-callout>}}^J,{\mathrm{<figure-callout\; id="2"\; label="e\; J"\; filenames="C0312823800111.png"\; state="\{\{state\}\}">e</figure-callout>}}^J,...,{\mathrm{<figure-callout\; id="2"\; label="e\; J"\; filenames="C0312823800111.png"\; state="\{\{state\}\}">e</figure-callout>}}^J{]}^T$

S(m，n)＝[S₁(m，n)，S₂(m，n)，…，S_K(m，n)]^T

N(m，n)＝[N₁(m，n)，N₂(m，n)，…，N_M(m，n)]^T

Channel amplitude corrected g _i1 is ═ 1, i.e $G=\mathrm{diag}(e^{j{\mathrm{\&phi;}}_1},e^{j{\mathrm{\&phi;}}_2},...,e^{{\mathrm{j\&phi;}}_M}).$

The signal and noise components in the wig two-dimensional echo spectrum are the zero-mean stationary random process of each history, the signal and the noise are independent of each other, the noise of each channel is not related to each other, and the signals and the noise have the same variance sigma²The white Gaussian process of (1), the array covariance matrix of the two-dimensional echo spectrum is

R_ZZ＝E[Z(m，n)Z^H(m，n)]＝GAR_SSA^HG^H+σ²I (23)

Wherein R is_SS＝E[S(m，n)S^H(m，n)]. To R_ZZIs subjected to eigenvalue decomposition to obtain

R_ZZ＝UDU^H (24)

D is a radical of R_ZZIs constructed by diagonal arrays of eigenvalues

D＝diag(λ₁，λ₂，…，λ_M)λ₁≥λ₂≥…≥λ_M (25)

U is a sum of the eigenvalues λ₁Matrix of corresponding feature vectors

U＝[μ₁，μ₂，…，μ_M]＝[U_S，U_N] (26)

Wherein, U_S＝[μ₁，μ₂，…，μ_K]Corresponding to K eigenvalues representing the signal, the space spanned by the column vectors is called signal subspace; u shape_N＝[μ_K+1，μ_K+2，…，μ_M]The space spanned by the column vectors, corresponding to the M-K eigenvalues representing noise, is then referred to as the noise subspace.

The spatial spectrum function of DOA estimation is obtained by using MUSIC algorithm (R.O.Schmidt, Multiple event location and signaling, IEEE trans. antennas Propagation, 1986, Vol.34, pp.276-280.) proposed by Schmidt

$R_{\mathrm{MU}}=\frac{1}{{\left[\mathrm{Ga}\left(\mathrm{\&theta;}\right)\right]}^H{\mathrm{\&Pi;}}^{\&perp;}\left[\mathrm{Ga}\left(\mathrm{\&theta;}\right)\right]}---\left(27\right)$

Therein, II^⊥For projection operators on noise subspaces

${\mathrm{\&Pi;}}^{\&perp;}=U_N{U}_N^H=I-A{\left(A^{H}A\right)}^{-1}{A}^H---\left(28\right)$

Let Z (m, n) in equation (22) be a two-dimensional spectrum array output of a frequency point corresponding to a known reflection source echo (which can be detected in a two-dimensional echo spectrum according to information such as the distance and speed of the reflection source), thereby obtaining n^⊥. As is clear from the formula (21), G is substantially θ₁Can be expressed as G (theta)₁) The arrival direction θ of the echo of the reflection source is θ₀Known, then P_MUBecomes theta₁Function of (2)

$P_{\mathrm{MU}}\left({\mathrm{\&theta;}}_1\right)=\frac{1}{\left[G\left({\mathrm{\&theta;}}_1\right)a\left({\mathrm{\&theta;}}_0\right){]}^H{\mathrm{\&Pi;}}^{\&perp;}\right[G\left({\mathrm{\&theta;}}_1\right)a\left({\mathrm{\&theta;}}_0\right)]}---\left(29\right)$

Will P_MU(θ₁) Theta corresponding to spectrum peak₁The value is used as DOA estimation of a single arrival direction frequency point in a two-dimensional echo spectrum and is substituted into the formula (21) to obtain the channel phase gain e^jφ2. E found from different single arrival direction frequency points with noise in the actual echo spectrum^Jφ2Unlike this, statistical averaging may be performed to improve estimation accuracy. After the channel phase gain is estimated, the echo data of each channel is divided by the respective phase gain, so that the channel phase correction can be realized.

Fig. 4 is a schematic diagram of a uniform linear array commonly used in high-frequency ground wave radar, in which case, the detection of a single arrival direction frequency point in a echo spectrum is more accurate, and the above active phase correction method can be simplified.

The M-element uniform linear array can be divided into M-1 translation-invariant array element pairs A₁-A_MAny two of the combinations can be used for detecting the frequency points in the single arrival direction to obtain a corresponding set. Whether the frequency points fall into the intersection of a plurality of sets is used as a criterion, so that the detection of the frequency points in the single arrival direction is more accurate.

The coordinate of each array element of the uniform linear array is (x)_i，y_i) By substituting ((i-1) d, 0) into the formula (21)

$e^{J{\mathrm{\&phi;}}_i}=[Z_i(m^{\&prime;},n^{\&prime;})/Z_1(m^{\&prime;},n^{\&prime;})]e^{-\mathrm{<figure-callout\; id="2"\; label="J"\; filenames="C0312823800111.png"\; state="\{\{state\}\}">J</figure-callout>}\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}(i-1)d\sin {\mathrm{\&theta;}}_1}---\left(30\right)$

When the value of i is 2, the ratio of i to i is, ${\mathrm{<figure-callout\; id="2"\; label="e\; J"\; filenames="C0312823800111.png"\; state="\{\{state\}\}">e</figure-callout>}}^J=[Z_2(m^{\&prime;},n^{\&prime;})/Z_1(m^{\&prime;},n^{\&prime;})]e^{-J{\mathrm{\&phi;}}_2},$ can be obtained by substituting the above formula

$e^{j{\mathrm{\&phi;}}_1}=Z_1(m^{\&prime;},n^{\&prime;})\left[Z_2(m^{\&prime;},n^{\&prime;}){]}^{1-i}\right[Z_1(m^{\&prime;},n^{\&prime;}){]}^{i-2}{e}^{J(i-1){\mathrm{\&phi;}}_2}=B_i{e}^{J(i-1){\mathrm{\&phi;}}_2}---\left(31\right)$

Wherein, $B_i=Z_i(m^{\&prime;},n^{\&prime;})\left[Z_2(m^{\&prime;},n^{\&prime;}){]}^{1-i}\right[Z_1(m^{\&prime;},n^{\&prime;}){]}^{i-2}=e^{J[{\mathrm{\&phi;}}_i-(i-1){\mathrm{\&phi;}}_2]}$ is a quantity independent of the echo direction corresponding to the single arrival direction frequency point. The actual system contains noise and disturbance, and B obtained according to different single arrival direction frequency points_iAre not the same, can be obtained by statistical averagingSubstituted into formula (31) having

$e^{J{\mathrm{\&phi;}}_i}={\stackrel{\&OverBar;}{B}}_i{e}^{j(i-1){\mathrm{\&phi;}}_2}---\left(32\right)$

From the above formula, G is phi₂Can be expressed as G (phi)₂) The arrival direction θ of the echo of the reflection source is θ₀Then P is_MUBecomes phi₂Function of (2)

$P_{\mathrm{MU}}\left({\mathrm{\&phi;}}_2\right)=\frac{1}{\left[G\left({\mathrm{\&phi;}}_2\right)a\left({\mathrm{\&theta;}}_0\right){]}^H{\mathrm{\&Pi;}}^{\&perp;}\right[G\left({\mathrm{\&phi;}}_2\right)a\left({\mathrm{\&theta;}}_0\right)]}---\left(33\right)$

Will P_MU(φ₂) Phi corresponding to spectral peak₂The value is used as the estimation of the channel phase error of the array element 2 (based on the array element 1), and the channel phase gain e can be obtained by substituting the value into the formula (32)^jφ2。

The active phase correction method for the uniform linear array only carries out spectrum search once, and the statistical average is placed before the spectrum search, so that the calculation amount is small.

Claims (2)

1. A method for utilizing ocean echo to carry on array channel correction, characterized by that to receive the ocean echo with the antenna array comprising at least two array element pairs of invariant translation, through the statistical analysis to the two-dimensional spectrum of ocean echo, detect the frequency point of single direction of arrival among them; carrying out statistical averaging on the echo spectrum amplitude corresponding to the frequency points in the single arrival direction, estimating the amplitude characteristic of each channel, and realizing amplitude correction; and estimating the phase characteristics of each channel by adopting a multi-signal classification MUSIC algorithm according to the echo spectrum phase corresponding to the frequency point in the single arrival direction and the echo information of the known reflection signal source, and performing statistical averaging on a plurality of results to realize phase correction.

2. The method of claim 1, wherein the antenna array is a uniform linear array, the M-ary uniform linear array is divided into M-1 translational invariant array couples a 1-AM, any two of which are combined together to detect single arrival direction frequency points to obtain a corresponding set, and whether the set falls into the intersection of a plurality of such sets is taken as a criterion.

Images