Strong sea clutter suppression method based on multi-domain combination

Abstract

The sea clutter suppression method based on multi-domain combination belongs to the field of radar signal processing and solves the problem of poor accuracy of sea surface target detection under the existing sea clutter background. The present invention obtains all output signals of each antenna array; decomposes each output signal of each antenna array, extracts Doppler dimension information, distance dimension information and angle dimension information in the complex exponential component; utilizes the Doppler dimension information and the ROOT cyclic cancellation algorithm to eliminate the complex exponential component in the output signal; adopts the subspace decomposition method to remove the time-correlated clutter component of the output signal of the complex exponential component according to the distance dimension and the Doppler dimension; adopts the space-time adaptive processing algorithm to remove the spatial correlation residual clutter of the output signal of the time-correlated clutter component removed according to the Doppler dimension, distance dimension and angle dimension in the output signal of the clutter removal, and completes the suppression of strong sea clutter based on multi-domain combination. The present invention is suitable for sea clutter suppression in marine signal reception.

Description

Technical Field

[0001] The invention belongs to the field of radar signal processing, and in particular relates to a method for suppressing strong sea clutter signals. Background Art

[0002] Sea clutter is the backscattered echo of the local sea surface to the radar transmission signal, which is affected by multiple factors such as wind and waves. Using the characteristics of sea clutter to analyze the sea level conditions and realize the detection of sea level targets has important theoretical significance and practical value. Small target detection under the background of sea clutter is currently a hot topic and difficulty in research. At present, the detection of small targets under the background of sea clutter has the problem of poor accuracy of sea surface target detection due to the strong interference of high-frequency radar sea clutter on target detection. Summary of the invention

[0003] The purpose of the present invention is to solve the problem of poor accuracy of sea surface target detection under the existing sea clutter background, and proposes a strong sea clutter suppression method based on multi-domain combination.

[0004] The sea clutter suppression method based on multi-domain combination of the present invention comprises:

[0005] Step 1: According to the radar echo signal and the antenna array structure, all output signals of each antenna array are obtained;

[0006] Step 2: Decompose each output signal of each antenna array, obtain the complex exponential component in each output signal, and extract the Doppler dimension information, distance dimension information and angle dimension information in the complex exponential component;

[0007] Step 3: using the Doppler dimension information and the ROOT cyclic cancellation algorithm to eliminate the complex exponential component in the output signal;

[0008] Step 4: Using the subspace decomposition method, the output signal of the complex exponential component is subjected to time-correlated clutter removal according to the range dimension and the Doppler dimension;

[0009] Step 5: Adopting the space-time adaptive processing algorithm, the spatial correlation residual clutter removal is performed on the output signal of the clutter removal according to the Doppler dimension, distance dimension and angle dimension to remove the time-correlated clutter component output signal, so as to complete the suppression of strong sea clutter based on multi-domain joint.

[0010] Furthermore, in the present invention, in step 1, the method for obtaining all output signals of each antenna array according to the radar echo signal and the antenna array structure is:

[0011] The transmission signal of the nth transmitting element in the transmitting array:

[0012]

[0013] Where E is the total transmitted energy, T is the radar pulse duration, N is the total number of antenna arrays, and the baseband waveform φ n (t):

[0014]

[0015] in, is the baseband waveform φ n (t), assuming that the electromagnetic wave propagates independently in free space, in bidirectional propagation, the first antenna array element is selected as the reference point, and for any target, the received echo y of the mth receiving element of the antenna m (t) is:

[0016]

[0017] Where ξ represents the complex reflection coefficient of the target, f d,n represents the Doppler frequency, f d,n =2v t f n / c,v t Refers to the radial velocity of the target; f n is the operating frequency of the nth transmitting antenna, τ n,T and τ m,R are the sending and receiving time delays, respectively;

[0018]

[0019]

[0020] Among them, d T is the element spacing of the transmitting array, d R is the element spacing of the receiving array, τ 0 = 2r / c, θ represents the angle of the target, measured from the normal direction to the target direction, r represents the distance of the target from the receiving array, and c is the speed of light;

[0021] In the receiving array, the received signal is decomposed by N matched filters to obtain N isolated transmission signals. Then the nth output of the mth receiving antenna is:

[0022]

[0023] f d Indicates the Doppler frequency.

[0024] Furthermore, in the present invention, in step 2, the specific method for extracting the Doppler dimension information, the distance dimension information and the angle dimension information in the complex exponential component is:

[0025] Let the phase difference f dτ 0 =0:

[0026]

[0027] Extract three e index factors respectively, is the Doppler complex exponential component of the output signal, is the distance-dimensional complex exponential component of the output signal, is the angle-dimensional complex exponential component of the output signal.

[0028] Furthermore, in the present invention, in step 3, the specific method for eliminating the complex exponential component in the output signal is:

[0029] Step 31: Find the peak value of the spectrum and its corresponding frequency in the Doppler complex exponential component of the output signal;

[0030] Step 32: Establish an energy error function, search for the angle on [0, 2π] that minimizes the energy error function value, and estimate the initial phase of the complex exponential component in each output signal;

[0031] Step 32: Eliminate the complex exponential component of the output signal using the peak value of the spectrum found in step 31, its corresponding frequency and the estimated initial phase, and determine whether the elimination reaches the iteration number threshold. If so, complete the elimination of the complex exponential component in the output signal. Otherwise, return to execute step 31, where the iteration number threshold is determined according to the signal-to-noise ratio of the output signal.

[0032] Furthermore, in the present invention, in step 31, the process of searching for the angle on [0, 2π] that minimizes the energy error function to estimate the initial phase is:

[0033] Establish the energy error function ε(φ):

[0034]

[0035] Among them, x(n) is the Doppler spectrum of a single range gate, n corresponds to the Doppler unit, A is the highest peak of the spectral line of the entire Doppler spectrum, and f is the frequency corresponding to the peak. is the initial phase of the complex signal to be removed;

[0036] Estimate of the initial phase:

[0037]

[0038] is the initial phase estimate.

[0039] Furthermore, in the present invention, in step 4, the process of removing the time-correlated clutter component of the output signal of the complex exponential component according to the distance dimension and the Doppler dimension by using the subspace decomposition method is as follows:

[0040] Extract the echo column vector signal S containing the target signal range gate from the output signal after eliminating the complex exponential component T (k), k = 0, 1, 2, ... N-1;

[0041] S T =[S T (0),S T (1),S T (2),……,S T (N-1)] T

[0042] S i =[S i (0),S i (1),S i (2),……,S i (N-1)] T

[0043] Among them, S i is the echo column vector set of the range unit adjacent to the target, and the set contains M column vectors; S T is the echo column vector set containing the target signal range gate; the superscript T indicates vector transposition;

[0044] The covariance matrix is:

[0045]

[0046] Among them, the parameter γ i The solution formula is:

[0047]

[0048] The superscript T represents the matrix transpose, and the superscript H represents the conjugate transpose;

[0049] The covariance matrix R N×N Perform eigendecomposition:

[0050] R N×N =WΣW H

[0051] Get the eigenvalues, select the largest P eigenvalues ​​among the eigenvalues, where W represents the optimal solution of the window function and Σ is the sum symbol;

[0052] The linear space composed of the eigenvectors corresponding to the P eigenvalues ​​as a basis is used as a clutter subspace, the echo column vector containing the target signal is projected into the "clutter subspace" to estimate the clutter column vector, and the estimated clutter column vector is used to remove clutter from the output signal that eliminates the complex exponential component.

[0053] Further, in the present invention, in step 31, in step 5, a space-time adaptive processing algorithm is used to remove the spatial correlation residual clutter of the output signal of the clutter-removed output signal according to the Doppler dimension, the distance dimension and the angle dimension. The specific process is:

[0054] The output signal of the time-correlated clutter component is stacked into a column vector in the order of Doppler dimension and distance dimension to form a MN×1-dimensional space-time observation vector;

[0055] The specific splicing process is:

[0056] s(f,θ)=ves(f,θ)

[0057] Where vec(·) is a vectorization operation, and the two-dimensional data is stacked into a column in column order; s(f,θ) is the output signal after removing clutter; the output signal after removing clutter is expressed as a space-time steering vector:

[0058]

[0059] in, is the Kronecker product, a N (f) is the time orientation vector, v M (θ) is the time-oriented vector. The STAP algorithm is used to replace the spatial vector with the corresponding space-time vector:

[0060]

[0061] in, is the space-time covariance matrix of clutter.

[0062] Further, in the present invention, in step 31, the optimal solution W of the window function is:

[0063]

[0064] After the present invention adopts the full-space adaptive processing of three dimensions to suppress clutter, the clutter suppression effect is much better than the ROOT cyclic cancellation based on a single Doppler dimension and the subspace projection based on the Doppler dimension-range dimension. The components of sea clutter and ground clutter are also basically completely removed, and the signal-to-noise ratio (based on the maximum noise amplitude) is 13.40dB. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 is a flow chart of the method of the present invention;

[0066] Figure 2 It is a schematic diagram of the radar antenna transmitting signal system;

[0067] Figure 3 It is a schematic diagram of the radar antenna receiving signal system;

[0068] Figure 4 This is the sea clutter suppression effect diagram without using the subspace projection method;

[0069] Figure 5 This is the effect of suppressing sea clutter using the subspace projection method;

[0070] Figure 6 This is a schematic diagram of the one-dimensional range image target signal of the 158th Doppler unit of the measured data;

[0071] Figure 7 This is a schematic diagram of the one-dimensional range image sidelobe signal of the 158th Doppler unit of the measured data;

[0072] Figure 8 This is the effect of suppressing sea clutter without full space-time adaptive processing;

[0073] Fig. 9 This is the effect of using full space-time adaptive processing to suppress sea clutter. DETAILED DESCRIPTION

[0074] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0075] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0076] Specific implementation method 1: The following is combined Figure 1 This embodiment is described. The sea clutter suppression method based on multi-domain combination described in this embodiment includes:

[0077] Step 1: According to the radar echo signal and the antenna array structure, all output signals of each antenna array are obtained;

[0078] Step 2: Decompose each output signal of each antenna array, obtain the complex exponential component in each output signal, and extract the Doppler dimension information, distance dimension information and angle dimension information in the complex exponential component;

[0079] Step 3: using the Doppler dimension information and the ROOT cyclic cancellation algorithm to eliminate the complex exponential component in the output signal;

[0080] Step 4: Using the subspace decomposition method, the output signal of the complex exponential component is subjected to time-correlated clutter removal according to the range dimension and the Doppler dimension;

[0081] Step 5: Adopting the space-time adaptive processing algorithm, the spatial correlation residual clutter removal is performed on the output signal of the clutter removal according to the Doppler dimension, distance dimension and angle dimension to remove the time-correlated clutter component output signal, so as to complete the suppression of strong sea clutter based on multi-domain joint.

[0082] Furthermore, combined with Figure 2 and Figure 3 This embodiment is described. In this embodiment, in step 1, the method for obtaining all output signals of each antenna array according to the radar echo signal and the antenna array structure is:

[0083] The transmission signal of the nth transmitting element in the transmitting array:

[0084]

[0085] Where E is the total transmitted energy, T is the radar pulse duration, N is the total number of antenna arrays, and the baseband waveform φ n (t):

[0086]

[0087] in, is the baseband waveform φ n (t), assuming that the electromagnetic wave propagates independently in free space, in bidirectional propagation, the first antenna array element is selected as the reference point, and for any target, the received echo y of the mth receiving element of the antenna m (t) is:

[0088]

[0089] Where ξ represents the complex reflection coefficient of the target, f d,n represents the Doppler frequency, f d,n =2v t f n / c,v t Refers to the radial velocity of the target; f n is the operating frequency of the nth transmitting antenna, τn,T and τ m,R are the sending and receiving time delays, respectively;

[0090]

[0091]

[0092] Among them, d T is the element spacing of the transmitting array, d R is the element spacing of the receiving array, τ 0 = 2r / c, θ represents the angle of the target, measured from the normal direction to the target direction, r represents the distance of the target from the receiving array, and c is the speed of light;

[0093] In the receiving array (the receiving array corresponds to the transmitting array and the signal), the received signal is decomposed by N matched filters to obtain N isolated transmission signals. Then the nth output of the mth receiving antenna is:

[0094]

[0095] f d Indicates the Doppler frequency.

[0096] Further, in this implementation, in step 2, the specific method for extracting the Doppler dimension information, distance dimension information and angle dimension information in the complex exponential component is:

[0097] Let the phase difference f d τ 0 =0:

[0098]

[0099] Extract three e index factors respectively, is the Doppler complex exponential component of the output signal, is the distance-dimensional complex exponential component of the output signal, is the angle-dimensional complex exponential component of the output signal.

[0100] Furthermore, in the present invention, in step 3, the specific method for eliminating the complex exponential component in the output signal is:

[0101] Step 31: Find the peak value of the spectrum and its corresponding frequency in the Doppler complex exponential component of the output signal;

[0102] Step 32: Establish an energy error function, search for the angle on [0, 2π] that minimizes the energy error function value, and estimate the initial phase of the complex exponential component in each output signal;

[0103] Step 32: Eliminate the complex exponential component of the output signal using the peak value of the spectrum found in step 31, its corresponding frequency and the estimated initial phase, and determine whether the elimination reaches the iteration number threshold. If so, complete the elimination of the complex exponential component in the output signal. Otherwise, return to execute step 31, where the iteration number threshold is determined according to the signal-to-noise ratio of the output signal.

[0104] Further, in this implementation, in step 31, the process of searching for the angle on [0, 2π] that minimizes the energy error function to estimate the initial phase is:

[0105] Establish the energy error function ε(φ):

[0106]

[0107] Among them, x(n) is the Doppler spectrum of a single range gate, n corresponds to the Doppler unit, A is the highest peak of the spectral line of the entire Doppler spectrum, and f is the frequency corresponding to the peak. is the initial phase of the complex signal to be removed;

[0108] Estimate of the initial phase:

[0109]

[0110] is the initial phase estimate.

[0111] Furthermore, in this embodiment, in step 4, the process of removing the time-correlated clutter component of the output signal of the complex exponential component according to the range dimension and the Doppler dimension by using the subspace decomposition method is as follows:

[0112] Extract the echo column vector signal S containing the target signal range gate from the output signal after eliminating the complex exponential component T (k), k = 0, 1, 2, ... U-1;

[0113] S T =[S T (0),S T (1),S T (2),……,S T (U-1)] T

[0114] S i =[S i (0),S i (1),S i (2),……,S i (U-1)] T

[0115] Among them, S iis the echo column vector set of the range unit adjacent to the target, and the set contains V column vectors; S T is the echo column vector set containing the target signal range gate; the superscript T indicates vector transposition;

[0116] The covariance matrix is:

[0117]

[0118] Among them, the parameter γ i The solution formula is:

[0119]

[0120] The superscript T represents the matrix transpose, and the superscript H represents the conjugate transpose;

[0121] The covariance matrix R U×U Perform eigendecomposition:

[0122] R U×U =WΣW H

[0123] Get the eigenvalues, select the largest P eigenvalues ​​among the eigenvalues, where W represents the optimal solution of the window function and Σ is the sum symbol;

[0124] The linear space composed of the eigenvectors corresponding to the P eigenvalues ​​as a basis is used as a clutter subspace, the echo column vector containing the target signal is projected into the "clutter subspace" to estimate the clutter column vector, and the estimated clutter column vector is used to remove clutter from the output signal that eliminates the complex exponential component.

[0125] Further, in this embodiment, in step 31, in step 5, a space-time adaptive processing algorithm is used to remove the spatial correlation residual clutter of the output signal of the clutter removal according to the Doppler dimension, the distance dimension and the angle dimension of the output signal of the time-correlated clutter component removal. The specific process is:

[0126] The output signal of the time-correlated clutter component is stacked into a column vector in the order of Doppler dimension and distance dimension to form a MN×1-dimensional space-time observation vector;

[0127] The specific splicing process is:

[0128] s(f,θ)=ves(f,θ)

[0129] Where vec(·) is a vectorization operation, and the two-dimensional data is stacked into a column in column order; s(f,θ) is the output signal after removing clutter; the output signal after removing clutter is expressed as a space-time steering vector:

[0130]

[0131] in, is the Kronecker product, a N (f) is the time orientation vector, v M (θ) is the time-oriented vector. The STAP algorithm is used to replace the spatial vector with the corresponding space-time vector:

[0132]

[0133] in, is the space-time covariance matrix of clutter.

[0134] Further, in this implementation, in step 31, the optimal solution W of the window function is:

[0135]

[0136] The multi-domain joint sea clutter suppression method of the present invention can make the best use of the correlation of sea clutter in multiple dimensions to achieve the best effect of clutter suppression and target detection. Considering that there are three important dimensions in modern radar spectra - fast time dimension, slow time dimension, and angle dimension. Based on the objective existence of these three dimensions, this paper analyzes and compares the ROOT loop cancellation based on a single dimension (slow time dimension), the subspace projection method based on the joint action of two dimensions (slow time dimension and fast time dimension), and the space-time adaptive processing method (STAP method) based on the joint action of three dimensions (slow time dimension, fast time dimension and angle dimension).

[0137] In a general radar system of modern architecture, the transmitting array has N array elements and the receiving array has M array elements. Figure 2 As shown in the figure, in this model, the N elements can be regarded not only as N transmitting array element antennas irradiating the same target, but also as a certain transmitting source irradiating N different targets to generate different reflected signal waves. Assume that the element spacing of the transmitting array is d T , the element spacing of the receiving array is d R The operating frequency of the nth transmitting antenna in the transmitting array is f n , the baseband signal is recorded as Then the transmission signal of the nth transmitting array element can be expressed as

[0138]

[0139] Where E is the total transmitted energy and T is the radar pulse duration. We assume the baseband waveform Satisfy the orthogonality condition

[0140]

[0141] Assume that electromagnetic waves propagate independently in free space. We consider the case of two-way propagation and select the first array element as the reference point. Therefore, for any target, the received echo of the mth receiving array element can be written as:

[0142]

[0143] Where ξ represents the complex reflection coefficient of the target, f d,n represents the Doppler frequency, i.e. f d,n =2v t f n / c,v t refers to the radial velocity of the target. Of course, in the second context, f d,n It can also be expressed as the Doppler frequencies corresponding to n different moving targets. In addition, τ n,T and τ m,R are the sending and receiving time delays, respectively, and are given by:

[0144]

[0145]

[0146] Among them, τ 0 =2r / c, θ represents the angle of the target (measured from the normal direction to the target direction), and r represents the distance of the target from the receiving array; in the receiving array, the measured signal is down-converted, matched filtered and stored. The received signal can be decomposed by N matched filters to produce N isolated transmission signals. After matched filtering, the nth output of the mth receiving antenna is:

[0147]

[0148] Generally speaking, the speed of a slow-moving target will not accumulate into a large displacement in the extremely short sampling time interval of the radar. From the perspective of signal processing, the Doppler frequency corresponding to the slow target accumulates the phase difference f in the distance delay. d τ 0 ≈0, which is why we often call the sampling within the pulse "fast time", so Equation 4 can be further simplified to

[0149]

[0150] Analyzing the above formula, excluding the first constant factor, it can be decomposed into three e exponential factors, each of which represents an important information. represents the slow time dimension (Doppler dimension), represents the fast time dimension (distance dimension), represents the angle dimension. Based on these three dimensions, three different clutter suppression methods are proposed.

[0151] First, the ROOT cyclic cancellation algorithm based on the slow time dimension. After the ionospheric phase contamination correction, the broadened first-order Bragg peak of the high-frequency ground wave radar echo signal is sharpened, and the characteristics of local large values ​​will appear. The effect of ocean currents and sea breezes causes the spectrum peaks on the left and right sides of the sea clutter to shift in the same direction. Under ideal conditions, if the transmitter and receiver are deployed in a very close position, the echo signal including the clutter can be regarded as a linear combination of complex exponential signals of different frequencies; the traditional ROOT cyclic cancellation algorithm is to estimate these frequencies distributed near the target frequency, with complex exponential signal components with different amplitudes and initial phases. Through cyclic iteration, each time the amplitude, frequency and initial phase of a complex exponential component are estimated, the component is subtracted from the original time domain signal to achieve clutter suppression.

[0152] The estimation of the amplitude A is done by finding the peak and position of the spectrum in the Doppler dimension. The peak is identified as the amplitude of the complex exponential signal to be removed, and the peak position is identified as the frequency f of the complex exponential signal to be removed. The initial phase It is determined by searching on [0,2π] to find the angle that minimizes the "energy error function". The expression of the "energy error function" is:

[0153]

[0154] That is, the estimated value of the initial phase is:

[0155]

[0156] Among them, x(n) is the Doppler spectrum of a single range gate, n corresponds to the Doppler unit, n = 1, 2, ... N, A and f are the estimated amplitude and frequency. After estimating the amplitude, frequency and initial phase of the complex exponential signal, the first component to be removed is obtained. Subtracting this component from the original signal can achieve clutter suppression.

[0157] Based on the previous step, the next step is a sea clutter suppression method based on the combination of slow time dimension and fast time dimension - subspace decomposition method, which is also called matrix diagonalization or eigenvalue decomposition in mathematics. Using this matrix transformation method, the covariance matrix of the signal column vector set is eigendecomposed. It can be obtained that the "clutter subspace" is composed of the eigenvector representing the clutter component as the basis; and the "noise subspace" is composed of the eigenvector representing the noise component as the basis. Then, using the orthogonal nature of the eigenvectors, the echo signal containing the target is projected to obtain the clutter component, thereby removing the clutter and achieving the purpose of sea clutter suppression. The core idea of ​​the algorithm is: when the distance range to be studied is short, the sea clutter has a strong correlation in the distance dimension. The specific solution method is roughly: the range gate where the target is located, and several surrounding range gates that only contain sea clutter and do not contain target signals, form a column vector set, and solve its eigenvalue decomposition to obtain all eigenvalues ​​and eigenvectors. As we all know, the target echo signal received by the high-frequency ground wave radar has three characteristics: "low signal-to-clutter ratio", "high signal-to-noise ratio" and "high clutter-to-noise ratio". In other words, the energy of the clutter component is much stronger than the target signal and noise. We know that the eigenvalue obtained after the eigenvalue decomposition of the covariance matrix represents the energy or amplitude of the corresponding component. Therefore, we believe that the vector space composed of the eigenvector corresponding to the larger eigenvalue is the "clutter subspace"; the vector space composed of the remaining eigenvectors is regarded as the "noise subspace". Since the amplitude of the target signal and noise is much smaller than that of the clutter signal, for the convenience of research, we can put the target signal and noise together and call them the "noise subspace".

[0158] Next, the purpose of suppressing sea clutter can be achieved by removing the components that can be projected into the "clutter subspace" from the range unit echo containing both the target and sea clutter. It can be seen that the subspace projection method is based on the premise that clutter is unrelated to noise and targets / is independent of each other. Because of the existence of independence, the part of the range gate echo column vector containing the target that is projected into the clutter subspace completely represents the sea clutter, and then removing it can achieve clutter suppression.

[0159] Let S T (k) is the echo column vector signal containing the target signal range gate, k = 0, 1, 2, ... U-1. i (k)} is the echo column vector set of its adjacent distance unit, and the set contains V column vectors in total, that is, i = 1, 2, ..., V, then

[0160] S T =[S T (0),S T (1),S T(2),……,S T (U-1)] T

[0161] S i =[S i (0),S i (1),S i (2),……,S i (U-1)] T

[0162] The expression of the covariance matrix is:

[0163]

[0164] Among them, the parameter γ i The solution formula is

[0165]

[0166] The superscript T stands for matrix transpose, and the superscript H stands for conjugate transpose.

[0167] The covariance matrix R U×U Perform feature decomposition, that is

[0168] R U×U =W∑W H

[0169] Where W represents a U×U unitary matrix, which contains all eigenvectors, and Σ is a diagonal matrix with eigenvalues ​​on the diagonal and 0s in the rest of the positions. In the echo, clutter has a higher energy than noise, so the eigenvectors corresponding to the larger eigenvalues ​​in the matrix Σ represent the discrete "components" in the field of the clutter signal from a special perspective. The linear space composed of these eigenvectors is called the "clutter subspace"; the opposite is the "noise subspace". Therefore, by projecting the echo column vector containing the target signal into the "clutter subspace", the clutter column vector can be estimated, and the clutter can be removed by subtracting the two vectors.

[0170] A sea clutter suppression method based on the combination of slow time dimension, fast time dimension and angle dimension, full space-time adaptive processing. When the data has significant space-time characteristics, space-time adaptive processing (STAP) can often achieve certain results. The most general fully adaptive STAP processes the three dimensions of array element, distance and pulse of the entire data block at the same time.

[0171] Assume that a two-dimensional data block of a range gate is X, and its dimension is V×U. Usually X includes internal and external noise B, clutter component C and potential target echo component S, that is, X=S+C+B.

[0172] Stacking the column vectors in the two-dimensional data X into a column vector in slow time order will form an MN×1-dimensional space-time observation vector, also called a space-time snapshot. Within an accumulation cycle, there are as many space-time snapshots as there are range gates. Naturally, the space-time snapshot data also has its corresponding space-time steering vector. Its dimension is also MN×1,

[0173] The classic STAP algorithm is nothing more than a broadening of the MVDR algorithm. It is exactly the same in form. It can be obtained by replacing the spatial vectors with the corresponding space-time vectors.

[0174] Compared with the simple Doppler or fast and slow time domain joint algorithm, the full space-time joint domain processing algorithm (3D-STAP) can theoretically achieve a higher processing performance ceiling. When constructing the covariance, we need to select several distance-angle straightening vectors close to the processing target to construct, but we must take into account that the target signal is not a single isolated point in the distance unit, and there are side lobes. In other words, there is a certain width, such as Figure 6 and Figure 7 As shown. However, there is a correlation between the side lobe and the main peak of the signal, which will affect the effect of suppressing clutter. Therefore, we use a sliding window method to select the target distance and the upper and lower k range gates (the window width is 2k+1), select the adjacent distance-angle straightening vector, and exclude the vectors in the window. Specific embodiment:

[0176] After the first stage, the ROOT loop cancellation algorithm is processed. Figure 2 The measured data is tested and multiple cycles of cancellation are performed. The Doppler spectrum of any distance unit is selected as a case study to demonstrate the effect. After 4 cancellations, we find that the Bragg peak on the right and the ground clutter spectrum peak in the middle are significantly suppressed, but the strength of the target signal is still lower than the clutter. After 6 cancellations, the ground clutter has been completely eliminated, and the spectrum peak of the sea clutter cannot overwhelm the target. The spectrum line representing the target has become the highest peak of the entire spectrum. At this time, the cancellation effect has reached the best.

[0177] Subspace projection method. We select 10 range units above and below the range gate containing the target, a total of 40 column vectors to calculate the correlation. The suppression results are as follows Figure 4 and Figure 5 shown.

[0178] The subspace projection method combining Doppler dimension and range dimension is adopted. The clutter suppression effect is much better than ROOT cyclic cancellation based on single Doppler dimension. The components of sea clutter and ground clutter are almost completely removed, but at the expense of signal-to-noise ratio. The signal-to-noise ratio (based on the maximum noise amplitude) is 9.24dB.

[0179] The effect of STAP in suppressing clutter is as follows: Figure 8 and Fig. 9 Shown

[0180] We selected the original complete data and performed the same processing on all vectors of the range gates from 20 to 70. After the sea clutter was suppressed, the slow-moving targets distributed near the sea clutter appeared on the RD image in the form of scattered points.

[0181] After adopting the full-space adaptive processing of three dimensions for clutter suppression, the clutter suppression effect is much better than the ROOT cyclic cancellation based on a single Doppler dimension and the subspace projection based on the Doppler dimension-range dimension. The components of sea clutter and ground clutter are also basically completely removed, and the signal-to-noise ratio (based on the maximum noise amplitude) is 13.40dB.

[0182] Although the present invention is described herein with reference to specific embodiments, it should be understood that these embodiments are merely examples of the principles and applications of the present invention. It should therefore be understood that many modifications may be made to the exemplary embodiments and that other arrangements may be devised without departing from the spirit and scope of the present invention as defined by the appended claims. It should be understood that the various dependent claims and features described herein may be combined in a manner different from that described in the original claims. It should also be understood that features described in conjunction with individual embodiments may be used in other described embodiments.

Claims

1. A sea clutter suppression method based on multi-domain joint, characterized in that: include: Step 1: According to the radar echo signal and the antenna array structure, all output signals of each antenna array are obtained; Step 2: Decompose each output signal of each antenna array, obtain the complex exponential component in each output signal, and extract the Doppler dimension information, distance dimension information and angle dimension information in the complex exponential component; Step 3: using the Doppler dimension information and the ROOT cyclic cancellation algorithm to eliminate the complex exponential component in the output signal; Step 4: Using the subspace decomposition method, the output signal of the complex exponential component is subjected to time-correlated clutter removal according to the range dimension and the Doppler dimension; Step 5: Adopting the space-time adaptive processing algorithm, the spatial correlation residual clutter removal is performed on the output signal of the clutter removal according to the Doppler dimension, distance dimension and angle dimension to remove the time-correlated clutter component output signal, so as to complete the suppression of strong sea clutter based on multi-domain joint.

2. The sea clutter suppression method based on multi-domain joint according to claim 1 is characterized in that: In step 1, according to the radar echo signal and the antenna array structure, the method for obtaining all the output signals of each antenna array is: The transmission signal of the nth transmitting element in the transmitting array: Where E is the total transmitted energy, T is the radar pulse duration, N is the total number of antenna arrays, and the baseband waveform φ n (t): in, is the baseband waveform φ n (t), assuming that the electromagnetic wave propagates independently in free space, in bidirectional propagation, the first antenna array element is selected as the reference point, and for any target, the received echo y of the mth receiving element of the antenna m (t) is: Where ξ represents the complex reflection coefficient of the target, f d,n represents the Doppler frequency, f d,n =2v t f n / c,v t Refers to the radial velocity of the target; f n is the operating frequency of the nth transmitting antenna, τ n,T and τ m,R are the sending and receiving time delays, respectively; Among them, d T is the element spacing of the transmitting array, d R is the element spacing of the receiving array, τ0 = 2r / c, θ represents the angle of the target, measured from the normal direction to the target direction, r represents the distance of the target from the receiving array, and c is the speed of light; In the receiving array, the received signal is decomposed by N matched filters to obtain N isolated transmission signals. Then the nth output of the mth receiving antenna is: f D Indicates the Doppler frequency.

3. The sea clutter suppression method based on multi-domain joint according to claim 2 is characterized in that: In step 2, the specific method of extracting the Doppler dimension information, distance dimension information and angle dimension information in the complex exponential component is: Let the phase difference f d τ0=0: Extract three e index factors respectively, is the Doppler complex exponential component of the output signal, is the distance-dimensional complex exponential component of the output signal, is the angle-dimensional complex exponential component of the output signal.

4. The sea clutter suppression method based on multi-domain joint according to claim 3 is characterized in that: In step 3, the specific method of eliminating the complex exponential component in the output signal is: Step 31: Find the peak value of the spectrum and its corresponding frequency in the Doppler complex exponential component of the output signal; Step 32: Establish an energy error function, search for the angle on [0, 2π] that minimizes the energy error function value, and estimate the initial phase of the complex exponential component in each output signal; Step 32: Eliminate the complex exponential component of the output signal using the peak value of the spectrum found in step 31, its corresponding frequency and the estimated initial phase, and determine whether the elimination reaches the iteration number threshold. If so, complete the elimination of the complex exponential component in the output signal. Otherwise, return to execute step 31, where the iteration number threshold is determined according to the signal-to-noise ratio of the output signal.

5. The sea clutter suppression method based on multi-domain joint according to claim 4 is characterized in that: In step 31, the process of searching for the angle on [0,2π] that minimizes the energy error function to estimate the initial phase is: Establish the energy error function ε(φ): Where x(n) is the Doppler spectrum of a single range gate, n corresponds to the Doppler unit, A is the highest peak of the entire Doppler spectrum, and f is the frequency corresponding to the peak. is the initial phase of the complex signal to be removed; Estimate of the initial phase: is the initial phase estimate.

6. The sea clutter suppression method based on multi-domain joint according to claim 5 is characterized in that: In step 4, the subspace decomposition method is used to remove the time-correlated clutter components of the output signal of the complex exponential component according to the range dimension and the Doppler dimension. The process is: Extract the echo column vector signal S containing the target signal range gate from the output signal after eliminating the complex exponential component T (k), k = 0, 1, 2, ... U-1; S T =[S T (0),S T (1),S T (2),……,S T (U-1)] T S i =[S i (0),S i (1),S i (2),……,S i (U-1)] T Among them, S i is the echo column vector set of the range unit adjacent to the target, and the set contains V column vectors; S T is the echo column vector set containing the target signal range gate; the superscript T indicates vector transposition; The covariance matrix is: Among them, the parameter γ i The solution formula is: The superscript T represents the matrix transpose, and the superscript H represents the conjugate transpose; The covariance matrix R U×U Perform eigendecomposition: R U×U =WΣW H Get the eigenvalues, select the largest P eigenvalues ​​among the eigenvalues, where W represents the optimal solution of the window function and Σ is the sum symbol; The linear space composed of the eigenvectors corresponding to the P eigenvalues ​​as the basis is used as the clutter subspace, the echo column vector containing the target signal is projected into the "clutter subspace" to estimate the clutter column vector, and the estimated clutter column vector is used to remove clutter from the output signal that eliminates the complex exponential component.

7. The sea clutter suppression method based on multi-domain joint according to claim 6 is characterized in that: In step 5, the space-time adaptive processing algorithm is used to remove the spatial correlation residual clutter from the output signal of the clutter removal according to the Doppler dimension, distance dimension and angle dimension. The specific process is: The output signal of the time-correlated clutter component is stacked into a column vector in the order of Doppler dimension and distance dimension to form a MN×1-dimensional space-time observation vector; The specific splicing process is: s(f,θ)=ves(f,θ) Where vec(·) is a vectorization operation, and the two-dimensional data is stacked into a column in column order; s(f,θ) is the output signal after removing clutter; the output signal after removing clutter is expressed as a space-time steering vector: in, is the Kronecker product, a N (f) is the time orientation vector, v M (θ) is the time-oriented vector, and the STAP algorithm is used to replace the spatial vector with the corresponding space-time vector: in, is the space-time covariance matrix of clutter.

8. The sea clutter suppression method based on multi-domain joint according to claim 7 is characterized in that: The optimal solution W of the window function is:

Images