Radar additive composite interference space-time combined suppression method and system based on blind source separation

Abstract

The invention discloses a blind source separation-based radar additive composite interference space-time combined suppression method and system. The method comprises the following steps of: S1, estimating a DOA (Direction of Arrival) of a composite interference signal according to an observation signal; s2, performing coherent accumulation on the radar echo signals to improve the signal-to-noise ratio; s3, constructing a blocking matrix by using the DOA information of the interference signal, and suppressing composite interference; s4, whitening processing is carried out on the signals output in the S3, and correlation among different channel signals is removed; s5, suppressing residual interference signals of the whitened signals by adopting a BCA algorithm; and S6, constructing a signal similarity measurement basis for the signals output in the S5 by using a mutual function, and selecting a real echo signal from the basis. According to the method, the space domain anti-interference method and the time domain anti-interference method are effectively combined, radar composite interference is suppressed, and compared with an existing blind source separation anti-interference method, the effect is better, and the consumed time is shorter.

Description

Technical Field

[0001] The invention belongs to the technical field of radar countermeasures, and in particular relates to a method and system for space-time joint suppression of radar additive composite interference based on blind source separation (Blind Signal Separation, BSS). Background Art

[0002] Radar, as an important detection device, can detect targets and determine their positions in space using electromagnetic signals. It is widely used in many fields, especially in the military field. With the development of technology, electronic warfare (EW) has become increasingly fierce, and radar faces serious interference threats. Using strong electronic interference to suppress the opponent's radar intelligence system has become the most effective and most commonly used electromagnetic countermeasure, resulting in great challenges to the real-time and authenticity of radar system information acquisition.

[0003] The interference effect of using only one interference style is often not good. The jammer needs to combine multiple interference styles to form a composite interference to the radar. The commonly used composite interference is the additive combination of noise interference and deception interference, which can not only suppress the real target but also enable the radar to detect false targets to achieve the purpose of deception. The impact of the composite interference electromagnetic environment on the radar system is very serious, which seriously affects the radar system's perception and information acquisition capabilities on the battlefield. In order to enable the radar to function normally in the future battlefield environment and provide information security for its own side, it is necessary to study radar anti-composite interference technology. However, under the threat of additive composite interference faced by radar, the radar anti-composite interference method used in the existing technology has the problems of high complexity and slow convergence. Summary of the invention

[0004] Aiming at the additive composite interference threat faced by radar, the existing radar anti-composite interference method has the problems of high complexity and slow convergence. The present invention proposes a radar additive composite interference space-time joint suppression method and system based on blind source separation (BSS), which improves the convergence speed.

[0005] Suppose the following confrontation scenario: a single array radar uses a uniform linear array (ULA) antenna, M antenna elements are arranged on the same straight line at uniform intervals, and the target echo signal and the interference signal differ in spatial azimuth. The observation signal model can be expressed as:

[0006] X=AS+N=[x1,x2,…,x M ] T

[0007] Where X represents the M×L dimensional observation signal matrix, and L is the number of signal snapshots. The M×K dimensional steering matrix A=[a(θ1),a(θ2),…,a(θ K )] any column vector a(θ k ) is the incident angle θ k The steering vector of the spatial source signal (k=1,…,K) is mathematically expressed as a(θ k )=[1,…,exp(j2π(M-1)(d / λ)sin(θ k ))] T S=[s1,s2,…,s K ] T is a K×L-dimensional spatial source signal matrix, N=[n1,n2,…,n M ] T is an M×L dimensional noise matrix.

[0008] In order to achieve the above-mentioned invention object, the present invention adopts the following technical scheme:

[0009] A space-time joint suppression method for radar additive composite interference based on BSS, comprising the following steps:

[0010] S1. Estimate the direction of arrival (DOA) of the interference signal based on the observed signal data.

[0011] S2, accumulate the pulse signal to improve the signal and interference to noise ratio (SINR);

[0012] S3, using the DOA information of the interference signal obtained in step S1 to construct a blocking matrix, and eliminating the interference signal energy in the accumulated signal in step S2 through blocking matrix preprocessing (Blocking Matrix Preprocessing, BMP), so as to suppress the interference signal;

[0013] S4, performing whitening processing on the output signal after the blocking matrix preprocessing in step S3 to remove the correlation between signals of different channels;

[0014] S5, using a blind source separation algorithm to suppress the residual interference in the signal whitened in step S4;

[0015] S6. Select the echo signal from the separated signals.

[0016] As a preferred solution, in step S1, estimating the DOA of the interference signal includes the following steps:

[0017] S11, the radar first calculates the covariance matrix C based on the observed signal X XX Right now:

[0018]

[0019] In the formula, (·) H is the conjugate transpose, and L is the number of signal snapshots.

[0020] S12, observation signal covariance matrix C XX The eigenvalue decomposition is:

[0021] C XX =QΛQ H

[0022] In the formula, Q is the characteristic matrix and Λ is the eigenvalue matrix. The eigenvalue matrix Λ=diag(λ1,…,λ M ) Arrange in descending order by element size, that is, λ1≥λ2≥…≥λ M The eigenvectors in the characteristic matrix Q are rearranged into Q′ according to the corresponding eigenvalue order; the first K+1 large eigenvalues ​​of the eigenvalue matrix Λ are retained, and the average value of the remaining eigenvalues ​​is:

[0023]

[0024] The reconstructed eigenvalue matrix Λ′=diag(λ1,…,λ K+1 ,σ 2 ,…,σ 2 ), the reconstructed covariance matrix C′ XX Right now:

[0025] C′ XX =Q′Λ′Q′ H

[0026] Where Λ′ is the reconstructed eigenvalue matrix, and Q′ is the rearranged feature matrix.

[0027] S13. In order to improve the accuracy of DOA estimation, the reconstructed covariance matrix C′ is used XX The DOA of the interference signal is estimated by combining a high-resolution arrival angle estimation algorithm (such as the Root-MUSIC algorithm).

[0028] As a preferred solution, in step S2, the signal accumulation adopts coherent accumulation, and the coherently accumulated signal is X C , the process is expressed as:

[0029]

[0030] In the formula, x C is a signal output by coherent accumulation, x kis the echo signal of the kth pulse, N is the number of coherent pulses, e is a natural constant, j is an imaginary unit, and φ is the phase difference.

[0031] As a preferred solution, the blocking matrix preprocessing in step S3 includes the following steps:

[0032] S31. Calculate the blocking matrix B of the interference signal according to the DOA of the interference signal; suppose the number of interference sources is N, and calculate the interference weight according to the DOA of the interference signal:

[0033]

[0034] In the formula, u i is the weight of the i-th interference source, θ i is the DOA of the i-th interference source; let μ be a row vector containing N+1 elements, where the elements are expressed as:

[0035] μ i =(-1) i-1 sum{e sum[nchoosek(u,i-1)]},i=1,…,N+1

[0036] Where sum(·) is the cumulative sum function, nchoosek(·) is the combination function; the blocking matrix B is initialized as a (MN)×M all-zero matrix, and the elements filled into it are:

[0037] B(i,i:i+N)=μ,i=1,…,MN

[0038] S32, performing blocking matrix preprocessing on the accumulated signal, namely:

[0039] X B =BX C

[0040] Where, is the B blocking matrix, X B is the signal obtained after preprocessing.

[0041] As a preferred solution, the whitening process in step S4 includes the following steps:

[0042] S41, calculate preprocessing signal X B Whitening matrix W;

[0043] Signal X B The covariance matrix C B Shure breaks it down into:

[0044]

[0045] Among them, Q and Q ⊥ are highly semi-orthogonal matrices, and the column vectors of the two matrices span the signal subspace and the noise subspace respectively; Is a diagonal matrix containing n non-zero eigenvalues; the signal X B The whitening matrix is ​​expressed as:

[0046] W=Σ -1 Q H

[0047] S42, using the whitening matrix W to pre-process the signal X B The whitening process is:

[0048] Z=WX B

[0049] Where Z is the signal matrix after whitening processing.

[0050] As a preferred solution, in step S5, the blind source separation algorithm uses a bounded component analysis algorithm (BCA) to suppress residual interference. The initial parameters of the algorithm are set to: 100 iterations, 0.1 initial algorithm accuracy, and 10 final accuracy. -5 .

[0051] As a preferred solution, in step S6, a signal similarity measurement basis is constructed using a mutual function for the signal output in step S5, and a true echo signal is selected therefrom, which specifically includes the following steps:

[0052] S61, calculate the cross-correlation function between the separated signal and the original pulse signal, namely:

[0053]

[0054] In the formula, R i is the cross-correlation function of the i-th signal, s is the reference pulse signal, is the separated i-th signal;

[0055] S62, taking the maximum absolute value of the cross-correlation function as the signal similarity measurement basis, namely:

[0056] ρ i =max(R i |)

[0057] In the formula, ρ i Signal similarity measure basis;

[0058] S63, selecting the signal with the greatest signal similarity as the echo signal, that is:

[0059] i0=argmax(ρ i )

[0060] Therefore, the echo signal

[0061] The present invention also discloses a radar additive composite interference space-time joint suppression system based on blind source separation, which is used to execute the above method and includes the following modules:

[0062] DOA estimation module: estimates the direction of arrival (DOA) of the interference signal based on the observed signal data;

[0063] Pulse accumulation module: accumulates pulse signals to improve the signal-to-noise ratio;

[0064] Blocking matrix preprocessing module: Use the DOA information of the interference signal to construct a blocking matrix, eliminate the interference signal energy in the accumulated signal through blocking matrix preprocessing, and achieve interference signal suppression;

[0065] Whitening processing module: whitening the output signal after blocking matrix preprocessing;

[0066] Blind source separation module: uses blind source separation algorithm to suppress the residual interference in the whitened signal;

[0067] Signal selection module: selects the echo signal from the separated signals.

[0068] Compared with the prior art, the advantages of the present invention are as follows:

[0069] (1) The BCA algorithm is applied to the field of radar anti-interference for the first time. Compared with the existing radar anti-interference method using the blind source separation algorithm, the present invention has low complexity and short running time;

[0070] (2) The present invention proposes a spatial and time domain joint anti-radar composite interference solution. By combining the spatial domain anti-interference method with the time domain anti-interference method, a better anti-composite interference effect can be obtained.

[0071] (3) The present invention proposes a method for deriving a blocking matrix under multiple interference sources, which can derive a blocking matrix under multiple interference sources and effectively suppress interference in a scenario where multiple interference sources generate complex interference.

[0072] (4) The present invention proposes a signal selection method based on the cross-correlation function, which can effectively extract the echo signal from the multi-path signal after anti-interference processing through experiments;

[0073] In view of the composite interference threat faced by radar and the low operating efficiency and poor anti-composite interference effect of existing methods, the radar additive composite interference space-time joint suppression method and system based on blind source separation introduce the BCA algorithm into the radar anti-interference problem, and solve the problem that active deceptive interference is difficult to suppress by combining DOA estimation and blocking matrix preprocessing. At the same time, the complexity and running time of the anti-composite interference method are reduced by the BCA algorithm. In addition, the present invention also proposes a method for selecting an echo signal from a multi-path signal, which can effectively select the correct echo signal. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] Figure 1 It is a flow chart of a BSS-based radar additive composite interference space-time joint suppression method provided by a preferred embodiment of the present invention;

[0075] Figure 2 It is a signal processing flow chart of a BSS-based radar additive composite interference space-time joint suppression method provided by a preferred embodiment of the present invention;

[0076] Figure 3 It is a comparison diagram before and after the suppression of noise amplitude modulation and dense false target interference additive composite interference provided by the preferred embodiment of the present invention;

[0077] Figure 4 It is a comparison diagram before and after the suppression of noise amplitude modulation, dense false target interference and intermittent sampling and forwarding additive composite interference provided by the preferred embodiment of the present invention;

[0078] Figure 5 is a performance comparison diagram of the preferred method of the present invention and other interference suppression methods;

[0079] Figure 6 It is a block diagram of a BSS-based radar additive composite interference space-time joint suppression system provided by a preferred embodiment of the present invention. DETAILED DESCRIPTION

[0080] The technical solution of the present invention is further explained below by way of examples.

[0081] See also Figure 1 and Figure 2 , Figure 1 is a method flow chart provided by a preferred embodiment of the present invention, Figure 2 1 is a signal processing overall block diagram provided by a preferred embodiment of the present invention. Specifically, a BSS-based radar additive composite interference space-time joint suppression method of this embodiment includes the following steps:

[0082] S1. Estimate the direction of arrival (DOA) of the interference signal based on the observed signal data. This step is as follows:

[0083] S11, the radar first calculates the covariance matrix C based on the observed signal X XX Right now:

[0084]

[0085] In the formula, (·) H is the conjugate transpose, and L is the number of signal snapshots.

[0086] S12, observation signal covariance matrix C XX The eigenvalue decomposition is:

[0087] C XX =QΛQ H

[0088] In the formula, Q is the characteristic matrix and Λ is the eigenvalue matrix. The eigenvalue matrix Λ=diag(λ1,…,λ M ) Arrange in descending order by element size, that is, λ1≥λ2≥…≥λ M The eigenvectors in the characteristic matrix Q are rearranged into Q′ according to the corresponding eigenvalue order. The first K+1 large eigenvalues ​​of the eigenvalue matrix Λ are retained, and the average value of the remaining eigenvalues ​​is:

[0089]

[0090] The reconstructed eigenvalue matrix Λ′=diag(λ1,…,λ K+1 ,σ 2 ,…,σ 2 ), the reconstructed covariance matrix C′ XX for:

[0091] C′ XX =Q′Λ′Q′ H

[0092] S13. In order to improve the accuracy of DOA estimation, the reconstructed covariance matrix C′ is used XX The DOA of the interference signal is estimated by combining a high-resolution arrival angle estimation algorithm (such as the Root-MUSIC algorithm).

[0093] S2, perform coherent accumulation on the pulse signal to improve the signal-to-noise ratio. The coherently accumulated signal is X C , the process can be expressed as:

[0094]

[0095] In the formula, x C is a signal output by coherent accumulation, x i is the echo signal of the kth pulse, N is the number of coherent pulses, and φ is the phase difference.

[0096] S3, using the DOA information of the interference signal obtained in step S1 to construct a blocking matrix, and eliminating the interference signal energy in the accumulated signal through blocking matrix preprocessing to achieve suppression of the interference signal. This step is specifically as follows:

[0097] S31. Calculate the blocking matrix B of the interference signal according to the DOA of the interference signal. Assume that the number of interference sources is N, and calculate the interference weight according to the DOA of the interference signal:

[0098]

[0099] In the formula, u i is the weight of the i-th interference source, θ i is the DOA of the ith interference source. Let μ be a row vector containing N+1 elements, where the elements can be expressed as:

[0100]

[0101] In the formula, sum(·) is the cumulative sum function, and nchoosek(·) is the combination function. The blocking matrix B is initialized as a (MN)×M all-zero matrix, and the elements filled into it are:

[0102] B(i,i:i+N)=μ,i=1,…,MN

[0103] S32, performing blocking matrix preprocessing on the accumulated signal, namely:

[0104] X B =BX C

[0105] Where, is the B blocking matrix, X B is the signal obtained after preprocessing.

[0106] S4, whitening the output signal after the blocking matrix preprocessing in step S3. This step is as follows:

[0107] S41, calculate preprocessing signal X B Whitening matrix W. Signal X B The covariance matrix C B Shure breaks it down into:

[0108]

[0109] Among them, Q and Q ⊥ are highly semi-orthogonal matrices, and the column vectors of the two matrices span the signal subspace and the noise subspace respectively. Is a diagonal matrix containing n non-zero eigenvalues. To reduce the dimensionality of the signal, the signal X BThe whitening matrix can be expressed as:

[0110] W=Σ -1 Q H

[0111] S42, using the whitening matrix W to pre-process the signal X B The whitening process is:

[0112] Z=WX B

[0113] Where Z is the signal matrix after whitening processing.

[0114] S5, using a blind source separation algorithm to suppress the residual interference in the signal whitened in step S4. The process of using the BCA algorithm to suppress the interference signal from the signal after whitening can be expressed as:

[0115]

[0116] In the formula, u m is the mth column vector of the source signal extraction matrix U. L ] Projected onto the hyperplane can be recorded as a point set:

[0117] Y={(Re(y l ),Im(y l ))l=1,…,L}

[0118] Where Re(·) is the real part of the complex number, Im(·) is the imaginary part of the complex number, and L is the number of signal snapshots. Obviously, the point set Y is a convex set, and its boundary can be defined as a convex hull. The mathematical representation of the convex hull is:

[0119]

[0120] In the formula, is the convex hull of the convex set Y, It records the elements that make up the convex hull, and conv(·) is the convex hull calculation.

[0121] The perimeter of the convex hull can be written as

[0122]

[0123] Contrast function of the nth iteration constructed from the perimeter of the convex hull It can be expressed as:

[0124]

[0125] In the formula, To extract vector The L2 norm of is used to normalize the contrast function. Therefore Δy i =u m Δz i , i=1,…,V, The gradient is:

[0126]

[0127] The step size of gradient descent is set to:

[0128]

[0129] Where eta is the algorithm accuracy. Extract vector u m The update can be expressed as:

[0130]

[0131] Here the new extraction vector Unitized. Extracted signal It can be expressed as:

[0132]

[0133] S6, selecting the echo signal from the separated signals. The specific process of this step is as follows:

[0134] S61, calculate the cross-correlation function between the separated signal and the original pulse signal, namely:

[0135]

[0136] In the formula, R i is the cross-correlation function of the i-th signal, s is the reference pulse signal, is the separated i-th signal;

[0137] S62, taking the maximum absolute value of the cross-correlation function as the signal similarity measurement basis, namely:

[0138] ρ i =max(R i |)

[0139] In the formula, ρ i Signal similarity measure basis;

[0140] S63, selecting the signal with the greatest signal similarity as the echo signal, that is:

[0141] i0=argmax(ρ i )

[0142] Therefore, the echo signal

[0143] The test was conducted according to the above embodiment, where the number of space targets was 1 and the number of space sources was known. The signal-to-noise ratio (SNR) was 10dB, the real target was set at 2000 range bins from the radar, and the incident angle of the echo signal was 0°; the composite interference was noise amplitude modulation (NAM) interference and dense false target (DFTJ) interference, which were generated by two interference sources located at different spatial positions and added and compounded at the radar end, where the jamming-to-signal ratio (JSR) of NAM was 20dB and the incident angle was 1°; the JSR of DFTJ was 10dB and the incident angle was 0.1°, and the composite interference suppression test was conducted, see Figure 3 . Figure 3 It is a comparison diagram of the composite interference suppression effect provided by the preferred embodiment of the present invention, wherein (a), (b), and (c) are respectively the time domain diagram, time-frequency diagram, and pulse pressure diagram of the observed signal before interference suppression; (d), (e), and (f) are respectively the time domain diagram, time-frequency diagram, and pulse pressure diagram of the selected echo signal after interference suppression. It can be seen that the method provided by the present invention can effectively counteract the additive composite of two interference sources. The time-frequency domain image of the selected echo signal is clean and clear, and the signal time-frequency characteristics can be clearly seen. After pulse compression, the target peak is clear. After calculation, after taking anti-interference measures, the signal-to-noise ratio gain of the output signal compared with the observed signal is 34.60dB.

[0144] When the SNR is 10dB, the real target is set at 2000 range bins away from the radar, and the incident angle of the echo signal is 0°; the composite interference is noise modulation interference, dense false target interference and interrupted-sampling Repeater Jamming (ISRJ) interference generated by three interference sources located in different spatial positions at the radar end, among which the JSR of NAM is 20dB and the incident angle is -1°; the JSR of DFTJ is 10dB and the incident angle is 0.2°; the JSR of ISRJ is 15dB and the incident angle is 0.1°, a composite interference suppression test is carried out, see Figure 4 . Figure 4It is a comparison diagram of the composite interference suppression effect provided by the preferred embodiment of the present invention, wherein (a), (b), and (c) are respectively the time domain diagram, time-frequency diagram, and pulse pressure diagram of the observed signal before interference suppression; (d), (e), and (f) are respectively the time domain diagram, time-frequency diagram, and pulse pressure diagram of the selected echo signal after interference suppression. After using the method of the present invention for anti-interference, the recovered real echo signal is clearer, in sharp contrast to that before anti-interference. The comparison between (f) and (c) shows that the false target peak is obviously suppressed, which will not affect the detection of the real target peak. After calculation, compared with the observed signal, the signal-to-interference-to-noise ratio gain of the recovered real echo signal is 40.14dB.

[0145] When the incident angle of the real echo signal is 0°; the JSR of NAM is 15dB, the incident angle is 1°; the JSR of DFTJ is 10dB, the incident angle is 0.3°; the JSR of ISRJ is 15dB, the incident angle is -0.5°; the SNR is controlled to change from 5dB to 15dB, and 50 Monte Carlo experiments are performed at each SNR. Please combine Figure 5 . Among them, (a), (b), and (c) are respectively the curves of signal-to-noise ratio gain changing with signal-to-noise ratio (SNR), the curves of false target suppression rate changing with signal-to-noise ratio (SNR), and the curves of algorithm running time changing with signal-to-noise ratio (SNR). (a) shows that compared with the anti-interference method (EMP-JADE) of feature projection combined with feature matrix joint similarity diagonalization algorithm and the anti-interference method (MusicMethod) of blind source separation combined with multi-signal classification, the SINR gain of the method (BMP-BCA) provided by the present invention is slightly lower than that of MusicMethod, and the effect is similar to that of EMP-JADE. (b) shows that with the increase of SNR, the false target suppression rate of the method provided by the present invention and MusicMethod increases significantly, the false target suppression rate of EMP-JADE fluctuates, and the false target suppression rate of the method provided by the present invention is better than that of the other two methods. (c) shows that the running time of the method of the present invention is much shorter than the two compared methods, and the algorithm efficiency is higher.

[0146] like Figure 6 As shown, the present invention also discloses a radar additive composite interference space-time joint suppression system based on blind source separation, which is used to execute the above method and includes the following modules:

[0147] DOA estimation module: estimates the DOA of the interference signal according to the observed signal X, and provides the DOA information of the interference signal to the blocking matrix preprocessing module;

[0148] Pulse accumulation module: improves the signal-to-noise ratio by using coherent accumulation according to the observed signal;

[0149] Blocking matrix preprocessing module: Use the DOA of the interference signal to perform blocking matrix preprocessing on the observed signal X to suppress the composite interference;

[0150] Whitening processing module: eliminates the correlation between signals from different channels, making the characteristics of each independent signal source more obvious;

[0151] Blind source separation module: optimizes signals from different channels and suppresses residual interference signals and noise introduced by blocking matrix preprocessing;

[0152] Signal selection module: selects the echo signal as output from the multi-path signals separated by blind source separation.

[0153] For other contents of this embodiment, please refer to the above method embodiment.

[0154] In summary, the above content is a further detailed description of the present invention in combination with specific implementation methods, but it cannot be determined that the specific implementation of the present invention is limited to these descriptions. The DOA estimation method based on covariance matrix reconstruction, the blocking matrix derivation method under multiple interference sources, and the signal selection method based on the cross-correlation function are the core technical contents of the present invention. All the semi-supervised transfer learning identification methods of radar composite interference under small samples formed by corresponding modifications, replacements, improvements, etc. within the above-mentioned design principles and implementation points of the present invention should be included in the protection scope of the present invention.

Claims

1. A space-time joint suppression method for radar additive composite interference based on blind source separation, characterized in that: The following steps are involved: S1. Estimate the direction of arrival (DOA) of the interference signal based on the observed signal data; S2, accumulate the pulse signal to improve the signal-to-interference-noise ratio; S3, using the DOA information of the interference signal obtained in step S1 to construct a blocking matrix, and eliminating the interference signal energy in the signal accumulated in step S2 through blocking matrix preprocessing to achieve suppression of the interference signal; S4, performing whitening processing on the output signal after the blocking matrix preprocessing in step S3; S5, using a blind source separation algorithm to suppress residual interference on the signal whitened in step S4; S6. Select the echo signal from the separated signals.

2. The method for joint space-time suppression of radar additive composite interference based on blind source separation according to claim 1 is characterized in that: Step S1 includes the following steps: S11, radar calculates the covariance matrix C based on the observed signal X XX : In the formula, (·) H is the conjugate transpose, L is the number of signal snapshots; S12, observation signal covariance matrix C XX Perform eigenvalue decomposition: C XX =QΛQ H Where Q is the characteristic matrix and Λ is the eigenvalue matrix; the eigenvalue matrix Λ=diag(λ1,…,λ M ) Arrange in descending order by element size, that is, λ1≥λ2≥…≥λ M The eigenvectors in the characteristic matrix Q are rearranged into Q′ according to the corresponding eigenvalue order; the first K+1 large eigenvalues ​​of the eigenvalue matrix Λ are retained, and the average value of the remaining eigenvalues ​​is: The reconstructed eigenvalue matrix Λ′=diag(λ1,…,λ K+1 ,σ 2 ,…,σ 2 ), the reconstructed covariance matrix C′ XX for: C′ XX =Q′Λ′Q′ H S13. Use the reconstructed covariance matrix C′ XX The DOA of the interference signal is estimated by combining a high-resolution arrival angle estimation algorithm.

3. The method for joint space-time suppression of radar additive composite interference based on blind source separation according to claim 2 is characterized in that: In step S2, the pulse signal is coherently accumulated to improve the signal-to-interference-noise ratio. The coherently accumulated signal is X C , the process is expressed as: In the formula, x C is a signal output by coherent accumulation, x k is the echo signal of the kth pulse, N is the number of coherent pulses, e is a natural constant, j is an imaginary unit, and φ is the phase difference.

4. The method for space-time joint suppression of radar additive composite interference based on blind source separation according to claim 3 is characterized in that: Step S3 includes the following steps: S31. Calculate the blocking matrix B of the interference signal according to the DOA of the interference signal; suppose the number of interference sources is N, and calculate the weight of the interference according to the DOA of the interference signal: In the formula, u i is the weight of the i-th interference source, θ i is the DOA of the i-th interference source; let μ be a row vector containing N+1 elements, where the elements are expressed as: μ i (-1) i-1 sum{e sum[nchoosek(u,i-1)] },i=1,…,N+1 Where sum(·) is the cumulative sum function, nchoosek(·) is the combination function; the blocking matrix B is initialized as a (MN)×M all-zero matrix, and the elements filled into it are: B(i,i:i+N)=μ,i=1,…,MN S32, performing blocking matrix preprocessing on the accumulated signal: X B =BX C Where B is the blocking matrix, X B is the signal obtained after preprocessing.

5. The method for joint space-time suppression of radar additive composite interference based on blind source separation according to claim 4 is characterized in that: Step S4 is specifically as follows: S41, calculate preprocessing signal X B The whitening matrix W of Signal X B The covariance matrix C B Shure breaks it down into: Among them, Q and Q ⊥ are highly semi-orthogonal matrices, and the column vectors of the two matrices span the signal subspace and the noise subspace respectively; Is a diagonal matrix containing n non-zero eigenvalues; the signal X B The whitening matrix is ​​expressed as: W=Σ -1 Q H S42, using the whitening matrix W to pre-process the signal X B To perform whitening: Z=WX B Where Z is the signal matrix after whitening processing.

6. The method for joint space-time suppression of radar additive composite interference based on blind source separation according to claim 5 is characterized in that: Step S5 is specifically as follows: The process of suppressing interference signals from the whitened signal using the bounded component analysis algorithm is expressed as: In the formula, u m is the mth column vector of the source signal extraction matrix U; the signal y=[y1,…,y L ] projected onto the hyperplane and recorded as a point set: Y={(Re(y l ),Im(y l ))l=1,…,L} Where Re(·) is the real part of the complex number, Im(·) is the imaginary part of the complex number, and L is the number of signal snapshots. The point set Y is a convex set, and its boundary is defined as the convex hull. The mathematical representation of the convex hull is: In the formula, is the convex hull of the convex set Y, Records the elements that make up the convex hull, conv(·) is the convex hull calculation; The perimeter of the convex hull is Contrast function of the nth iteration constructed from the perimeter of the convex hull It is expressed as: In the formula, To extract vector The L2 norm of Therefore, Δy i =u m Δz i , i=1,…,V, The gradient is: The step size of gradient descent is set to: Where eta is the algorithm accuracy; extract vector u m The update is expressed as: Here the new extraction vector Unitized; Extracted signal It is expressed as:

7. The method for joint space-time suppression of radar additive composite interference based on blind source separation according to claim 6 is characterized in that: Step S6 includes the following steps: S61, calculating the cross-correlation function between the separated signal and the original pulse signal: In the formula, R i is the cross-correlation function of the i-th signal, s is the reference pulse signal, is the separated i-th signal; S62. Take the maximum absolute value of the cross-correlation function as a measure of signal similarity: r i =max(R i |) In the formula, ρ i is the signal similarity measure; S63, selecting the signal with the greatest signal similarity as the echo signal: i0=argmax(ρ i ) Therefore, the echo signal 8. A radar additive composite interference space-time joint suppression system based on blind source separation, the system being used to execute the method according to any one of claims 1 to 7, characterized in that: The system includes the following modules: DOA estimation module: estimates the direction of arrival (DOA) of the interference signal based on the observed signal data; Pulse accumulation module: accumulates pulse signals to improve the signal-to-noise ratio; Blocking matrix preprocessing module: Use the DOA information of the interference signal to construct a blocking matrix, eliminate the interference signal energy in the accumulated signal through blocking matrix preprocessing, and achieve interference signal suppression; Whitening processing module: whitening the output signal after blocking matrix preprocessing; Blind source separation module: uses blind source separation algorithm to suppress the residual interference in the whitened signal; Signal selection module: selects the echo signal from the separated signals.

Images