A Mainlobe Interference Suppression Method Based on Projection Matrix Preprocessing
The Taylor estimation and sparse reconstruction method suppresses main lobe interference in radar signal processing, solves the problems of incorrect main lobe direction and high computing burden in the prior art, and achieves efficient main lobe interference suppression and secondary lobe level reduction .
Patent Information
- Application Number
- CN202211062722.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-08-31
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2042-08-31
AI Technical Summary
The prior art is difficult to keep the main lobe direction correct and not offset when suppressing the main lobe interference in the radar signal, while increasing the computing burden and it is difficult to effectively suppress the side lobe interference.
The Taylor estimation method is used to obtain a covariance matrix with higher accuracy, perform eigenvalue decomposition and feature projection preprocessing, reconstruct the covariance matrix using sparse reconstruction method, and obtain the optimal weight vector of the signal within the main lobe range to ensure that The weight vector projection remains unchanged, thereby achieving effective suppression of main lobe interference.
The calculation amount is reduced, ensuring the correct direction of the main lobe and the secondary lobe level is reduced, which significantly improves the calculation efficiency and effectively suppresses side lobe interference while suppressing the main lobe interference.
Smart Images

Figure CN115327490B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of radar signal processing, and in particular to a main lobe interference suppression method for projection matrix preprocessing. Background Art
[0002] Adaptive beamforming technology is widely used in radar, communication and sonar. The adaptive beamforming method can show good characteristics when the interference is in the sidelobe area. However, when the interference is in the mainlobe area, the conventional broadband array adaptive beamforming algorithm has main beam peak offset, high sidelobe and beam distortion, which greatly affects the performance of the algorithm.
[0003] In recent years, a large number of literatures and studies have shown that the method based on blocking matrix (BMP) and eigenprojection matrix pre-processing (EMP) applied to broadband arrays can effectively suppress mainlobe interference in the frequency domain, but it is still difficult to achieve accurate mainlobe conformity and mainlobe offset. Therefore, it is increasingly important to achieve mainlobe conformity and mainlobe non-offset under the premise of strengthening the suppression of mainlobe interference and sidelobe interference.
[0004] Li Shuai et al. published a method for suppressing main lobe interference by using constant beamforming in their paper “Main lobe interference suppression method using broadband constant beamforming” (2016(06):21-24.DOI:10.19297 / j.cnki.41-1228 / tj.2016.06.005.). This method can ensure that the width of the main lobe of the radiation pattern remains unchanged and can effectively suppress the main lobe interference. However, it cannot suppress the side lobe interference and introduces second-order cone programming to obtain the weight coefficient, thereby increasing the computational burden.
[0005] Hu Haitao from the National University of Defense Technology proposed a method for reconstructing the covariance matrix based on the EMP algorithm in his paper "Research on Radar Anti-Main Lobe Interference Algorithm" (2019.DOI:10.27052 / d.cnki.gzjgu.2019.000710.). Step 1: First obtain the covariance matrix of the received signal, perform eigenvalue decomposition on it and construct the eigenprojection matrix. Step 2: Preprocess the received signal with the eigenprojection matrix. Step 3: Reconstruct the covariance matrix of the preprocessed signal. Finally, adaptive beamforming is performed. However, the entire algorithm has limitations in the selection of scalar parameters for the correlation between the eigenvector and the steering vector, which will cause the final beam to be offset. Summary of the invention
[0006] In view of the deficiencies in the prior art, the present invention aims to provide a main lobe interference suppression method for projection matrix preprocessing, which reduces the amount of calculation and ensures that the main lobe direction in the radiation pattern is correct while reducing the side lobe level.
[0007] In order to achieve the above object, the technical solution adopted by the present invention is as follows:
[0008] A main lobe interference suppression method for projection matrix preprocessing, the method comprising the following steps:
[0009] S1 uses Taylor estimation method to obtain a more accurate covariance matrix for the received signal model;
[0010] S2 performs eigenvalue decomposition on the covariance matrix of the received signal;
[0011] S3 constructs a feature projection matrix according to the decomposed eigenvalues and eigenvectors, and performs feature projection preprocessing on the received signal;
[0012] S4 reconstructs the covariance matrix using the sparse reconstruction method;
[0013] S5 obtains the optimal weight vector of the signal under the premise that the weight vector projection remains unchanged within the main lobe range; finally, the preprocessed signal and the weight are multiplied to obtain the final result.
[0014] It should be noted that the covariance matrix with higher accuracy in step S1 is:
[0015]
[0016] in, is the covariance matrix of the received signal, and the A matrix is the steering vector a(θ i ) is a matrix composed of A = [a(θ1),…,a(θ m )], m is the sum of the main lobe and side lobe interference numbers; and for the suppression of an interference in the main lobe, h i is the power of the ith interference, is the noise power.
[0017] It should be noted that the covariance matrix with higher accuracy is decomposed in step S2 to obtain:
[0018]
[0019] Among them, λ i for The eigenvalues of u, i = 1, 2, ... m. i is the eigenvector corresponding to the i-th eigenvalue. g is the interference subspace U g The eigenvalue diagonal matrix, Λ n is the noise subspace U n Eigenvalue diagonal matrix, p is the number of array elements.
[0020] It should be noted that, in step S3, it is assumed that um is the eigenvector corresponding to the main lobe interference, the eigenprojection matrix
[0021]
[0022] Where I is the identity matrix, |u m H a(θ0)| 2 ≥ρ|a(θ0)| 2 , ρ is a scalar parameter that limits the correlation between the eigenvector and the steering vector; u that satisfies the above formula m That is, it is the characteristic vector corresponding to the main lobe interference, otherwise it is the characteristic vector of the side lobe interference;
[0023] The feature projection matrix preprocesses the original signal x(k), and the processed signal is:
[0024] X2(k)=Bx(k)
[0025] The signal after the above processing is reconstructed through step S4.
[0026] It should be noted that in step S4, the covariance matrix of step S3 is reconstructed, specifically: from X2(k) of step S3, it can be known that
[0027]
[0028] R y Perform eigenvalue decomposition to obtain:
[0029]
[0030] Among them, λ i For R y eigenvalues, i = 1, 2, ... m, and λ1 ≥ λ2 ≥ ... λ i-1 ,u i is the eigenvector corresponding to the i-th eigenvalue;
[0031] At the eigenvalue λ i Select the eigenvalue λ corresponding to the noise i-d , i-d+1 , …λ i-1 , d is the number of eigenvalues corresponding to the noise, expression:
[0032]
[0033] According to the number of noise corresponding eigenvalues d, calculate the noise eigenvalue λ i-d , i-d+1 , …λ i-1 The average
[0034]
[0035] Using noise mean Correct the noise eigenvalue and set the noise eigenvalue λ i-d , i-d+1 , …λ i-1 Equal to the noise mean Get the corrected signal covariance matrix The expression is:
[0036]
[0037] in, is the diagonal matrix composed of the corrected eigenvalues, U is the eigenvector matrix, Y=[y1,y2,…y P-1 ].
[0038] It should be noted that step S5 includes:
[0039] S5.1 Calculate the optimal weight vector W under the minimum variance criterion under the constraint that the projection of the main lobe subspace remains unchanged, according to the formula:
[0040] H = ∫ φ a(θ)a(θ) H dθ
[0041]
[0042] Determine the main lobe region subspace, which is the column vector of U1. i is the i-th eigenvalue, v i is the corresponding eigenvector, Λ1 is the diagonal matrix composed of the first K2 eigenvalues, U1 is the matrix composed of the eigenvectors corresponding to the eigenvalues in Λ1, Λ2 is composed of P-K2 eigenvalues, and U2 is the matrix composed of the eigenvectors corresponding to the eigenvalues in Λ2;
[0043] Construct the main lobe subspace projection matrix E according to the main lobe subspace U1;
[0044] E=U(U H U) -1 U H
[0045] With the projection of the weight vector in the main lobe subspace unchanged as the constraint condition, the modified signal covariance matrix obtained in step 4 is Calculate the optimal weight vector under the condition of minimum variance
[0046]
[0047] Where a(θ) is the steering vector in the desired direction θ
[0048] Finally solve the formula:
[0049]
[0050] S5.2 Signal weighting
[0051] The preprocessed signal obtained in step S3 is multiplied by the weight obtained in step S5.1:
[0052] Z=W H X=W H BX.
[0053] The beneficial effects of the present invention are:
[0054] 1. Compared with the traditional broadband constant beam width main lobe suppression method, the present invention does not need to perform covariance matrix inversion to avoid optimization problems such as second-order cone programming in the process of solving the optimal weights, thereby effectively shortening the operation time and significantly improving the calculation efficiency;
[0055] 2. Compared with the method of suppressing main lobe interference by the EMP algorithm, the weight vector projection of the present invention is determined in the main lobe subspace, which can not only achieve the main lobe shape preservation but also ensure that the main lobe peak direction is correct and the main lobe width remains unchanged. BRIEF DESCRIPTION OF THE DRAWINGS
[0056] Figure 1 It is a flow chart of the implementation of the present invention;
[0057] Figure 2 This is a comparison diagram of the main lobe suppression method of the present invention and the traditional broadband constant beam when the main lobe interference is 2°;
[0058] Figure 3 This is a comparison diagram between the present invention and the method for suppressing main lobe interference using the EMP algorithm when the main lobe interference is at 2°. DETAILED DESCRIPTION
[0059] The present invention will be further described below in conjunction with the accompanying drawings. It should be noted that this embodiment is based on the technical solution and provides a detailed implementation method and a specific operation process, but the protection scope of the present invention is not limited to this embodiment.
[0060] like Figure 1 As shown, the present invention is a main lobe interference suppression method for projection matrix preprocessing, and the method comprises the following steps:
[0061] S1 uses Taylor estimation method to obtain a more accurate covariance matrix for the received signal model;
[0062] S2 performs eigenvalue decomposition on the covariance matrix of the received signal;
[0063] S3 constructs a feature projection matrix according to the decomposed eigenvalues and eigenvectors, and performs feature projection preprocessing on the received signal;
[0064] S4 reconstructs the covariance matrix using the sparse reconstruction method;
[0065] S5 obtains the optimal weight vector of the signal under the premise that the weight vector projection remains unchanged within the main lobe range; finally, the preprocessed signal and the weight are multiplied to obtain the final result.
[0066] It should be noted that the covariance matrix with higher accuracy in step S1 is:
[0067]
[0068] in, is the covariance matrix of the received signal, and the A matrix is the steering vector a(θ i ) is a matrix composed of A = [a(θ1),…,a(θ m )], m is the sum of the main lobe and side lobe interference numbers; and for the suppression of an interference in the main lobe, h i is the power of the ith interference, is the noise power.
[0069] It should be noted that the covariance matrix with higher accuracy is decomposed in step S2 to obtain:
[0070]
[0071] Among them, λ i for The eigenvalues of u, i = 1, 2, ... m. i is the eigenvector corresponding to the i-th eigenvalue. g is the interference subspace U g The eigenvalue diagonal matrix, Λ n is the noise subspace U n Eigenvalue diagonal matrix, p is the number of array elements.
[0072] It should be noted that, in step S3, it is assumed that u m is the eigenvector corresponding to the main lobe interference, the eigenprojection matrix
[0073]
[0074] Where I is the identity matrix, |u m H a(θ0)| 2 ≥ρ|a(θ0)| 2 , ρ is a scalar parameter that limits the correlation between the eigenvector and the steering vector; u that satisfies the above formulam That is, it is the characteristic vector corresponding to the main lobe interference, otherwise it is the characteristic vector of the side lobe interference;
[0075] The feature projection matrix preprocesses the original signal x(k), and the processed signal is:
[0076] X2(k)=Bx(k)
[0077] The signal after the above processing is reconstructed through step S4.
[0078] It should be noted that in step S4, the covariance matrix of step S3 is reconstructed, specifically: from X2(k) of step S3, it can be known that
[0079]
[0080] R y Perform eigenvalue decomposition to obtain:
[0081]
[0082] Among them, λ i For R y eigenvalues, i = 1, 2, ... m, and λ1 ≥ λ2 ≥ ... λ i-1 ,u i is the eigenvector corresponding to the i-th eigenvalue;
[0083] At the eigenvalue λ i Select the eigenvalue λ corresponding to the noise i-d , i-d+1 , …λ i-1 , d is the number of eigenvalues corresponding to the noise, expression:
[0084]
[0085] According to the number of noise corresponding eigenvalues d, calculate the noise eigenvalue λ i-d , i-d+1 , …λ i-1 The average
[0086]
[0087] Using noise mean Correct the noise eigenvalue and set the noise eigenvalue λ i-d , i-d+1 , …λ i-1 Equal to the noise mean Get the corrected signal covariance matrix The expression is:
[0088]
[0089] in, is the diagonal matrix composed of the corrected eigenvalues, U is the eigenvector matrix, Y=[y1,y2,…y P-1 ].
[0090] It should be noted that step S5 includes:
[0091] S5.1 Calculate the optimal weight vector W under the minimum variance criterion under the constraint that the projection of the main lobe subspace remains unchanged, according to the formula:
[0092] H = ∫ φ a(θ)a(θ) H dθ
[0093]
[0094] Determine the main lobe region subspace, which is the column vector of U1. i is the i-th eigenvalue, v i is the corresponding eigenvector, Λ1 is the diagonal matrix composed of the first K2 eigenvalues, U1 is the matrix composed of the eigenvectors corresponding to the eigenvalues in Λ1, Λ2 is composed of P-K2 eigenvalues, and U2 is the matrix composed of the eigenvectors corresponding to the eigenvalues in Λ2;
[0095] Construct the main lobe subspace projection matrix E according to the main lobe subspace U1;
[0096] E=U(U H U) -1 U H
[0097] With the projection of the weight vector in the main lobe subspace unchanged as the constraint condition, the modified signal covariance matrix obtained in step 4 is Calculate the optimal weight vector under the condition of minimum variance
[0098]
[0099] Where a(θ) is the steering vector in the desired direction θ
[0100] Finally solve the formula:
[0101]
[0102] S5.2 Signal weighting
[0103] The preprocessed signal obtained in step S3 is multiplied by the weight obtained in step S5.1:
[0104] Z=W HX=W H BX.
[0105] Example
[0106] In order to make the purpose, technical solutions and advantages of the present invention more clearly described, they will be verified through the following simulation experiments.
[0107] The simulation parameters are shown in Table 1:
[0108] parameter Parameter Value parameter Parameter Value Number of array elements 16 Signal-to-Noise Ratio 0dB Element spacing Half wavelength Sampling frequency 120MHz Beam pointing 0° Drying ratio 60dB Center frequency 1GHz Main lobe interference direction 2° bandwidth 50MHz Pulse repetition period 200us Pulse Width 50us Number of dwell cycles 5 Target direction 0° Noise Type Gaussian white noise
[0109] Simulation content
[0110] Simulation 1: Under the simulation conditions in Table 1, the antenna radiation patterns of the main lobe suppression method of the present invention and the broadband constant beam forming method are compared. The results are as follows: Figure 2 shown.
[0111] Figure 2 The blue line in the middle represents the antenna pattern obtained by the main lobe suppression method of broadband constant beam forming, and the red line represents the antenna pattern obtained by the present invention. Figure 2 It can be seen that the antenna pattern of the present invention has a significant suppression effect on the sidelobe level, which is basically maintained above 10dB. In terms of the amount of calculation, since the present invention does not require matrix inversion, the calculation time is also greatly reduced. The original technology takes 9.193 seconds to calculate, and the present invention takes 0.368824 seconds to calculate.
[0112] Simulation 2: Under the simulation conditions in Table 1, the antenna radiation patterns of the present invention and the main lobe suppression method based on the EMP algorithm are compared. The results are as follows: Figure 3 shown.
[0113] Figure 3 The dotted line represents the antenna pattern obtained by the main lobe suppression method based on the EMP algorithm, and the red line represents the antenna pattern obtained by the present invention. The main lobe of the main lobe interference suppression based on the EMP algorithm deviates approximately by about 3°. The main lobe pointing of the present invention meets the simulation conditions and optimizes the sidelobe level suppression by about 10dB. Although the original algorithm also uses projection matrix preprocessing, the present invention uses the weight vector projection unchanged in the main lobe subspace based on the weight constraint. As a result, our main lobe shape is not deformed and the pointing is more accurate.
[0114] For those skilled in the art, various other corresponding changes and modifications can be made according to the technical solutions and concepts described above, and all of these changes and modifications should fall within the protection scope of the claims of the present invention.
Claims
1. A main lobe interference suppression method using projection matrix preprocessing, characterized in that: The method comprises the following steps: S1 uses Taylor estimation method to obtain a more accurate covariance matrix for the received signal model; S2 performs eigenvalue decomposition on the covariance matrix of the received signal with higher accuracy; S3 constructs a feature projection matrix according to the decomposed eigenvalues and eigenvectors, and performs feature projection preprocessing on the received signal; S4 reconstructs the covariance matrix of the signal preprocessed in step 3 using a sparse reconstruction method to obtain a modified covariance matrix; S5. Under the premise that the weight vector projection remains unchanged within the main lobe range, the optimal weight vector of the signal is obtained according to the modified covariance matrix obtained in step 4; finally, the preprocessed signal and the weight are multiplied to obtain the final result; The step S5 comprises: S5.1 Computing the optimal weight vector under the minimum variance criterion under the constraint that the projection of the main lobe subspace remains unchanged , according to the formula: ; ; Determine the main lobe area subspace, which is The column vector of is the i-th eigenvalue, is the corresponding eigenvector, For the front The diagonal matrix of eigenvalues, for The matrix of eigenvectors corresponding to the eigenvalues in , Depend on The eigenvalues are composed of for The matrix consisting of the eigenvectors corresponding to the eigenvalues in ; According to the main lobe subspace Construct the main lobe subspace projection matrix ; ; With the projection of the weight vector in the main lobe subspace unchanged as the constraint condition, the modified signal covariance matrix obtained in step 4 is , calculate the optimal weight vector under the minimum variance condition ; in for Direction of the steering vector; Finally solve the formula: ; Among them, a q is the steering vector in the desired direction; S5.2 Signal weighting The preprocessed signal obtained in step S3 is multiplied by the weight obtained in step S5.1: 。 2. The main lobe interference suppression method of projection matrix preprocessing according to claim 1, characterized in that: The covariance matrix with higher accuracy in step S1 is: ; in, is the covariance matrix of the received signal, and the A matrix is the steering vector The matrix composed of , is the direction of interference, m is the sum of the main lobe and side lobe interference; and for the suppression of an interference in the main lobe, , is the power of the ith interference, is the noise power.
3. The main lobe interference suppression method of projection matrix preprocessing according to claim 1, characterized in that: The result obtained by decomposing the covariance matrix with higher accuracy in step S2 is: ; in, for The eigenvalues of , i=1,2,...m, is the eigenvector corresponding to the i-th eigenvalue, The interference subspace The eigenvalue diagonal matrix, is the noise subspace Eigenvalue diagonal matrix, p is the number of array elements.
4. The main lobe interference suppression method of projection matrix preprocessing according to claim 1, characterized in that: In step S3, it is assumed that is the eigenvector corresponding to the main lobe interference, the eigenprojection matrix ; in, is the identity matrix, , is a scalar parameter that limits the correlation between the eigenvector and the steering vector; That is, it is the characteristic vector corresponding to the main lobe interference, otherwise it is the characteristic vector of the side lobe interference; The steering vector in the direction of the first disturbance; The feature projection matrix is the original signal Preprocessing is performed, and the processed signal is: ; The signal after the above processing is reconstructed through step S4.
5. The main lobe interference suppression method of projection matrix preprocessing according to claim 4, characterized in that: In step S4, the covariance matrix of step S3 is reconstructed, specifically: Knowable ; in, is the number of training samples, Perform eigenvalue decomposition to obtain: ; in, for The eigenvalues of , i = 1, 2, ... m, and , is the eigenvector corresponding to the i-th eigenvalue; In the eigenvalue Select the eigenvalue corresponding to the noise , d is the number of eigenvalues corresponding to the noise, expression: ; Calculate the noise eigenvalue according to the number of eigenvalues corresponding to the noise d The average , ; Using the noise mean Correct the noise characteristic value and make the noise characteristic value Equal to the noise mean , and obtain the corrected signal covariance matrix The expression is: ; in, is the diagonal matrix composed of the corrected eigenvalues, ; .
Citation Information
Patent Citations
Self-adaption wave beam forming method of broadband array under main lobe jamming
CN109490850A
Main lobe interference suppression algorithm based on feature projection preprocessing and covariance matrix sparse reconstruction
CN109959899A
Array antenna main lobe interference suppression method and system
CN110196410A
Main lobe interference suppression method based on characteristic oblique projection covariance matrix reconstruction
CN113406576A
Distributed radar main lobe interference suppression method based on EMP-CMR algorithm
CN113917420A
