阅读说明：本技术 一种基于特征波束的自适应波束域fsa的杂波抑制降维方法 (Clutter suppression dimension reduction method of self-adaptive beam domain FSA based on characteristic beam ) 是由 杨志伟 李婧雅 田敏 雷赫 黄帅 于 2019-07-31 设计创作，主要内容包括：本发明公开了一种基于特征波束的自适应波束域FSA的杂波抑制降维方法,包括：获取FSA数据矢量；获取特征空间；根据所述特征空间对所述FSA数据矢量进行波束域自适应降维处理得到降维后的数据矢量；对所述降维后的数据矢量进行杂波抑制和自适应滤波处理得到杂波抑制后的距离-多普勒域数据。本发明提供的杂波抑制降维方法具有一定的自适应性,在实现杂波抑制的同时,降低了自适应处理维数,降低了挑选样本的标准,减少了处理所需的均匀杂波样本数和运算量,提高了算法的实用性。(The invention discloses a clutter suppression and dimension reduction method of self-adaptive beam domain FSA based on characteristic beams, which comprises the following steps: obtaining an FSA data vector; acquiring a feature space; carrying out beam domain self-adaptive dimensionality reduction on the FSA data vector according to the feature space to obtain a dimensionality reduced data vector; and performing clutter suppression and self-adaptive filtering processing on the data vector after the dimension reduction to obtain the range-Doppler domain data after the clutter suppression. The clutter suppression dimension reduction method provided by the invention has certain self-adaptability, reduces the self-adaptive processing dimension while realizing clutter suppression, reduces the standard of selecting samples, reduces the number of uniform clutter samples and the operand required by processing, and improves the practicability of the algorithm.)

1. A clutter suppression dimension reduction method of self-adaptive beam domain FSA based on eigenbeams is characterized by comprising the following steps:

obtaining an FSA data vector;

acquiring a feature space;

carrying out beam domain self-adaptive dimensionality reduction on the FSA data vector according to the feature space to obtain a dimensionality reduced data vector;

and performing clutter suppression and self-adaptive filtering processing on the data vector after the dimension reduction to obtain the range-Doppler domain data after the clutter suppression.

2. The clutter suppression dimension reduction method according to claim 1, wherein said obtaining the FSA data vector comprises:

carrying out A/D sampling on N space equivalent array elements of the radar to obtain K +2 pulse echo data in one CPI; wherein the K +2 pulse echo data contain echo information of L samples;

performing time domain sliding window processing on the K +2 pulse data for three times in each space equivalent array element to obtain three groups of data consisting of 1-K, 2-K +1 and 3-K +2 pulses;

respectively carrying out K-point DFT on the three groups of data to convert the data to a Doppler domain to obtain an FSA data vector; for the same Doppler channel, the three groups of data respectively arrange L sample data of N space equivalent array elements into a data vector of NxL dimension.

3. The clutter suppression dimension reduction method according to claim 2, wherein the expression of the FSA data vector is:

x(k)＝[x₀(k)^T,x₁(k)^T,x₂(k)^T]^T；

where x (k) represents the FSA data vector, x₀(k) Data vector, x, corresponding to the unsliding window of data in the kth Doppler channel₁(k) Data vector, x, corresponding to a pulse representing data slip in the kth Doppler channel₂(k) Represents the data vector corresponding to two pulses of data sliding in the kth Doppler channel, (-)^TIndicating transposition.

4. The clutter suppression dimension reduction method according to claim 1, wherein the obtaining the feature space comprises:

respectively calculating covariance matrixes of non-sliding windows, sliding one pulse and sliding two pulse data in the kth Doppler channel to obtain three covariance matrixes;

and performing characteristic decomposition on the covariance matrix and extracting a limited number of characteristic beams to construct a characteristic space.

5. The clutter suppression dimension reduction method according to claim 4, wherein the covariance matrix is calculated by the following formula:

wherein R is₀(k) Covariance matrix, R, representing the correspondence of unsmooth windowed data₁(k) Representing the covariance matrix, R, corresponding to the sliding of one pulse data₂(k) Representing a covariance matrix corresponding to the sliding two pulse data, (-)^HRepresenting a conjugate transpose.

6. The clutter suppression dimension reduction method according to claim 4, wherein performing eigen decomposition on the covariance matrix and extracting a finite number of eigen-beam formation eigenspaces comprises:

respectively carrying out eigenvalue decomposition on the three covariance matrixes to obtain eigenvalues;

sorting the eigenvalues from big to small to respectively obtain the first M (M is more than or equal to 1 and less than or equal to N) big eigenvalues lambda_i,jCorresponding feature vector v_i,j(ii) a Wherein λ is_i,j(i-0, 1,2, j-1, 2.., M) represents the ith set of data covariance matrices R_i(k) J-th large eigenvalue, v_i,jRepresenting a characteristic value λ_i,jA corresponding eigenbeam;

according to the characteristic beam v_i,jA feature space of three sets of data is constructed.

7. The clutter suppression dimension reduction method according to claim 6, wherein the expression of the feature space is:

C₀(k)＝[v_0,1,...,v_0,M]；

C₁(k)＝[v_1,1,...,v_1,M]；

C₂(k)＝[v_2,1,...,v_2,M]；

wherein, C₀(k) Representing clutter feature space, C, corresponding to non-sliding window data₁(k) Representing the clutter feature space corresponding to a sliding pulse data, C₂(k) Watch (A)And displaying clutter feature space corresponding to the two pulse data.

8. The clutter suppression dimension reduction method according to claim 1, wherein performing beam-domain adaptive dimension reduction on the FSA data vector according to the feature space to obtain a dimension-reduced data vector comprises:

constructing a beam domain projection matrix of three groups of data according to the feature space;

obtaining a beam domain dimension reduction matrix according to the beam domain projection matrix;

and performing beam domain dimension reduction processing on the FSA data vector according to the beam domain dimension reduction matrix to obtain a dimension-reduced data vector.

9. The clutter suppression dimension reduction method according to claim 8, wherein constructing a beam domain projection matrix of three sets of data from the eigenspace comprises:

spatial steering s according to beam center_sAnd constructing a beam domain projection matrix of three groups of data by the characteristic space, wherein the expression of the beam projection matrix is as follows:

P₀(k)＝[s_s,C₀(k)]；

P₁(k)＝[s_s,C₁(k)]；

P₂(k)＝[s_s,C₂(k)]；

wherein, P₀(k) Representing the beam projection matrix, P, corresponding to the unsmooth window data₁(k) Representing the projection matrix of the beam corresponding to sliding a pulse of data, P₂(k) Representing the beam projection matrix for sliding two pulse data.

Technical Field

The invention belongs to the technical field of motion platform radars, and particularly relates to a clutter suppression and dimension reduction method of a self-adaptive beam domain FSA based on a characteristic beam.

Background

Radars are electronic devices that detect objects using electromagnetic waves. In order to effectively improve the aerial surveillance capability, a radar system is generally mounted on an airplane or a satellite. Due to the fact that the airborne/spaceborne radar platform has the characteristic of high-speed movement, when the radar works in a downward-looking state, a scene main lobe clutter Doppler spectrum is seriously widened, a weak slow target is submerged in the clutter, and direct detection cannot be achieved. Therefore, the premise of utilizing the airborne/spaceborne radar to realize the moving target detection task is to suppress strong clutter in a scene, so that an originally submerged weak moving target is highlighted, the signal-to-noise-and-noise ratio (SCNR) of the target is improved, and the detection probability is improved. Due to the fact that a clutter spectrum under a motion platform has a space-time coupling characteristic, a target is difficult to distinguish from a clutter only through one-dimensional space domain or time domain filtering, the clutter suppression is required to be carried out on a space-time two-dimensional plane, and meanwhile, due to the fact that the clutter changes along with the environment and the terrain, a self-adaptive processing mode is generally adopted.

Brennan, Reed et al, in the document "Theory of Adaptive radar" (IEEE Transactions on Adaptive and Electronic Systems, 1973, 9(2)), extend the one-dimensional spatial filtering technique into the Space-Time two-dimensional domain, and propose a Space-Time Adaptive Processing (STAP) method, i.e. a full-dimensional Space-Time Adaptive Processing method; the filter can form clutter suppression notches at space-time two-dimensional clutter distribution positions, clutter is effectively suppressed, and the Minimum Detectable Velocity (MDV) of a target is reduced. Brennan et al, in The document of comprehensive of space-time adaptive processing using an experimental airborne radar data (The Record of 1993IEEE National Radarconference, Mass., USA, 1993: 176-.

However, the full-dimensional space-time adaptive processing method utilizes all space equivalent array elements and time domain pulse sampling, so that the system dimension is large, the operation complexity is high, the requirement on training samples is high, and the method is difficult to realize in actual processing. When actual measurement data in engineering are processed, the system dimension is increased along with the increase of the number of spatial channels, and the processing method still has larger operation amount during processing.

Disclosure of Invention

In order to solve the problems in the prior art, the invention provides a clutter suppression and dimension reduction method of an adaptive beam domain FSA based on a characteristic beam. The technical problem to be solved by the invention is realized by the following technical scheme:

a clutter suppression dimension reduction method of self-adaptive beam domain FSA based on eigenbeams comprises the following steps: obtaining an FSA data vector;

acquiring a feature space;

carrying out beam domain self-adaptive dimensionality reduction on the FSA data vector according to the feature space to obtain a dimensionality reduced data vector;

and performing clutter suppression and self-adaptive filtering processing on the data vector after dimension reduction to obtain clutter suppressed distance-Doppler domain data.

In one embodiment of the present invention, the acquiring the FSA data vector comprises:

carrying out A/D sampling on N space equivalent array elements of the radar to obtain K +2 pulse echo data in one CPI; wherein the K +2 pulse echo data comprise echo information of L samples;

performing time domain sliding window processing on the K +2 pulse data for three times in each space equivalent array element to obtain three groups of data consisting of 1-K, 2-K +1 and 3-K +2 pulses;

respectively carrying out K-point DFT on the three groups of data to convert the data to a Doppler domain to obtain an FSA data vector; for the same Doppler channel, the three groups of data respectively arrange L sample data of N space equivalent array elements into a data vector of NxL dimension.

In one embodiment of the present invention, the expression of the FSA data vector is:

x(k)＝[x₀(k)^T,x₁(k)^T,x₂(k)^T]^T；

where x (k) represents the FSA data vector, x₀(k) Data vector, x, corresponding to the data unsliding window in the kth Doppler channel₁(k) Data vector, x, corresponding to a pulse representing data sliding in the kth Doppler channel₂(k) Data vectors corresponding to two pulses representing data slip in the kth Doppler channel, (. C)^TIndicating transposition.

In one embodiment of the present invention, the acquiring the feature space includes:

respectively calculating covariance matrixes of non-sliding windows, sliding one pulse and sliding two pulse data in the kth Doppler channel to obtain three covariance matrixes;

and performing characteristic decomposition on the covariance matrix and extracting a limited number of characteristic beams to construct a characteristic space.

In one embodiment of the present invention, the covariance matrix is calculated by the following formula:

wherein R is₀(k) Covariance matrix, R, representing the correspondence of unsmooth windowed data₁(k) Representing a covariance matrix, R, corresponding to sliding a pulse data₂(k) Representing a covariance matrix corresponding to the sliding two pulse data, (-)^HRepresenting a conjugate transpose.

In one embodiment of the present invention, performing eigen decomposition on the covariance matrix and extracting a limited number of eigenbeam configuration eigenspaces comprises:

respectively carrying out eigenvalue decomposition on the three covariance matrixes to obtain eigenvalues;

sorting the eigenvalues from big to small to respectively obtain the first M (M is more than or equal to 1 and less than or equal to N) big eigenvalues lambda_i,jCorresponding feature vector v_i,j(ii) a Wherein λ is_i,j(i-0, 1,2, j-1, 2.., M) represents the ith set of data covariance matrices R_i(k) J-th large eigenvalue, v_i,jRepresenting a characteristic value λ_i,jA corresponding eigenbeam;

according to the characteristic beam v_i,jA feature space of three sets of data is constructed.

In one embodiment of the present invention, the expression of the feature space is:

C₀(k)＝[v_0,1,...,v_0,M]；

C₁(k)＝[v_1,1,...,v_1,M]；

C₂(k)＝[v_2,1,...,v_2,M]；

wherein, C₀(k) Representing clutter feature space, C, corresponding to non-sliding window data₁(k) Representing the clutter feature space corresponding to a sliding pulse data, C₂(k) Representing a clutter feature space corresponding to the sliding two-pulse data.

In an embodiment of the present invention, performing beam domain adaptive dimension reduction on the FSA data vector according to the feature space to obtain a dimension-reduced data vector includes:

constructing a beam domain projection matrix of three groups of data according to the feature space;

obtaining a beam domain dimension reduction matrix according to the beam domain projection matrix;

and carrying out beam domain dimension reduction processing on the FSA data vector according to the beam domain dimension reduction matrix to obtain a dimension-reduced data vector.

In one embodiment of the present invention, constructing a beam domain projection matrix of three sets of data from the eigenspace comprises:

spatial steering s according to beam center_sAnd the beam domain projection moments of the three groups of data are constructed by the characteristic spaceAn array, the expression of the beam projection matrix being:

P₀(k)＝[s_s,C₀(k)]；

P₁(k)＝[s_s,C₁(k)]；

P₂(k)＝[s_s,C₂(k)]；

wherein, P₀(k) Representing the beam projection matrix, P, corresponding to the unsmooth window data₁(k) Representing the projection matrix of the beam corresponding to sliding a pulse of data, P₂(k) Representing the beam projection matrix for sliding two pulse data.

The invention has the beneficial effects that:

1. the clutter suppression dimension reduction method provided by the invention greatly reduces the system dimension of the processor while suppressing clutter, reduces the operation amount, saves the operation resource and has better processing efficiency;

2. the clutter suppression dimension reduction method provided by the invention reduces the number of training samples required when constructing the covariance matrix due to the reduction of the processor dimension, reduces the standard of selecting samples, and has stronger practicability;

3. the dimension reduction matrix in the clutter suppression dimension reduction method provided by the invention is formed by the characteristic wave beam after the characteristic decomposition of the original data covariance matrix, and the problem that the dimension reduction matrix in the prior art does not depend on data is solved, so that the method provided by the invention has certain self-adaptability.

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

Drawings

Fig. 1 is a schematic flowchart of a clutter suppression dimension reduction method for adaptive beam space domain FSA based on eigenbeams according to an embodiment of the present invention;

FIG. 2 is a schematic flow chart of another clutter suppression dimension reduction method according to an embodiment of the present invention;

FIG. 3 is a graph of computational complexity as a function of number of spatial channels for conventional FSA processing and methods of the present invention provided by embodiments of the present invention;

FIGS. 4 a-4 b are graphs of target output signal-to-noise-ratio (SCNR) processed by a conventional FSA and the method of the present invention according to an embodiment of the present invention;

FIG. 5 is a graph of output SCNR as a function of sample number after the spur suppression process provided by an embodiment of the present invention.

Detailed Description

The present invention will be described in further detail with reference to specific examples, but the embodiments of the present invention are not limited thereto.