Range-velocity two-dimensional super-resolution method for radar swarm targets in strong ground clutter environment

Abstract

The present invention belongs to the technical field of radar detection and tracking of swarm UAV targets, and specifically relates to a two-dimensional super-resolution method for distance and speed of radar swarm targets in a strong ground clutter environment, which suppresses emission leakage and ground clutter by fast time dimension bandpass filtering and slow time dimension MTI filtering, expands the number of samples by sliding rearrangement of radar echoes, and finally obtains a two-dimensional super-resolution capability of distance and speed that breaks through the Rayleigh limit by a super-resolution digital signal processing method. The present invention can perform clutter suppression and two-dimensional super-resolution processing of distance and speed on LFMCW radar swarm UAV target echoes under a strong ground clutter background, thereby obtaining a two-dimensional super-resolution capability of distance and speed that breaks through the Rayleigh limit under the condition that the radar instantaneous bandwidth resources and accumulation time resources are limited.

Description

Technical Field

[0001] The invention belongs to the technical field of radar detection and tracking of swarm UAV targets, and specifically relates to a two-dimensional super-resolution method for radar swarm targets in a strong ground clutter environment. The method can suppress clutter and perform two-dimensional super-resolution processing on LFMCW radar swarm UAV target echoes under a strong ground clutter background, thereby obtaining a two-dimensional super-resolution capability of distance and speed that breaks through the Rayleigh limit when the radar instantaneous bandwidth resources and accumulation time resources are limited. Background Art

[0002] Because continuous wave radar has the advantages of strong clutter suppression capability, no close-range blind spots, strong close-range small target detection capability and high distance-speed two-dimensional resolution capability, it has certain applications in fields such as low-altitude target monitoring.

[0003] In recent years, swarm drones have brought new challenges to modern radar precision detection. Swarm drones have the characteristics of a large number of targets, staggered positions, small radar cross-section (RCS), and strong maneuverability. It is very difficult to use radar detection to obtain information such as the characteristics, number, and individual spatial trajectory of the group targets. Actual radar equipment needs to monitor and track all targets of interest within its coverage area, and its bandwidth resources and accumulation time resources are very precious and limited. This makes the range resolution and velocity resolution of radars using the classic range-velocity two-dimensional matched filtering method limited by the radar signal bandwidth and accumulation time, which seriously limits the radar's ability to resolve swarm drones.

[0004] With the rapid development of digital signal processing capabilities, the resolution can be significantly improved through super-resolution digital signal processing methods. At present, the research results of super-resolution are mainly applied to the fields of imaging and angle super-resolution, but rarely involved in the field of swarm target detection. The current super-resolution algorithm has obvious shortcomings in the following two aspects: 1) The research only focuses on the super-resolution algorithm itself, and does not consider the adaptability to target environments such as strong ground clutter; 2) The research is based on the assumption that the sample data is sufficient. When the number of samples is small, the estimation accuracy of the sampling covariance matrix will be significantly deteriorated. Swarm targets are typical low, slow and small targets, and radars are bound to face extremely strong ground clutter; in addition, the relative positions between individuals in the swarm are dynamic and changeable, and the movement process is difficult to reproduce. In addition, the bandwidth and accumulation time resources allocated by the radar to a single swarm are limited, so the number of samples is relatively limited. The above two problems have brought difficulties to the application of super-resolution algorithms in the field of swarm detection. Summary of the invention

[0005] The present invention proposes a two-dimensional super-resolution method for range-velocity of LFMCW radar swarm UAV targets under strong ground clutter background. Fast time-dimensional bandpass filtering and slow time-dimensional MTI filtering are used to suppress emission leakage and ground clutter. The number of samples is expanded by sliding rearrangement of radar echoes. Finally, a two-dimensional super-resolution capability of range-velocity that breaks through the Rayleigh limit is obtained through a super-resolution digital signal processing method.

[0006] The problem to be solved by the present invention is achieved in this way:

[0007] A two-dimensional super-resolution method for range-velocity of radar swarm targets in a strong ground clutter environment comprises the following steps:

[0008] Step 1: Under the guidance of the search radar, the LFMCW tracking radar generates and transmits the LFMCW waveform, and uses the analog receiving channel to perform orthogonal mixing and de-skewing processing on the received echo signal, and then performs ADC sampling, and performs motion compensation on the digital echo signal according to the guidance information of the search radar;

[0009] Step 2: According to the search radar guidance information and the prior target parameter information determined by the radar design requirements, a fast time dimension difference bandpass filter and a slow time dimension Doppler MTI filter are designed, and the motion compensated signal is subjected to fast time dimension bandpass filtering and slow time dimension MTI filtering respectively. The steady-state response output after two-dimensional filtering is recorded as s(m,n);

[0010] Step 3: Use a two-dimensional rectangular window to slide s(m,n) two-dimensionally at a set time interval to obtain multiple two-dimensional sub-data blocks, and rearrange all the two-dimensional sub-data blocks uniformly to obtain a data matrix X;

[0011] Step 4: Perform singular value decomposition on the data matrix X to obtain a signal subspace matrix S consisting of eigenvectors and a diagonal matrix Λ consisting of corresponding positive singular values;

[0012] Step 5: Define the filter coefficient vector a and calculate the projection a of a on the signal subspace matrix S s =S H a, and further calculate the two-dimensional super-resolution output power Among them, ||·|| 2 Represents the sum of the squares of the 2-norms of vectors.

[0013] Furthermore, in step 1, the motion compensated signal is expressed as:

[0014]

[0015] In the formula, m is the pulse number, n is the fast time dimension sampling number, K is the number of individuals in the drone swarm, and A is i is the complex amplitude of the echo from the i-th UAV in the swarm, τ0i is the echo delay of the i-th UAV at time 0, Δt is the fast time dimension sampling interval, and f di is the Doppler frequency of the i-th UAV relative to the radar, T r is the pulse interval of the transmitted waveform, is the fixed phase associated with the i-th UAV.

[0016] Furthermore, in step 2, the passband range of the fast time-dimensional beat bandpass filter is f bmin ~f bmax ,and The purpose is to suppress transmission leakage, clutter and interference signals outside the swarm distance range; the order of the slow-time Doppler MTI filter does not exceed 1 / 5 of the total number of pulses, and the purpose is to suppress stationary ground clutter and ultra-low-speed clutter within the swarm distance range; where τ 0i is the echo delay of the i-th UAV at time 0.

[0017] Furthermore, the step 3 specifically includes the following process:

[0018] Use a two-dimensional rectangular window to slide s(m,n) in two dimensions, where the size of the two-dimensional rectangular window matrix is ​​P=M1×N1, M1 is the slow time dimension of the two-dimensional rectangular window, N1 is the fast time dimension of the two-dimensional rectangular window, and the step length of the two-dimensional rectangular window sliding along the two dimensions is 1 data value, and a total of L=L1×L2 sub-data blocks are slid out, L1=M-M1+1, L2=N-N1+1, N is the number of steady-state response output points in the fast time dimension, and M is the number of steady-state response output points in the slow time dimension;

[0019] The sub-data block matrix of size M1×N1 obtained by sliding through l1 data points along the slow time dimension and l2 data points along the fast time dimension is recorded as:

[0020]

[0021] Wherein, M1>M / 2, N1>N / 2, l1=0,1,...,L1-1, l2=0,1,...,L2-1;

[0022] Will Arrange them in columns to form a column vector of length P, denoted as but The expression is:

[0023]

[0024] Then L First arrange along l1, then arrange along l2, and finally arrange into a data matrix X of size P×L. The expression of X is:

[0025]

[0026] Furthermore, the expression for defining the filter coefficient vector a in step 5 is:

[0027]

[0028] In the formula, is the beat frequency search value, is the Doppler frequency search value, represents the Kronecker product.

[0029] Compared with the background technology, the present invention has the following advantages:

[0030] (1) Compared with the swarm detection radar using the classic range-velocity two-dimensional matched filtering method, the method of the present invention can significantly improve the range resolution and velocity resolution;

[0031] (2) The super-resolution method of the present invention is applicable to strong ground clutter environments;

[0032] (3) The super-resolution method described in the present invention can be applied to scenarios with a relatively limited number of samples. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 is a flow chart of an embodiment of the present invention. DETAILED DESCRIPTION

[0034] Now combined with the attached Figure 1 The present invention is described in further detail.

[0035] A two-dimensional super-resolution method for range-velocity of radar swarm targets in a strong ground clutter environment, characterized by comprising the following steps:

[0036] Step 1: Under the guidance of the search radar, the LFMCW tracking radar intelligently generates and transmits the LFMCW waveform, uses the analog receiving channel to perform orthogonal mixing and de-skewing processing on the received echo signal, and then performs ADC sampling. The digital echo signal is motion compensated according to the guidance information of the search radar. The signal after motion compensation can be approximately expressed as

[0037]

[0038] In the formula, m is the pulse number, n is the fast time dimension sampling number, K is the number of individuals in the drone swarm, and A is i is the complex amplitude of the echo from the i-th UAV in the swarm, τ 0i is the echo delay of the i-th UAV at time 0, Δt is the fast time dimension sampling interval, and f di is the Doppler frequency of the i-th UAV relative to the radar, Tr is the pulse interval of the transmitted waveform, is the fixed phase associated with the i-th UAV;

[0039] Step 2: According to the search radar guidance information and the prior target parameter information determined by the radar design requirements, the fast time dimension beat bandpass filter and the slow time dimension Doppler MTI filter are designed online in real time; wherein the passband range of the fast time dimension beat bandpass filter is f bmin ~f bmax ,and Its purpose is to suppress transmission leakage, clutter and interference signals outside the swarm distance range; the order of the slow-time Doppler MTI filter is recommended to be no more than 1 / 5 of the total number of pulses, and its purpose is to suppress stationary ground clutter and ultra-low-speed clutter within the swarm distance range;

[0040] Perform fast-time bandpass filtering and slow-time MTI filtering on p(m,n) respectively, and record the steady-state response output after two-dimensional filtering as s(m,n), where the number of fast-time steady-state response output points is N, and the number of slow-time steady-state response output points is M;

[0041] Step 3: Use a two-dimensional rectangular window of size P = M1 (slow time dimension) × N1 (fast time dimension) to slide s(m,n) in two dimensions, where the step length of the two-dimensional rectangular window sliding along the two dimensions is 1 data value, and a total of L = L1 × L2 (L1 = M-M1+1, L2 = N-N1+1) sub-data blocks can be slid out; the sub-data block matrix of size M1 × N1 obtained after sliding through l1 data points along the slow time dimension and l2 data points along the fast time dimension is recorded as

[0042]

[0043] Where, l1=0,1,...,L1-1,l2=0,1,...,L2-1;

[0044] Will Arrange them in columns to form a column vector of length P, denoted as but The expression is

[0045]

[0046] In the formula, T represents transpose;

[0047] Then L First arrange along l1, then arrange along l2, and finally arrange into a data matrix X of size P×L, where L<P, and the expression of X is

[0048]

[0049] Step 4: Perform singular value decomposition on X to obtain the signal subspace matrix S composed of L eigenvectors and the corresponding diagonal matrix Λ composed of L positive singular values. Then the covariance matrix R of X can be expressed as

[0050] R=XX H =SΛ 2 S H

[0051] In the formula, H represents conjugate transpose;

[0052] Step 5: Define the filter coefficient vector a, which is expressed as

[0053]

[0054] In the formula, is the beat frequency search value, is the Doppler frequency search value, represents the Kronecker product;

[0055] Calculate the projection a of a on the signal subspace matrix S s =S H a, the two-dimensional super-resolution output power can be further calculated as

[0056]

[0057] In the formula, ||·|| 2 Represents the sum of the squares of the 2-norms of vectors.

Claims

1. A two-dimensional super-resolution method for range and velocity of radar swarm targets in a strong ground clutter environment, characterized in that: The following steps are involved: Step 1: Under the guidance of the search radar, the LFMCW tracking radar generates and transmits the LFMCW waveform, and uses the analog receiving channel to perform orthogonal mixing and de-skewing processing on the received echo signal, and then performs ADC sampling, and performs motion compensation on the digital echo signal according to the guidance information of the search radar; Step 2: According to the search radar guidance information and the prior target parameter information determined by the radar design requirements, a fast time dimension beat bandpass filter and a slow time dimension Doppler MTI filter are designed, and the motion compensated signal is subjected to fast time dimension bandpass filtering and slow time dimension MTI filtering respectively. The steady-state response output after two-dimensional filtering is recorded as s(m,n); Step 3: Use a two-dimensional rectangular window to slide s(m,n) two-dimensionally at a set time interval to obtain multiple two-dimensional sub-data blocks, and uniformly rearrange all the two-dimensional sub-data blocks to obtain a data matrix X; Step 4: Perform singular value decomposition on the data matrix X to obtain a signal subspace matrix S consisting of eigenvectors and a diagonal matrix Λ consisting of corresponding positive singular values; Step 5: Define the filter coefficient vector a and calculate the projection a of a on the signal subspace matrix S s =S H a, and further calculate the two-dimensional super-resolution output power Among them, ||·|| 2 Represents the square of the vector 2 norm; Wherein, in step 1, the motion compensated signal is expressed as: In the formula, m is the pulse number, n is the fast time dimension sampling number, K is the number of individuals in the drone swarm, and A is i is the complex amplitude of the echo from the i-th UAV in the swarm, τ 0i is the echo delay of the i-th UAV at time 0, Δt is the fast time dimension sampling interval, and f di is the Doppler frequency of the i-th UAV relative to the radar, T r is the pulse interval of the transmitted waveform, is the fixed phase associated with the i-th UAV; Wherein, in step 2, the passband range of the fast time dimension beat bandpass filter is f bmin ~f bmax ,and The purpose is to suppress transmission leakage, clutter and interference signals outside the swarm distance range; the order of the slow-time Doppler MTI filter does not exceed 1 / 5 of the total number of pulses, and the purpose is to suppress stationary ground clutter and ultra-low-speed clutter within the swarm distance range; where τ 0i is the echo delay of the i-th UAV at time 0; Wherein, the step 3 specifically includes the following process: Use a two-dimensional rectangular window to slide s(m,n) in two dimensions, where the size of the two-dimensional rectangular window matrix is ​​P=M1×N1, M1 is the slow time dimension of the two-dimensional rectangular window, N1 is the fast time dimension of the two-dimensional rectangular window, and the step length of the two-dimensional rectangular window sliding along the two dimensions is 1 data value, and a total of L=L1×L2 sub-data blocks are slid out, L1=M-M1+1, L2=N-N1+1, N is the number of steady-state response output points in the fast time dimension, and M is the number of steady-state response output points in the slow time dimension; The sub-data block matrix of size M1×N1 obtained by sliding through l1 data points along the slow time dimension and l2 data points along the fast time dimension is recorded as: Wherein, M1>M / 2, N1>N / 2, l1=0,1,...,L1-1, l2=0,1,...,L2-1; Will Arrange them in columns to form a column vector of length P, denoted as but The expression is: Then L First arrange along l1, then arrange along l2, and finally arrange into a data matrix X of size P×L. The expression of X is: Among them, the expression for defining the filter coefficient vector a in step 5 is: In the formula, is the beat frequency search value, is the Doppler frequency search value, represents the Kronecker product.

Images