CN106970358B - An Optimal Method for Angular Doppler Registration of Clutter Spectrum of Non-Side-Looking Array Radar - Google Patents

Abstract

The invention belongs to the technical field of radars, and discloses an optimization method for angular Doppler registration of clutter spectrum of a non-front side-view array radar, which comprises the following steps: obtaining clutter data of each range gate received by the non-front side view array radar, determining a space Doppler frequency position where a maximum power point of each training sample unit is located, determining a clutter spectrum center, calculating a correction matrix corresponding to the clutter data of each training sample unit by taking the Doppler frequency and the space frequency of a distance unit to be detected as references, and calculating a corrected clutter covariance matrix of the distance unit to be detected, so that a filtering weight vector is obtained, and clutter suppression is performed on the clutter data of the non-front side view array radar, so that clutter of the non-front side view array radar is more uniform after the clutter spectrum processing.

Description

An Optimal Method for Angular Doppler Registration of Clutter Spectrum of Non-Side-Looking Array Radar

technical field

The invention belongs to the field of radar technology, and in particular relates to an optimization method for angle Doppler registration of a non-side-looking array radar clutter spectrum, which is suitable for ideally performing radar clutter spectrum when errors exist in known radar configuration parameters. compensation.

Background technique

Airborne radar has the functions of air warning, reconnaissance, control and guided weapons, etc., and plays an important role in my country's national defense construction. When the airborne radar is looking down, due to the wide distribution and high intensity of the ground clutter, the moving target is completely covered and cannot be identified. How to effectively suppress ground clutter is an important content of airborne radar signal processing. Space-time adaptive processing (STAP) method is an effective means to suppress ground clutter, compensate for clutter spectrum broadening caused by platform motion, and detect ground slow targets. The clutter covariance matrix (Clutter CovarianceMatrix, CCM) is the decisive factor of the clutter suppression performance of the space-time adaptive processing method. In practical applications, the clutter covariance matrix of the unit to be detected is estimated by the sample covariance inversion (Sample MatrixInversion, SMI) method by using the adjacent training sample units that are independent and identically distributed (IID) with the unit to be detected In order to achieve adaptive clutter suppression, in order to make the output signal-to-noise ratio loss of the space-time filter less than 3dB, the number of independent and identically distributed training samples used to estimate the clutter covariance matrix must be greater than two degrees of freedom of the system times. For non-side-looking arrays, the ground clutter Doppler frequency has a serious distance dependence, especially in short-range conditions, the distance dependence is more significant, so that the clutter distribution does not meet the independent and identical distribution conditions, and it is impossible to use training samples to Accurately estimating the clutter covariance matrix of the unit to be detected degrades the clutter suppression performance of the space-time adaptive processing method.

At present, there are many methods for compensating distance dependence, including Doppler Warping (Doppler Warping, DW) method, Angle Doppler Compensation (Angle Doppler Compensation, ADC) method, Adaptive Angle Doppler Compensation (Adaptive Angle Doppler Compensation, A2DC) method and registration based compensation (RegistrationBased Compensation, RBC) and other methods. Among them, the DW method essentially compensates the clutter Doppler frequency difference between the training sample and the detection range unit caused by different pitch angles, so that the clutter in the training sample is approximately stable. In addition to compensating for the Doppler difference between the training sample and the range unit to be detected, the ADC method and the A2DC method will also compensate for the spatial frequency difference caused by different pitch angles. The RBC method extracts the peak value of radar clutter spectrum by using time-domain smoother snapshots, which can realize the complete compensation of radar clutter, and can obtain good radar clutter suppression performance under ideal conditions.

However, among the above compensation methods, the DW method only compensates the center of the main lobe clutter spectrum in the Doppler domain, and the compensation effect obtained is not particularly ideal. The ADC method registers the clutter characteristics of each training sample unit and the sample unit to be tested through compensation in the space angle domain and the Doppler domain, and calculates the clutter conversion matrix, thereby suppressing the clutter distance dependence to a certain extent , for the main lobe clutter, the compensation effect of this method is more obvious. However, in practical applications, the clutter transformation matrix of each training sample needs to be based on the flight configuration parameters (such as pitch angle, azimuth angle, yaw angle and speed, etc.) Therefore, the error of any configuration parameters will have a certain impact on the performance of the clutter distance-dependent compensation of the method. Although the A2DC method does not need to know the configuration parameters, and the radar clutter suppression performance is also very good under ideal conditions, in practice, the covariance matrix of the radar clutter in each range unit is extremely unstable, resulting in a large drop in radar clutter suppression performance. The RBC method uses time-domain smoother snapshots to extract the peak value of the radar clutter spectrum, so that the estimation accuracy of the peak value of the radar clutter spectrum and the robustness of the RBC method will affect the performance of the STAP technology.

Contents of the invention

For the problems existing in the above-mentioned prior art, the object of the present invention is to provide a kind of optimization method of the angle Doppler registration of the non-side-looking array radar clutter spectrum, so that after the non-side-looking array radar clutter spectrum is processed, its clutter more uniform.

The invention proposes an optimization method for angle Doppler registration of non-side-looking array radar clutter spectrum. This method estimates the clutter spectrum of the non-front-side-looking array radar through the adaptive iterative method (IAA, Iterative Adaptive Approach), obtains the approximate value of the clutter power of each training sample, and determines the Doppler frequency position of the maximum power point of each training sample and the spatial frequency position, determine the center of the clutter spectrum, and then combine the Angle Doppler Compensation (ADC, Angle Doppler Compensation) method to perform two-dimensional compensation in the air domain and Doppler domain, and obtain the training consistent with the clutter statistical characteristics of the unit to be detected The sample data makes the clutter of the non-side-view array radar more uniform after processing the clutter spectrum, and then improves the clutter performance of the space-time adaptive processing (STAP) technology in the application of the non-side-view array radar.

In order to achieve the above object, the present invention adopts the following technical solutions to achieve.

A method for optimizing the angle Doppler registration of a non-side-looking array radar clutter spectrum, said method comprising the steps of:

Step 1. Obtain the clutter data of L distance units received by the non-frontal side-view array radar, and set the kth distance unit as the distance unit to be detected, then the other L-1 distance units except the distance unit to be detected The unit is the training sample unit; L represents the total number of distance units contained in the clutter data received by the non-side-looking array radar;

Step 2, setting the normalized spatial frequency interval [-1, 1] of the non-frontal side-view array radar and the normalized Doppler frequency interval [-1, 1] of the non-frontal side-view array radar ; And the normalized spatial frequency interval of the non-frontal side-view array radar and the normalized Doppler frequency interval of the non-frontal side-view array radar constitute the space-time two-dimensional plane of the non-frontal side-view array radar;

Step 3, let l=1, l∈{1, 2, ..., L}, l≠k, l represents the lth training sample unit;

Step 4: Sampling the clutter data of the lth training sample unit on the space-time two-dimensional plane of the non-side-view array radar to obtain K _s ×K _t sampling points; K _s represents the lth training The number of sampling points of the clutter data of the sample unit in the normalized spatial frequency interval of the non-side-view array radar, K _t represents the normalization of the clutter data of the lth training sample unit in the non-side-view array radar The number of sampling points in the Doppler frequency interval;

Step 5, for the (p, q)th sampling point of the clutter data of the lth training sample unit on the space-time two-dimensional plane of the non-side-view array radar, obtain the (p, q)th sampling point The space-time steering vector at the point, p∈{1, 2,..., K _t }, q∈{1, 2,..., K _s }; so that the clutter data of the lth training sample unit is on the non-positive side Space-time steering vectors at all K _s ×K _t sampling points on the space-time two-dimensional plane of line-of-sight radar;

Step 6, according to the clutter data of the lth training sample unit and the space-time steering vector at the (p, q)th sampling point, calculate the initial power at the (p, q)th sampling point p ∈ {1, 2, ..., K _t }, q ∈ {1, 2, ..., K _s }; and determine the clutter of the clutter data of the l-th training sample unit according to the initial power at each sampling point power spectrum P ⁽⁰⁾ ;

And set the initial value of the number of iterations n to 1;

Step 7, according to the power at the (p, q)th sampling point And the space-time steering vector at the (p, q)th sampling point constructs the covariance matrix at the (p, q)th sampling point p ∈ {1, 2, ..., K _t }, q ∈ {1, 2, ..., K _s };

Step 8, according to the clutter data of the lth training sample unit, the space-time steering vector at the (p, q) sampling point and the covariance matrix at the (p, q) sampling point Calculate the power at the (p,q)th sample point after the nth iteration p ∈ {1, 2, ..., K _t }, q ∈ {1, 2, ..., K _s }; and according to the power at each sampling point after the nth iteration Determine the clutter power spectrum P ⁽ⁿ⁾ after the nth iteration of the clutter data of the l training sample unit;

Step 9, if And 0 _< n < num, then add 1 to the value of n, repeat steps 7 and 8; num represents the maximum value of the set iteration number, express request norm;

like Or n>num, then stop the iteration, and the power at each sampling point after the nth iteration As the final power at each sampling point in the clutter data of the lth training sample unit;

Step 10, determine the maximum value of the final power at each sampling point in the clutter data of the lth training sample unit, and obtain the normalized Doppler frequency and normalized spatial frequency at the sampling point corresponding to the maximum value , which are respectively used as the normalized Doppler frequency and the normalized spatial frequency corresponding to the power spectrum of the clutter data of the lth training sample unit;

Step 11, determine the normalized Doppler frequency and the normalized spatial frequency corresponding to the clutter power spectrum of the clutter data of the distance unit to be detected; Normalized Doppler frequency and normalized spatial frequency, normalized Doppler frequency and normalized spatial frequency corresponding to the clutter power spectrum of the clutter data of the distance unit to be detected, calculate the lth training sample unit Corrected Doppler frequency and corrected spatial frequency corresponding to power spectrum of clutter data;

Step 12, according to the corrected Doppler frequency and the corrected spatial frequency corresponding to the power spectrum of the clutter data of the lth training sample unit, calculate the corresponding Doppler domain of the power spectrum of the clutter data of the lth training sample unit transformation matrix and spatial transformation matrix, and then according to the power spectrum of the clutter data of the lth training sample unit corresponding to the Doppler domain transformation matrix and the spatial domain transformation matrix, the power spectrum correspondence of the clutter data of the lth training sample unit is calculated correction matrix;

According to the clutter data of the lth training sample unit and the corresponding correction matrix of the power spectrum of the clutter data of the lth training sample unit, the clutter data of the lth training sample unit after correction is obtained;

In step 13, add 1 to the value of l, and repeat steps 4 to 12, so as to obtain L-1 corrected clutter data, and then calculate the to-be-detected clutter data based on L-1 corrected clutter data The covariance matrix of the clutter data for the range cell;

Obtain the space-time steering vector of the distance unit to be detected, thereby obtain the filter weight vector according to the space-time steering vector of the distance unit to be detected and the covariance matrix of the clutter data of the distance unit to be detected, and obtain the filter weight vector according to the filter weight vector Angle Doppler registration is performed on the clutter data of the L range units respectively.

Beneficial effects of the present invention: the method of the present invention utilizes the IAA method to estimate the clutter spectrum of the non-side-looking array radar in the case of a single snapshot, obtains the clutter power approximate value of each training sample, determines its clutter spectrum center, and then combines the ADC method to carry out The two-dimensional compensation in the airspace and Doppler domain makes the clutter of the non-side-looking array radar more uniform after processing the clutter spectrum, and then improves the space-time adaptive processing (STAP) technology in the clutter of the non-side-looking array radar. performance.

Description of drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings that need to be used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only These are some embodiments of the present invention. Those skilled in the art can also obtain other drawings based on these drawings without creative work.

Fig. 1 is a schematic flow chart of an optimization method for angular Doppler registration of a non-side-looking array radar clutter spectrum provided by an embodiment of the present invention;

Figure 2 is a schematic diagram of the optimal clutter spectrum under the non-side-view array radar;

Figure 3 is a schematic diagram of the uncompensated clutter spectrum obtained under the non-side-looking array radar;

Fig. 4 is a schematic diagram of the clutter spectrum obtained by using the existing A2DC method under the non-side-view array radar;

Fig. 5 is the clutter spectrum schematic diagram that uses the method of the present invention to obtain under the non-side-looking array radar;

Fig. 6 is a schematic diagram of comparison of clutter suppression improvement factors of non-frontal side-looking array radars in the four cases shown in Fig. 2 , Fig. 3 , Fig. 4 , and Fig. 5 .

Detailed ways

The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some, not all, embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the protection scope of the present invention.

An embodiment of the present invention provides an optimization method for angular Doppler registration of a non-side-looking array radar clutter spectrum, as shown in FIG. 1 , the method includes the following steps:

Step 1. Obtain the clutter data of L distance units received by the non-frontal side-view array radar, and set the kth distance unit as the distance unit to be detected, then the other L-1 distance units except the distance unit to be detected The unit is the training sample unit; L represents the total number of distance units contained in the clutter data received by the non-side-looking array radar.

Generally, when the clutter data of L range units received by the non-side-view array radar contains target data, the range unit where the target is located is taken as the range unit to be detected.

In step 1, the clutter data of L range units received by the non-frontal side-view array radar are obtained, specifically:

The antenna of the non-front-side-view array radar is set to be a linear array composed of N array elements uniformly distributed, and the non-front-side-view array radar transmits M pulses within a coherent processing interval;

Sampling on the distance is carried out L times on each pulse and each array element channel respectively, then the actually received clutter data is three-dimensional data of N×M×L; the nth array of the lth distance unit The clutter data received by the m-th pulse of the element is denoted as x _nml , and the clutter data received by the n-th array element of the l-th distance unit is denoted as x _nl =[x _n1l , x _n2l ,..., x _nMl ] ^T , arrange the clutter data received by the N array elements of the l-th distance unit into an NM×1-dimensional column vector X _l , so as to obtain the clutter data Xl=[x _1l ^T of the l-th distance unit , x _2l ^T ,…,x _Nl ^T ] ^T , where l∈{1, 2,…, L}.

Generally, the number of selected training sample units is greater than 2NM.

Step 2, setting the normalized spatial frequency interval [-1, 1] of the non-frontal side-view array radar and the normalized Doppler frequency interval [-1, 1] of the non-frontal side-view array radar ; and the normalized spatial frequency interval of the non-frontal side-view array radar and the normalized Doppler frequency interval of the non-frontal side-view array radar constitute a space-time two-dimensional plane of the non-frontal side-view array radar.

The clutter data of a certain distance unit is the data on a two-dimensional plane constructed by the number of array elements and the number of pulses, which is the most original acquired echo data; the space-time two-dimensional plane is the angle after normalization , Doppler frequency to construct a two-dimensional plane, the angle and Doppler frequency are calculated based on known configuration parameters.

Step 3, let l=1, l∈{1, 2, ..., L}, l≠k, l represents the lth training sample unit.

Step 4: Sampling the clutter data of the lth training sample unit on the space-time two-dimensional plane of the non-side-view array radar to obtain K _s ×K _t sampling points; K _s represents the lth training The number of sampling points of the clutter data of the sample unit in the normalized spatial frequency interval of the non-side-view array radar, K _t represents the normalization of the clutter data of the lth training sample unit in the non-side-view array radar The number of sampling points in the Doppler frequency range.

In step 4, K _s =10N, k _t =10M;

Among them, N represents the number of array elements contained in the antenna of the non-side-view array radar, M represents the number of pulses transmitted by the non-side-view array radar in a coherent processing interval, K _s represents the miscellaneous value of the lth training sample unit The number of sampling points of the wave data in the normalized spatial frequency interval of the non-side-view array radar, K _t represents the normalized Doppler frequency of the clutter data of the lth training sample unit in the non-side-view array radar The number of sampling points in the interval.

In order to reconstruct the covariance matrix more accurately, here the airspace of the space-time plane is divided into K _s grid points, and the Doppler domain is divided into K _t grid points, then the corresponding normalization of each grid point The normalized spatial frequency can be expressed as f _{s, q} , q ∈ {1, 2, ..., K _s }, and the normalized Doppler frequency can be expressed as f _{d, p} , p ∈ {1, 2, ..., K _t }, according to the Cramereau lower bound theory, increasing the number of grid distributions indefinitely will not significantly improve the performance of spectral estimation. Therefore, K _s =10N, k _t =10M are selected here, and the space-time plane can be divided into K=K _s K _t grid points, that is, 100NM grid points, so the clutter data of the current training sample can be represented by the angle Doppler data at the grid sampling point. The normalized spatial frequency interval [-1, 1] is uniformly sampled to obtain K _s sampling points; the normalized Doppler frequency interval [-1, 1] is uniformly sampled to obtain K _t sampling points.

Step 5, for the (p, q)th sampling point of the clutter data of the lth training sample unit on the space-time two-dimensional plane of the non-side-view array radar, obtain the (p, q)th sampling point The space-time steering vector at the point, p∈{1, 2,..., K _t }, q∈{1, 2,..., K _s }; so that the clutter data of the lth training sample unit is on the non-positive side Space-time steering vectors at all K _s ×K _t sampling points on the space-time two-dimensional plane of line-of-sight radar.

In step 5, the space-time steering vector at the (p, q)th sampling point is obtained, specifically:

Obtain the steering vector of the (p, q)th sampling point on the normalized spatial frequency interval of the non-side-view array radar

Obtain the steering vector of the (p, q)th sampling point on the normalized Doppler frequency interval of the non-frontal side-view array radar

Thus, the space-time steering vector at the (p, q)th sampling point

Among them, f _{d, p} represent the normalized Doppler frequency of the pth sampling point on the normalized Doppler frequency interval of the non-frontal side-looking array radar, and f _{s, q} represent the normalized Doppler frequency of the non-frontal side-looking array radar. The normalized spatial frequency of the qth sampling point on the normalized spatial frequency interval; Indicates kronecker product, (·) ^T indicates transpose.

Step 6, according to the clutter data of the lth training sample unit and the space-time steering vector at the (p, q)th sampling point, calculate the initial power at the (p, q)th sampling point p ∈ {1, 2, ..., K _t }, q ∈ {1, 2, ..., K _s }; and determine the clutter of the clutter data of the l-th training sample unit according to the initial power at each sampling point power spectrum P ⁽⁰⁾ ;

And set the initial value of the number of iterations n to 1.

In step 6, use the following formula to calculate the initial power at the (p, q)th sampling point

Among them, S(f _{d, p} , f _{s, q} ) represents the space-time steering vector at the (p, q)th sampling point, X _l represents the clutter data of the l-th training sample unit, ( ) ^H represents Conjugate transpose, |·| means to take the absolute value operation;

According to the initial power at each sampling point, determine the clutter power spectrum P ⁽⁰⁾ of the clutter data of the lth training sample unit as diag(·) represents a diagonal matrix, that is, the diagonal elements are The diagonal matrix of .

Step 7, according to the power at the (p, q)th sampling point And the space-time steering vector at the (p, q)th sampling point constructs the covariance matrix at the (p, q)th sampling point p ∈ {1, 2, ..., K _t }, q ∈ {1, 2, ..., K _s }.

In step 7, according to the power at the (p, q)th sampling point And the space-time steering vector at the (p, q)th sampling point constructs the covariance matrix at the (p, q)th sampling point as follows:

Among them, S(f _{d, p} , f _{s, q} ) represents the space-time steering vector at the (p, q)th sampling point.

Step 8, according to the clutter data of the lth training sample unit, the space-time steering vector at the (p, q) sampling point and the covariance matrix at the (p, q) sampling point Calculate the power at the (p,q)th sample point after the nth iteration p ∈ {1, 2, ..., K _t }, q ∈ {1, 2, ..., K _s }; and according to the power at each sampling point after the nth iteration Determine the clutter power spectrum P ⁽ⁿ⁾ after the nth iteration of the clutter data of the lth training sample unit.

In step 8, according to the clutter data of the lth training sample unit, the space-time steering vector at the (p, q) sampling point and the covariance matrix at the (p, q) sampling point Calculate the power at the (p,q)th sample point after the nth iteration as follows:

Among them, S(f _{d, p} , f _{s, q} ) represents the space-time steering vector at the (p, q)th sampling point, X _l represents the clutter data of the l-th training sample unit, ( ) ^H represents Conjugate transpose, |·| means to take the absolute value operation, (·) ^-1 means matrix inversion;

According to the power at each sampling point after the nth iteration Determine the clutter power spectrum P ⁽ⁿ⁾ after the nth iteration of the clutter data of the lth training sample unit as follows:

Among them, diag( ) represents a diagonal matrix, that is, the diagonal elements are The diagonal matrix of .

Step 9, if And 0 _< n < num, then add 1 to the value of n, repeat steps 7 and 8; num represents the maximum value of the set iteration number, express request norm;

like Or n>num, then stop the iteration, and the power at each sampling point after the nth iteration As the final power at each sampling point in the clutter data of the lth training sample unit.

Step 10, determine the maximum value of the final power at each sampling point in the clutter data of the lth training sample unit, and obtain the normalized Doppler frequency and normalized spatial frequency at the sampling point corresponding to the maximum value , which are respectively taken as the normalized Doppler frequency and normalized spatial frequency corresponding to the power spectrum of the clutter data of the l-th training sample unit.

Step 11, determine the normalized Doppler frequency and the normalized spatial frequency corresponding to the clutter power spectrum of the clutter data of the distance unit to be detected; Normalized Doppler frequency and normalized spatial frequency, normalized Doppler frequency and normalized spatial frequency corresponding to the clutter power spectrum of the clutter data of the distance unit to be detected, calculate the lth training sample unit The power spectrum of the clutter data corresponds to the corrected Doppler frequency and the corrected spatial frequency.

In step 11: determine the normalized Doppler frequency _{fd, k} and the normalized spatial frequency fs _{, k} corresponding to the clutter power spectrum of the clutter data of the range unit to be detected:

Among them, λ is the working wavelength of the non-side-looking array radar, is the array element spacing of the uniform linear array, and the carrier aircraft flies along the x-axis at the speed v, the angle between the plane perpendicular to the xy plane where the uniform linear array is located and the x-axis is θ _k represents the pitch angle corresponding to the clutter data of the range unit to be detected, Indicates the azimuth angle corresponding to the clutter data of the distance unit to be detected relative to the uniform linear array, the number of array elements is N, and the pulse repetition frequency is f _r ;

Calculate the corrected Doppler frequency Δf _d,l =f _d,l -f _d,k and the corrected spatial frequency Δf _s,l =f _s,l -f corresponding to the power spectrum of the clutter data of the lth training sample unit _{s, k} ;

Among them, f _{d, l} represent the normalized Doppler frequency corresponding to the power spectrum of the clutter data of the l-th training sample unit, and f _{s, l} represent the corresponding frequency of the power spectrum of the clutter data of the l-th training sample unit Normalized spatial frequency, f _{d, k} represent the normalized Doppler frequency corresponding to the clutter power spectrum of the clutter data of the range unit to be detected, f _{s, k} represent the clutter of the clutter data of the range unit to be detected The normalized spatial frequency corresponding to the power spectrum.

Step 12, according to the corrected Doppler frequency and the corrected spatial frequency corresponding to the power spectrum of the clutter data of the lth training sample unit, calculate the corresponding Doppler domain of the power spectrum of the clutter data of the lth training sample unit transformation matrix and spatial transformation matrix, and then according to the power spectrum of the clutter data of the lth training sample unit corresponding to the Doppler domain transformation matrix and the spatial domain transformation matrix, the power spectrum correspondence of the clutter data of the lth training sample unit is calculated correction matrix;

According to the clutter data of the lth training sample unit and the corresponding correction matrix of the power spectrum of the clutter data of the lth training sample unit, the corrected clutter data of the lth training sample unit is obtained.

In step 12, the power spectrum of the clutter data of the lth training sample unit is calculated corresponding to the Doppler domain transformation matrix T _{d, l} and the space domain transformation matrix T _{s, l} :

Calculate the corresponding correction matrix T _IAA-ADC of the power spectrum of the clutter data of the lth training sample unit:

According to the clutter data of the lth training sample unit and the corresponding correction matrix of the power spectrum of the clutter data of the lth training sample unit, the corrected clutter data of the lth training sample unit is obtained

Among them, Δf _{d, l} represents the corrected Doppler frequency corresponding to the power spectrum of the clutter data of the l-th training sample unit, and Δf _{s, l} represents the correction space corresponding to the power spectrum of the clutter data of the l-th training sample unit frequency, Represents the kronecker product, N represents the number of array elements contained in the antenna of the non-side-view array radar, and M represents the number of pulses transmitted by the non-side-view array radar within a coherent processing interval.

In step 13, add 1 to the value of l, and repeat steps 4 to 12, so as to obtain L-1 corrected clutter data, and then calculate the to-be-detected clutter data based on L-1 corrected clutter data The covariance matrix of the clutter data for the range cell;

Obtain the space-time steering vector of the distance unit to be detected, thereby obtain the filter weight vector according to the space-time steering vector of the distance unit to be detected and the covariance matrix of the clutter data of the distance unit to be detected, and obtain the filter weight vector according to the filter weight vector Angle Doppler registration is performed on the clutter data of the L range units respectively.

In step 13, calculate the covariance matrix of the clutter data of the distance unit to be detected according to the L-1 corrected clutter data

in, Indicates the clutter data of the lth training sample unit after correction;

Acquire the space-time steering vector S(f _{d, k , f s, k ) of the distance unit to be detected, according to the space-time steering vector S(f d, k , f s, k ) of} the distance unit to be detected and the Covariance matrix of clutter data for range cells Find the filter weight vector which is

in, is the normalization constant.

Finally, the clutter data of the L range units received by the non-side-looking array radar are respectively suppressed through the filter weight vectors to obtain a uniform clutter spectrum.

What needs to be added is to calculate the improvement factor IF(f _dt ) of non-frontal side-view array radar clutter suppression, which is used to measure the angle Doppler compensation effect of non-frontal side-view array radar clutter spectrum.

Specifically, its expression is:

Among them, P _C represents the set non-side-looking array radar clutter input power, and _PN represents the set non-side-looking array radar noise input power.

The effects of the present invention can be further illustrated by the following simulation experiments.

(1) Description of simulation experiment data

When evaluating the suppression performance of non-side-view array radar clutter, in order to be able to compare with the traditional adaptive angle Doppler compensation method, the method simulation of the present invention adopts a uniform line array; select the appropriate repetition frequency, regardless of the distance For the fuzzy problem, the noise-to-noise ratio at the element level of the non-side-looking array radar is 60dB, and the No. 60 range unit is processed. In this part, the clutter power spectrum estimated by the sampling covariance inversion method without compensation, compensated by angle Doppler compensation method and compensated by the method of the present invention is compared with the optimal clutter power spectrum. The clutter power spectrum of different methods adopts Minimum Variance Distortionless Response (MVDR) spectrum. The simulation parameters of the non-front-side-looking array radar are shown in Table 1.

Table 1

(2) Simulation results and analysis

The simulation result of the present invention is as shown in Figure 2～Fig. 6; Wherein, Fig. 2 is the optimal clutter spectrum schematic diagram of No. 60 range unit under the non-frontal side-view array radar, and Fig. 3 is No. 60 distances under the non-frontal side-view array radar Schematic diagram of the clutter spectrum obtained by the unit without compensation. Figure 4 is a schematic diagram of the clutter spectrum obtained by using the A2DC method for the No. 60 range unit under the non-frontal side-looking array radar. Schematic diagram of the clutter spectrum obtained by the ADC; Figure 6 is a comparison chart of the clutter suppression improvement factor and the optimal clutter suppression improvement factor of the uncompensated, A2DC method, and IAA-ADC method.

Due to the insufficient number of samples in the non-uniform scene, the uncompensated radar clutter spectrum has high sidelobes and poor resolution; both the A2DC method and the IAA-ADC method can obtain a higher-resolution spectrum. But the covariance matrix of the radar clutter of each range unit is extremely unstable in the A2DC method in practice, causing the radar clutter suppression performance to drop a lot; The side-looking array radar clutter spectrum is more uniform after processing, and the performance of spectrum estimation is improved.

In summary, the simulation experiment has verified the correctness, effectiveness and reliability of the present invention.

Those of ordinary skill in the art can understand that all or part of the steps to realize the above method embodiments can be completed by hardware related to program instructions, and the aforementioned programs can be stored in computer-readable storage media. When the program is executed, the execution includes The steps of the above-mentioned method embodiments; and the aforementioned storage medium includes: ROM, RAM, magnetic disk or optical disk and other various media that can store program codes.

The above is only a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Anyone skilled in the art can easily think of changes or substitutions within the technical scope disclosed in the present invention. Should be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be determined by the protection scope of the claims.

Claims (10)

1. A method for optimizing the angular Doppler registration of the clutter spectrum of a non-forward side-looking array radar is characterized by comprising the following steps:

step 1, acquiring clutter data of L distance units received by a non-front side view array radar, setting a kth distance unit as a distance unit to be detected, and setting other L-1 distance units except the distance unit to be detected as training sample units; l represents the total number of distance units contained in clutter data received by the non-front side view array radar;

step 2, setting a normalized spatial frequency interval [ -1,1] of the non-obverse side view array radar and a normalized Doppler frequency interval [ -1,1] of the non-obverse side view array radar; the normalized space frequency interval of the non-front side view array radar and the normalized Doppler frequency interval of the non-front side view array radar form a space-time two-dimensional plane of the non-front side view array radar;

step 3, let L be 1, L be e {1,2, …, L }, L be not equal to k, and L represent the sequence number of the training sample unit;

step 4, sampling clutter data of the ith training sample unit on a space-time two-dimensional plane of the non-side view array radar to obtain K_s×K_tSampling points; k_sArray radar for expressing clutter data of I training sample unit in non-positive side viewNormalized spatial frequency interval of (1) number of sampling points, K_tThe number of sampling points of clutter data of the ith training sample unit in a normalized Doppler frequency interval of the non-side view array radar is represented;

step 5, for the (p, q) th sampling point of the clutter data of the ith training sample unit on the space-time two-dimensional plane of the non-positive side view array radar, obtaining the space-time steering vector at the (p, q) th sampling point, wherein p belongs to {1,2, …, K ∈ [ ]_t}，q∈{1,2,…,K_s}; thereby obtaining all K of clutter data of the first training sample unit on the space-time two-dimensional plane of the non-front side view array radar_s×K_tSpace-time steering vectors at each sampling point;

step 6, calculating the initial power at the (p, q) th sampling point according to the clutter data of the l training sample unit and the space-time guide vector at the (p, q) th sampling pointp∈{1,2,…,K_t}，q∈{1,2,…,K_s}; and determining a clutter power spectrum P of clutter data of the ith training sample unit according to the initial power at each sampling point⁽⁰⁾；

Setting the initial value of the iteration times i as 1;

step 7, according to the power at the (p, q) th sampling pointAnd the space-time steering vector at the (p, q) th sampling point constructs a covariance matrix at the (p, q) th sampling pointp∈{1,2,…,K_t}，q∈{1,2,…,K_s}；

Step 8, according to the clutter data of the ith training sample unit, the space-time guide vector at the (p, q) th sampling point and the covariance matrix at the (p, q) th sampling pointCalculating the power at the (p, q) th sampling point after the ith iterationp∈{1,2,…,K_t}，q∈{1,2,…,K_s}; and according to the power at the (p, q) th sampling point after the ith iterationDetermining clutter power spectrum P after ith iteration of clutter data of ith training sample unit⁽ⁱ⁾；

Step 9, ifAnd 0<i<num, adding 1 to the value of i, and repeatedly executing the step 7 and the step 8; num represents the maximum value of the set number of iterations,expression solutionA norm;

if it isOr i>num, stopping iteration and calculating the power at the (p, q) th sampling point after the ith iterationThe final power at the (p, q) th sampling point in the clutter data of the ith training sample unit;

step 10, determining the maximum value of the final power at each sampling point in the clutter data of the first training sample unit, acquiring the normalized doppler frequency and the normalized spatial frequency at the sampling point corresponding to the maximum value, and respectively using the normalized doppler frequency and the normalized spatial frequency as the normalized doppler frequency and the normalized spatial frequency corresponding to the power spectrum of the clutter data of the first training sample unit;

step 11, determining a normalized Doppler frequency and a normalized spatial frequency corresponding to a clutter power spectrum of clutter data of a distance unit to be detected; calculating a corrected Doppler frequency and a corrected spatial frequency corresponding to the power spectrum of the clutter data of the first training sample unit according to the normalized Doppler frequency and the normalized spatial frequency corresponding to the power spectrum of the clutter data of the first training sample unit and the normalized Doppler frequency and the normalized spatial frequency corresponding to the clutter power spectrum of the clutter data of the distance unit to be detected;

step 12, calculating a Doppler domain conversion matrix and a spatial domain conversion matrix corresponding to the power spectrum of the clutter data of the ith training sample unit according to the corrected Doppler frequency and the corrected spatial frequency corresponding to the power spectrum of the clutter data of the ith training sample unit, and further calculating a power spectrum corresponding correction matrix of the clutter data of the ith training sample unit according to the Doppler domain conversion matrix and the spatial domain conversion matrix corresponding to the power spectrum of the clutter data of the ith training sample unit;

obtaining clutter data of the first training sample unit after correction according to the clutter data of the first training sample unit and the power spectrum corresponding correction matrix of the clutter data of the first training sample unit;

step 13, adding 1 to the value of L, and repeatedly executing the steps 4 to 12 to obtain L-1 corrected clutter data, so as to calculate and obtain a covariance matrix of the clutter data of the distance unit to be detected according to the L-1 corrected clutter data;

and acquiring a space-time guide vector of the distance unit to be detected, so as to obtain a filtering weight vector according to the space-time guide vector of the distance unit to be detected and a covariance matrix of clutter data of the distance unit to be detected, and respectively carrying out angle Doppler registration on the clutter data of the L distance units according to the filtering weight vector to obtain a result after the clutter data of the L distance units are subjected to angle Doppler registration.

2. The method according to claim 1, wherein in step 1, clutter data of L range units received by the non-forward looking array radar is obtained, specifically:

setting an antenna of the non-front side view array radar to be a linear array consisting of N array elements which are uniformly distributed, wherein the non-front side view array radar transmits M pulses in a coherent processing interval;

sampling at a distance of L times on each pulse and each array element channel respectively, so that the actually received clutter data is three-dimensional data with dimensions of NxMxL; recording the clutter data received by the mth pulse of the nth array element of the z-th distance unit as x_nmzIf the array element is located in the z-th range unit, the clutter data received by the nth array element is marked as x_nz＝[x_n1z,x_n2z,…,x_nMz]^TArranging clutter data received by N array elements of the z-th distance unit into NM multiplied by 1 dimension column vector X_zSo as to obtain clutter data X of z-th distance unit_z＝[x_1z ^T,x_2z ^T,…,x_Nz ^T]^TWherein z ∈ {1,2, …, L }, (·)^TRepresenting a transpose; n-1, 2, …, N, M-1, 2, …, M; n represents the number of array elements contained in the antenna of the non-front side view array radar, and M represents the number of pulses transmitted by the non-front side view array radar in one coherent processing interval.

3. The method for optimizing the angular Doppler registration of the clutter spectrum of the non-forward looking array radar according to claim 1, wherein K is determined in step 4_s＝10N，K_t＝10M；

Wherein N represents the number of array elements contained in the antenna of the non-front side view array radar, M represents the number of pulses transmitted by the non-front side view array radar in a coherent processing interval, and K_sThe number of sampling points, K, of the clutter data of the ith training sample unit in the normalized space frequency interval of the non-side view array radar_tAnd the number of sampling points of the clutter data of the ith training sample unit in the normalized Doppler frequency interval of the non-positive side view array radar is represented.

4. The method according to claim 1, wherein in step 5, the space-time steering vector at the (p, q) th sampling point is obtained, specifically:

acquiring a steering vector of the (p, q) th sampling point on a normalized space frequency interval of the non-front side view array radar

Acquiring a steering vector of the (p, q) th sampling point on a normalized Doppler frequency interval of the non-normal side view array radar

Thus, the space-time steering vector at the (p, q) -th sampling point

Wherein f is_d,pIndicating the normalized Doppler frequency, f, of the p-th sampling point in the normalized Doppler frequency interval of the non-frontal view array radar_s,qThe normalized spatial frequency of the q sampling point on the normalized spatial frequency interval of the non-front side view array radar is represented;represents kronecker product (.)^TRepresenting a transpose; n represents the number of array elements contained in the antenna of the non-front side view array radar, and M represents the number of pulses transmitted by the non-front side view array radar in one coherent processing interval.

5. The method for optimizing the angular Doppler registration of the clutter spectrum of the non-forward looking array radar according to claim 1, wherein in step 6,

calculating the initial power at the (p, q) th sampling point by using the following formula

Wherein, S (f)_d,p,f_s,q) Representing the space-time steering vector at the (p, q) -th sampling point, X_lClutter data representing the ith training sample cell, (-)^HRepresenting conjugate transposition, |, representing absolute value operation; f. of_d,pIndicating the normalized Doppler frequency, f, of the p-th sampling point in the normalized Doppler frequency interval of the non-frontal view array radar_s,qThe normalized spatial frequency of the q sampling point on the normalized spatial frequency interval of the non-front side view array radar is represented;

determining a clutter power spectrum P of clutter data of the ith training sample unit according to the initial power at each sampling point⁽⁰⁾，diag (-) denotes a diagonal matrix, i.e. the diagonal element isThe diagonal matrix of (a).

6. The method for optimizing the angular Doppler registration of the clutter spectrum of the non-forward looking array radar according to claim 1, wherein in step 7,

according to the power at the (p, q) th sampling pointAnd the space-time steering vector at the (p, q) th sampling point constructs a covariance matrix at the (p, q) th sampling pointThe following were used:

wherein, S (f)_d,p,f_s,q) Representing the space-time steering vector at the (p, q) -th sampling point, f_d,pIndicating the normalized Doppler frequency, f, of the p-th sampling point in the normalized Doppler frequency interval of the non-frontal view array radar_s,qAnd the normalized spatial frequency of the q sampling point on the normalized spatial frequency interval of the non-front side view array radar is represented.

7. The method for optimizing the angular Doppler registration of the clutter spectrum of the non-forward looking array radar according to claim 1, wherein in step 8,

clutter data according to the ith training sample unit, a space-time steering vector at the (p, q) th sampling point and a covariance matrix at the (p, q) th sampling pointCalculating the power at the (p, q) th sampling point after the ith iterationThe following were used:

wherein, S (f)_d,p,f_s,q) Representing the space-time steering vector at the (p, q) -th sampling point, X_lClutter data representing the ith training sample cell, (-)^HRepresenting conjugate transpose, | representing absolute value operation, (-)^-1Representing matrix inversion; f. of_d,pIndicating the normalized Doppler frequency, f, of the p-th sampling point in the normalized Doppler frequency interval of the non-frontal view array radar_s,qThe normalized spatial frequency of the q sampling point on the normalized spatial frequency interval of the non-front side view array radar is represented;

according to the (p, q) th sampling point after the ith iterationOf (2) isDetermining clutter power spectrum P after ith iteration of clutter data of ith training sample unit⁽ⁱ⁾The following were used:

wherein diag (-) denotes a diagonal matrix, i.e. the diagonal elements areThe diagonal matrix of (a).

8. The method for optimizing the angular doppler registration of the clutter spectrum of the non-forward looking array radar according to claim 1, wherein in step 11:

determining a normalized Doppler frequency f corresponding to a clutter power spectrum of clutter data of a distance unit to be detected_d,kAnd normalized spatial frequency f_s,k：

Wherein, the lambda is the working wavelength of the non-front side view array radar,the array element spacing of the uniform linear arrays is determined, the carrier where the non-positive side-view array radar is located flies along the x axis at a speed v, the plane where the uniform linear arrays are located is vertical to the x-y plane, and the plane where the uniform linear arrays are located and the x axis form an included angleθ_kIndicating a range to be measuredThe pitch angle corresponding to the clutter data from the unit,indicating the azimuth angle f corresponding to the clutter data of the distance unit to be detected relative to the uniform linear array_rIs the pulse repetition frequency;

calculating the corrected Doppler frequency Deltaf corresponding to the power spectrum of the clutter data of the ith training sample unit_d,l＝f_d,l-f_d,kAnd correcting the spatial frequency Δ f_s,l＝f_s,l-f_s,k；

Wherein f is_d,lNormalized Doppler frequency, f, corresponding to the power spectrum of the clutter data representing the ith training sample unit_s,lNormalized spatial frequency, f, corresponding to the power spectrum of the clutter data representing the ith training sample unit_d,kNormalized Doppler frequency, f, corresponding to a clutter power spectrum of clutter data representing a range cell to be detected_s,kAnd representing the normalized spatial frequency corresponding to the clutter power spectrum of the clutter data of the distance unit to be detected.

9. The method for optimizing the angular Doppler registration of the clutter spectrum of the non-forward looking array radar according to claim 1, wherein in step 12,

calculating a power spectrum corresponding Doppler domain conversion matrix T of clutter data of the ith training sample unit_d,lAnd spatial domain transformation matrix T_s,l：

Calculating to obtain a power spectrum corresponding correction matrix T of clutter data of the first training sample unit_IAA-ADC：

Obtaining the clutter data of the first training sample unit after correction according to the clutter data of the first training sample unit and the power spectrum corresponding correction matrix of the clutter data of the first training sample unit

Wherein, X_lRepresents clutter data,. DELTA.f, of the ith training sample unit_d,lA modified Doppler frequency,. DELTA.f, corresponding to the power spectrum of the clutter data representing the ith training sample cell_s,lRepresents the modified spatial frequency corresponding to the power spectrum of the clutter data of the ith training sample unit,the kronecker product is expressed, N represents the number of array elements contained in the antenna of the non-front side view array radar, and M represents the number of pulses transmitted by the non-front side view array radar in one coherent processing interval.

10. The method for optimizing the angular doppler registration of the clutter spectrum of the non-forward looking array radar according to claim 1, wherein in step 13,

calculating to obtain a covariance matrix of clutter data of a distance unit to be detected according to the L-1 corrected clutter data

Wherein,representing clutter data of the corrected I training sample unit;

obtaining space-time guiding vector S (f) of distance unit to be detected_d,k,f_s,k) According to the space-time steering vector S (f) of the distance unit to be detected_d，k,f_s，k) To be provided withAnd covariance matrix of clutter data of distance unit to be detectedCalculating a filter weight vector of a distance unit to be detected

Wherein f is_d,kNormalized Doppler frequency, f, corresponding to a clutter power spectrum of clutter data representing a range cell to be detected_s,kA normalized spatial frequency corresponding to a clutter power spectrum of clutter data representing a range bin to be detected,is a normalization constant.