CN106125039B - Improvement space-time adaptive Monopulse estimation method based on local Combined Treatment - Google Patents

Abstract

The invention discloses an improved space-time self-adaptive monopulse angle measurement method based on local joint processing. Angle, iteratively update the space-time domain dimensionality reduction matrix of the JDL algorithm and the space-time steering vector of target detection, which solves the problems of fewer channels, low Doppler resolution and large angle measurement error in multi-channel airborne radar systems. The invention can accurately estimate the target space angle only by two iterations, and is easy for engineering implementation.

Description

Improved space-time adaptive monopulse angle measurement method based on local joint processing

technical field

The invention relates to the field of airborne radar monopulse angle measurement, in particular to an improved space-time self-adaptive monopulse angle measurement method based on local joint processing.

Background technique

When the airborne radar looks down, ground clutter seriously causes Doppler broadening, which makes moving targets easily submerged by clutter and affects radar target detection performance.

In 1973, Brennan et al proposed the use of space-time two-dimensional adaptive signal processing (STAP) to suppress clutter. STAP performs space-time two-dimensional filtering, selects training samples through the adjacent distance unit of the unit to be detected (CUT), and adaptively calculates the weight of the filter. A powerful tool for clutter.

Although the application of STAP technology can improve the performance of target detection, it cannot estimate the target angle. The adaptive monopulse technique proposed by Nickel is a high-precision angle estimation method, and this method can be extended to two dimensions in space and time. STAP realizes adaptive clutter suppression and coherent accumulation of moving target signals in space-time two-dimensional space, which can theoretically achieve optimal processing, but the amount of computation required for full-dimensional processing is astonishing, assuming that the number of samples in the air and time domains are N and K, the obtained adaptive weights need to estimate and invert the NK×NK dimensional clutter correlation matrix, and the calculation amount is O(NK) ³ , so there are huge difficulties in real-time processing both in software and hardware.

Dimensionality reduction STAP, through the linear transformation of full-dimensional data, reduces the solution of the problem to a low-dimensional space, which can reduce the degree of freedom of the system.

Contents of the invention

The technical problem to be solved by the present invention is to provide an improved space-time self-adaptive monopulse angle measurement method based on local joint processing for the defects involved in the background technology, which solves the problem that the multi-channel airborne radar system has fewer channels The Doppler resolution is low and the angle measurement error is large.

The present invention adopts the following technical solutions for solving the problems of the technologies described above:

An improved space-time adaptive monopulse angle measurement method based on local joint processing, including the following steps:

Step 1), set h=0, and obtain the received signal z of each array element of the airborne radar detection distance unit according to the following formula:

z=bs+n

Among them, b represents the complex envelope of the target, n represents clutter plus noise, s is the space-time steering vector of the target,

Kronecker product, s _t = [1 e ^j2πv ... e ^j2πv(K-1) ] ^T , s _s = [1 e ^j2πu ... e ^j2πu(N-1) ] ^T , s _t and s _s respectively Corresponding to the time domain steering vector and the air domain steering vector, the superscript T represents the transpose operation;

v represents the normalized Doppler frequency of the target, u=d sinθ/λ represents the normalized spatial frequency of the target, d represents the array element spacing, λ represents the wavelength, θ represents the target azimuth space angle, and N represents the radar antenna element The number, K is the number of pulses in a coherent accumulation cycle;

Step 2), the detection space-time steering vector s ₀ of the angle-Doppler unit where the target to be detected is:

in, s _t0 and s _s0 respectively correspond to the time domain steering vector at the center of the Doppler unit to be detected and the space domain steering vector at the center of the angle unit to be detected;

v ₀ represents the normalized Doppler frequency of the center of the Doppler unit to be detected, u ₀ =d sinθ ₀ /λ represents the normalized spatial frequency of the center of the angle unit to be detected, and θ ₀ is the azimuth pointing angle of the transmitting antenna;

Step 3), according to the following formula, the local joint processing (JDL) dimensionality reduction matrix of the angle-Doppler unit to be detected is obtained:

Among them, T _t is the dimensionality reduction matrix in the time domain, and T _s is the dimensionality reduction matrix in the space domain:

Step 4), obtain the JDL dimension reduction data of the angle-Doppler unit to be detected according to the following formula:

z _T = T ^H z

Among them, the superscript ^H represents the complex conjugate transpose operation;

Step 5), using adjacent distance units as samples to perform maximum likelihood estimation to obtain the JDL dimensionality reduction clutter plus interference noise covariance matrix R _T ;

Step 6), calculate JDL and beam adaptation weight w _T according to the following formula:

in, Indicates the JDL dimensionality reduction detection space-time steering vector,

Step 7), respectively calculate the JDL azimuth difference beam adaptive weight and the JDL time domain difference beam adaptive weight;

Step 8), calculate the estimated value of the target normalized spatial frequency u according to the following formula Estimated value of target normalized Doppler frequency v

where r _u is the monopulse ratio of the azimuth difference beam to the sum beam, μ _u is the offset correction value of r _u , r _v is the monopulse ratio of the time domain difference beam to the sum beam, μ _v is the offset of r _v displacement correction value; is the slope matrix of the space-time adaptive monopulse ratio;

Step 9), make h=h+1;

Step 10), judge whether h is less than m, if h<m, m is a pre-set integer greater than 1, the detection space-time steering vector _s0 in step 1 is corrected as in:

And correct T _t and T _s in the JDL dimensionality reduction matrix T in step 1 as:

Step 11), repeatedly execute step 4) to step 10), until h=m;

Step 12), calculate the estimated value of the target azimuth space angle θ according to the following formula

Among them, arcsin( ) is an arcsine operation;

Step 13), output the estimated value of the target azimuth space angle.

As a further optimization scheme of the improved space-time adaptive monopulse angle measurement method based on local joint processing in the present invention, m in step 10) is equal to 2.

As a further optimization scheme of the improved space-time adaptive monopulse angle measurement method based on local joint processing in the present invention, the JDL azimuth difference beam adaptive weight and the JDL time domain difference beam adaptive weight calculation formula in step 7) They are as follows:

where d _T,u = T ^H d _u , d _T,v = T ^H d _v ,

Diagonal matrix D _N =diag(λπi[0 1 ... N-1]/d), D _K =diag(2πi[0 1 ... K -1]).

As a further optimization scheme of the improved space-time adaptive monopulse angle measurement method based on local joint processing in the present invention, r _v , μ _v , r _u , μ _u and The calculation formulas of each element in are as follows:

As a further optimization scheme of the improved space-time adaptive monopulse angle measurement method based on local joint processing in the present invention, the estimated value of RT described in step 5 ₎ for:

Among them, x _Ti =T ^H x _i represents the output of the i-th training sample x _i after dimensionality reduction processing by the dimensionality reduction matrix T, and L is the number of samples.

As a further optimization scheme of the improved space-time adaptive monopulse angle measurement method based on local joint processing in the present invention, L=27.

Compared with the prior art, the present invention adopts the above technical scheme and has the following technical effects:

1. When the target spatial frequency and Doppler frequency are mismatched at the same time, the method iteratively updates the space-time domain dimensionality reduction matrix and target space-time steering vector of the JDL algorithm by estimating the target Doppler frequency and azimuth angle, which can reduce the Doppler overcomes the loss, thereby improving the output signal-to-noise ratio, and can obtain higher angle measurement accuracy than conventional adaptive monopulse.

2. Fast convergence speed and easy engineering implementation.

Description of drawings

Figure 1 is a flow chart of the improved space-time adaptive monopulse angle measurement method based on JDL;

Figure 2 is the change curve of the target angle estimation of the JDL-STAM algorithm with the number of iterations;

Figure 3 is the change curve of the target angle estimation of the JDL-STAM and JDL-MSTAM algorithms with the number of iterations;

Figure 4 is the curve of the normalized target Doppler frequency estimated by the JDL-STAM and JDL-MSTAM algorithms with the number of iterations;

Figure 5 is the variation curve of target angle estimation RMSE with SCNR for JDL-STAM and JDL-MSTAM algorithms;

Fig. 6 is the change curve of target angle estimation RMSE versus SCNR of JDL-MSTAM algorithm under different pulse numbers.

Detailed ways

Below in conjunction with accompanying drawing, technical scheme of the present invention is described in further detail:

The invention discloses an improved space-time self-adaptive monopulse angle measurement method based on local joint processing. angle, iteratively updating the space-time dimensionality reduction matrix of the JDL algorithm and the space-time steering vector of target detection can significantly reduce the Doppler crossover loss, thereby improving the output signal-to-noise ratio. This method only needs two iterations to accurately estimate the target space angle.

Assuming that the radar antenna has N array elements, each array element antenna is isotropic, and the number of pulses in a coherent accumulation cycle is K, then the radar signal Doppler unit signal model of the detection unit can be expressed as follows:

z=bs+n

Among them, z represents the signal vector received by the detection unit array, b represents the complex envelope of the target, and n represents the clutter plus noise; assuming that the clutter plus noise n obeys the Gaussian distribution with a mean value of 0 and a covariance of R, the clutter, noise and The targets are not related to each other; s is the space-time steering vector of the target:

Among them, s _t =[1 e ^j2πv ... e ^j2πv(K-1) ] ^T , s _s =[1 e ^j2πu ... e ^j2πu(N-1) ] ^T , s _t and s _s respectively correspond to The domain steering vector and the space domain steering vector, the superscript T represents the transpose operation, v represents the normalized Doppler frequency of the target, u=d sinθ/λ represents the normalized spatial frequency of the target, d is the distance between the array elements, λ is the wavelength, and θ is the target azimuth space angle.

The detection space-time steering vector _s0 of the angle-Doppler unit where the target is to be detected is:

in, s _t0 and s _s0 respectively correspond to the time domain steering vector at the center of the Doppler unit to be detected and the space domain steering vector at the center of the angle unit to be detected, and v ₀ represents the normalized Doppler frequency of the center of the Doppler unit to be detected , u ₀ =d sinθ ₀ /λ represents the normalized spatial frequency of the center of the angular unit to be detected, and θ ₀ is the azimuth pointing angle of the transmitting antenna.

The JDL algorithm first transforms the space-time signal data into the angle-Doppler domain through the two-dimensional discrete Fourier transform (DFT). In the angle Doppler domain, since the radar emission energy is mainly concentrated in the observation direction, the angle Doppler units are grouped in the observation direction, and each group forms a local processing region (LPR). Assume that the radar array antenna has N columns, and the number of time-domain pulses in one coherent processing interval is K. It is assumed that there are 3 angle units and 3 Doppler units in the LPR, then the transformation mapped to the specified LPR is realized by matrix T. where T is a dimensionality-reduction transposition matrix composed of a series of Kronecker products of airspace and time-domain steering vectors, defined as:

In the formula, is the Kronecker product; T _t is the dimensionality reduction matrix in the time domain, namely:

T _s is the spatial domain dimensionality reduction matrix, namely:

It is assumed here that clutter and noise obey a Gaussian distribution with a mean value of 0 and the covariance matrix of clutter plus noise is defined as R. After matrix T operation, data conversion and dimensionality reduction are realized at the same time, in which the angle domain is divided into 1/N intervals, and the Doppler domain is divided into 1/K intervals, which is equivalent to the two-dimensional DFT transformation of the JDL algorithm.

After conversion, the JDL dimensionality reduction data z _T =T ^H z of the angle-Doppler unit to be detected, the clutter plus interference noise covariance matrix R _T =E{z _T z _T ^H }, E{ } represents the mathematical expectation Operation, superscript ^H indicates complex conjugate transpose operation, JDL and beam adaptive weights in Represents the space-time steering vector for JDL dimensionality reduction detection; since the noise characteristics are unknown, the calculation of the covariance matrix R _T of the above formula in practical applications is based on its maximum likelihood estimation form replace:

Where x _Ti =T ^H x _i represents the output of the i-th training sample x _i after dimensionality reduction processing by the dimensionality reduction matrix T. L is the number of samples. In order to ensure the estimation accuracy, the samples need to meet the independent and identical distribution condition statistically with the noise component of the unit to be tested. It is possible to take L=27 independent and identically distributed training samples.

If the normalized Doppler frequency of the target is strictly limited to the center Doppler frequency of the detection unit, and the normalized spatial frequency of the target is also at the center of the corresponding detection unit, then the space-time steering vector in the angle-Doppler domain is

s _T = [0 ... 0 1 0 ... 0] ^T

Among them, "1" represents the detection angle-Doppler unit, and the other units are "0". However, when there is a deviation, that is, when the target Doppler and angle both deviate from the center of the detection unit, the JDL transformation matrix T and the transformed target space-time guidance time domain must be mismatched. It further leads to the loss of target Doppler crossing loss and the loss of output signal-to-noise ratio, and the subsequent target detection error increases. That is to say, in order to further improve the target angle measurement accuracy, it is necessary to correct and compensate the target Doppler and angle parameters.

The data is processed by JDL, and the beam adaptive weights corresponding to the space-time adaptive monopulse corresponding to the space-time adaptive monopulse should be calculated in the angle-Doppler unit to be detected. Then the JDL azimuth difference beam adaptive weight and the JDL time domain difference beam adaptive weight are respectively:

where d _T,u = T ^H d _u , d _T,v = T ^H d _v , Diagonal matrix D _N =diag(λπi[0 1 ... N -1]/d), D _K =diag(2πi[0 1 ... K -1]).

The formulas for target azimuth space angle and Doppler frequency estimated by space-time adaptive monopulse are as follows:

in and are the estimated values of the target normalized spatial frequency u and the target normalized Doppler frequency v respectively, r _u is the monopulse ratio of the spatial domain difference beam and the sum beam, μ _u is the offset correction value of r _u , Their calculation formulas are:

r _v is the monopulse ratio of the difference beam and the sum beam in the time domain, μ _v is the offset correction value of r _v , and the calculation formulas are:

is the slope matrix of the space-time adaptive monopulse ratio, and the calculation formulas of its elements are:

In the case of variable determination, JDL-based space-time adaptive monopulse (herein referred to as JDL-STAM) can be used to estimate the target Doppler frequency and space angle. However, when the target Doppler frequency deviates from the center frequency of the detected Doppler unit and the spatial frequency does not match at the same time, the angle measurement error of the JDL-STAM algorithm increases accordingly. JDL-STAM achieves detection-steering vector matching after JDL processing, but it does not compensate for target Doppler crossing loss. Multi-step single pulses can further reduce the deviation of the estimated value from the true value. In order to further improve the accuracy of angle measurement, it should be able to simultaneously update the space-time steering vector _s0 and the space-time domain dimensionality reduction matrix T of the JDL algorithm according to the estimated spatial frequency and Doppler frequency, that is, the JDL improvement based on the present invention After the space-time adaptive monopulse algorithm (JDL-MSTAM) estimates u and v with the given observation direction spatial frequency and normalized Doppler frequency, it is used as the new initial value u ₀ and v ₀ , Update the detection time domain steering vector s _t0 and the space domain steering vector s _s0 , and redesign the space-time dimensionality reduction matrix of the JDL algorithm which is:

The corrected detection space-time steering vector s ₀ is in:

Re-estimate the sample covariance matrix and the weights of the corresponding adaptive sum beam, time-domain adaptive difference beam and space-domain adaptive difference beam according to the updated detection space-time steering vector and JDL dimensionality reduction matrix, and iteratively estimate the space-time adaptive Estimating the target space angle and Doppler frequency with a single pulse not only improves the target Doppler coherent accumulation gain, but also reduces the target Doppler cross-domain loss and the mismatch loss of the space-time steering vector, so better measurement results can be obtained. Angular accuracy. The simulation experiment results show that the target normalized spatial frequency u and normalized Doppler frequency v can be accurately estimated by two-step iterative operation. The estimated target azimuth space angle is

θ=arcsin(λu/d)

Among them, arcsin(·) is the arcsine operation.

In summary, the specific signal flow chart of the JDL-based improved space-time adaptive monopulse angle measurement method proposed by the present invention is shown in FIG. 1 . The performance of the algorithm is evaluated by computer simulation based on radar clutter simulation data. Radar system parameters refer to the table below:

Inject a target to be detected in the distance unit to be detected, its normalized Doppler frequency v=0, the normalized center Doppler frequency v ₀ =1/2K of the Doppler unit where the target is located, and the LPR is 3× 3.

Figure 2 shows the angle estimation diagram of the JDL-STAM algorithm at the center Doppler frequency of the target Doppler frequency offset detection unit. As shown in Figure 1, after two iterations, the JDL-STAM algorithm estimates the most accurate angle when the offset is the smallest, that is, when v=0.004. However, when the target normalized Doppler frequency v shifts from the center normalized Doppler frequency, the accuracy of target angle estimation decreases immediately, and as the Doppler frequency offset increases, its angle estimation error becomes larger correspondingly .

When v=0.002, the angle estimation results of each iteration of the JDL-STAM and JDL-MSTAM algorithms are shown in Figure 3. Different from the JDL-STAM algorithm, the angle estimation error of the JDL-MSTAM algorithm is smaller. The high precision of the JDL-MSTAM algorithm benefits from its joint estimation of the target Doppler frequency and azimuth space angle. By iteratively updating the JDL transformation matrix, the space-time steering vector is corrected to reduce the target Doppler crossing loss and the steering vector matching error. . Figure 4 shows the target normalized Doppler frequency estimated by each iteration of the JDL-STAM and JDL-MSTAM algorithms. As shown in Figure 4, the target Doppler frequency estimated by the JDL-MSTAM algorithm is more accurate.

Root mean square error (RMSE) is used to quantify the angle estimation accuracy of the adaptive monopulse processor studied in this paper. The root mean square error is defined as

In the formula, M is the number of Monte Carlo experiments, Indicates the target azimuth angle estimated by the mth time, and θ indicates the actual target azimuth angle. The results below are the average of 200 independent Monte Carlo experiments for 3-step space-time adaptive monopulse iterations. The variation curves of the RMSE of the target azimuth space angle estimated by the two methods with the signal-to-clutter-to-noise ratio (SCNR) are shown in Fig. 5 . Target normalized Doppler frequency v = 0.002, SCNR varied from -10dB to 30dB. It can be seen from the figure that the estimated RMSE of the two algorithms decreases with the increase of SCNR, but the error of the JDL-MSTAM algorithm is smaller.

Assuming that the target Doppler frequency deviates from the center frequency of the detection Doppler unit by 40%, Fig. 6 shows the change curve of target angle estimation RMSE versus SCNR under different pulse number conditions of JDL-MSTAM algorithm, namely K=128, 64 and 32. It can be seen from the figure that when the number of pulses is large, the Doppler resolution is improved, and the JDL-MSTAM algorithm can obtain higher angle measurement accuracy.

Those skilled in the art can understand that, unless otherwise defined, all terms (including technical terms and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. It should also be understood that terms such as those defined in commonly used dictionaries should be understood to have a meaning consistent with the meaning in the context of the prior art, and unless defined as herein, are not to be interpreted in an idealized or overly formal sense explain.

The specific embodiments described above have further described the purpose, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above descriptions are only specific embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims (5)

1. The improved space-time self-adaptive monopulse angle measurement method based on local joint processing, is characterized in that, comprises the steps: Step 1), set h=0, and obtain the received signal z of each array element of the airborne radar detection distance unit according to the following formula: z=bs+n Among them, b represents the complex envelope of the target, n represents clutter plus noise, s is the space-time steering vector of the target, Kronecker product, s _t = [1 e ^j2πv ... e ^j2πv(K-1) ] ^T , s _s = [1 e ^j2πu ... e ^j2πu(N-1) ] ^T , s _t and s _s respectively Corresponding to the time domain steering vector and the air domain steering vector, the superscript T represents the transpose operation; v represents the normalized Doppler frequency of the target, u=dsinθ/λ represents the normalized spatial frequency of the target, d represents the array element spacing, λ represents the wavelength, θ represents the target azimuth space angle, and N represents the number of radar antenna elements number, K is the number of pulses in a coherent accumulation cycle; Step 2), the detection space-time steering vector s ₀ of the angle-Doppler unit where the target to be detected is: in, s _t0 and s _s0 respectively correspond to the time domain steering vector at the center of the Doppler unit to be detected and the space domain steering vector at the center of the angle unit to be detected; v ₀ represents the normalized Doppler frequency of the center of the Doppler unit to be detected, u ₀ = dsinθ ₀ /λ represents the normalized spatial frequency of the center of the angle unit to be detected, and θ ₀ is the azimuth pointing angle of the transmitting antenna; Step 3), according to the following formula, the local joint processing (JDL) dimensionality reduction matrix of the angle-Doppler unit to be detected is obtained: Among them, T _t is the dimensionality reduction matrix in the time domain, and T _s is the dimensionality reduction matrix in the space domain: Step 4), obtain the JDL dimension reduction data of the angle-Doppler unit to be detected according to the following formula: z _T = T ^H z Among them, the superscript H represents the complex conjugate transpose operation; Step 5), using adjacent distance units as samples to perform maximum likelihood estimation to obtain the JDL dimensionality reduction clutter plus interference noise covariance matrix R _T ; Step 6), calculate JDL and beam adaptation weight w _T according to the following formula: in, Indicates the JDL dimensionality reduction detection space-time steering vector, Step 7), respectively calculate the JDL azimuth difference beam adaptive weight and the JDL time domain difference beam adaptive weight; Step 8), calculate the estimated value of the target normalized spatial frequency u according to the following formula Estimated value of target normalized Doppler frequency v where r _u is the monopulse ratio of the azimuth difference beam to the sum beam, μ _u is the offset correction value of r _u , r _v is the monopulse ratio of the time domain difference beam to the sum beam, μ _v is the offset of r _v displacement correction value; is the slope matrix of the space-time adaptive monopulse ratio; Step 9), make h=h+1; Step 10), judge whether h is less than m, if h<m, m is a pre-set integer greater than 1, the detection space-time steering vector _s0 in step 1 is corrected as in: And correct T _t and T _s in the JDL dimensionality reduction matrix T in step 1 as: Step 11), repeatedly execute step 4) to step 10), until h=m; Step 12), calculate the estimated value of the target azimuth space angle θ according to the following formula Among them, arcsin( ) is an arcsine operation; Step 13), output the estimated value of the target azimuth space angle. 2 . The improved space-time adaptive monopulse angle measurement method based on local joint processing according to claim 1 , wherein m in the step 10) is equal to 2. 3 . 3. Based on the improved space-time adaptive monopulse angle measurement method based on local joint processing according to claim 1, it is characterized in that the JDL azimuth difference beam adaptive weight and the JDL time domain difference beam described in step 7) The adaptive weight calculation formulas are as follows: where d _T,u = T ^H d _u , d _T,v = T ^H d _v , Diagonal matrix D _N =diag(λπi[0 1 ... N-1]/d), D _K =diag(2πi[0 1 ... K-1]). 4. Based on the improved space-time adaptive monopulse angle measurement method based on local joint processing according to claim 1, it is characterized in that the estimated value of R _T in step 5) for: Among them, x _Ti =T ^H x _i represents the output of the i-th training sample x _i after dimensionality reduction processing by the dimensionality reduction matrix T, and L is the number of samples. 5. Based on the improved space-time adaptive monopulse angle measurement method based on local joint processing according to claim 4, it is characterized in that L=27.