A radar adaptive detection method based on improved large-dimensional precision matrix estimation

Abstract

The present application relates to radar signal processing technical field, especially to a kind of radar adaptive detection method based on improved large-dimensional precision matrix estimation.The method for the low problem of existing radar detection accuracy is as follows: radar data is collected and radar data is preprocessed, to obtain the radar data after preprocessing;According to the radar data after preprocessing, determine the unit to be detected and reference unit;According to reference unit, construct sample covariance matrix, then inverse sample covariance matrix, to obtain sample precision matrix;According to sample precision matrix, obtain improved precision estimation matrix;According to the improved precision estimation matrix, judge whether there is target in the distance unit of the unit to be detected, the detection performance of radar system in the case where dimension N and sample size K are in the same order of magnitude can be significantly improved by using the present application.The low problem of radar detection accuracy caused by poor detection performance of adaptive detector in the large-dimensional asymptotic system of existing method is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of radar signal processing, and particularly relates to a radar adaptive detection method based on improved large-dimension precision matrix estimation. BACKGROUND

[0002] Radar target detection in strong clutter (noise) environment has always been one of the core problems in the field of radar signal processing. With the progress of electronic technology, various new system radars have emerged, especially phased array radars have developed rapidly, and due to the successful description of the spherically invariant random vector to the multi-dimensional non-Gaussian signal, the coherent processing of radar echo signal vector has become a research hotspot in the past three decades. Whether it is a multi-dimensional Gaussian or non-Gaussian clutter, the precision matrix (i.e. the inverse of the covariance matrix) is an important parameter to describe the amplitude and correlation characteristics of the signal. If the clutter precision matrix is known in advance, the optimal detector can be derived through the Neyman-Pearson (NP) criterion. However, in real scenarios, the prior information of the clutter (or noise) precision matrix cannot be obtained in advance. At this time, the real clutter precision matrix needs to be estimated through some auxiliary data near the unit to be detected. This kind of detector is called adaptive detector.

[0003] The design problem of the optimal adaptive detector has attracted widespread attention of researchers in the field of radar in the past forty years. Improving the detection ability of adaptive detectors in complex environments is the research focus in the field of radar target detection. Precision matrix estimation is of great significance to adaptive detectors, and a sufficiently accurate precision matrix is the key to improving the performance of adaptive detectors.

[0004] At present, in the radar signal processing flow, the precision matrix estimation is established on the basis of the classical asymptotic system, that is, the low-dimensional large-sample situation (in the classical asymptotic system, it is generally assumed that the dimension N is fixed and the sample number K tends to infinity, that is, N is fixed, and K tends to infinity.) For example, in the Gaussian clutter background, the sample precision matrix (also the maximum likelihood estimate) is often used to estimate the real clutter covariance matrix.

[0005] However, with the emergence of phased array radars and the use of large antenna arrays, and due to the limitations of data storage space and the real-time requirements of signal processing, the sample size K and sample dimension N of the observed data in actual radar signal processing are at the same order of magnitude. In addition, in some special scenarios, due to the non-uniformity of clutter, the sample size K may be less than the sample dimension N. For example, in a sky wave over-the-horizon radar, a coherent integration period often contains 128 or 256 samples, i.e. N = 128 or 256, but due to the non-uniformity of the sky wave over-the-horizon radar clutter, only 40-60 auxiliary data are available for clutter covariance matrix estimation in the range dimension, i.e. K = 40-60. This situation is referred to as a large-dimensional asymptotic system, i.e. the dimension N and the sample size K both tend to infinity, and the ratio of the two tends to a constant (N / K→c∈(0,1) when N,K→∞). In the large-dimensional asymptotic system, the sample precision matrix estimation is not an unbiased consistent estimate of the true covariance matrix.

[0006] The inconsistency of the precision matrix estimation ultimately leads to a large loss in the performance of the detector. Taking the generalized likelihood ratio detector as an example, when K = 2N, the performance of the GLRT detector has a 3dB signal-to-clutter-and-noise ratio loss compared to the optimal detector, which is unacceptable in actual scenarios.

[0007] Therefore, the existing radar detection method has the problem of low radar detection accuracy due to poor detection performance of the radar adaptive detector in the large-dimensional asymptotic system (i.e. N and K are at the same order of magnitude). SUMMARY

[0008] The purpose of the present application is to solve the problem of low radar detection accuracy due to poor detection performance of the radar adaptive detector in the large-dimensional asymptotic system (i.e. N and K are at the same order of magnitude) of the existing radar detection method. An improved large-dimensional precision matrix estimation-based radar adaptive detection method is invented, which comprises the following steps:

[0009] S1: initialize the radar parameters; collect the radar data and pre-process the radar data to obtain pre-processed radar data;

[0010] S2: determine the to-be-detected unit and the reference unit according to the pre-processed radar data;

[0011] S3: construct a sample covariance matrix according to the reference unit, and then invert the sample covariance matrix to obtain a sample precision matrix;

[0012] S4: obtain the eigenvalues and eigenvectors of the sample precision matrix according to the sample precision matrix,

[0013] S5: constructing an improved precision estimation matrix according to the radar parameters set in S1 and the eigenvalues and eigenvectors of the sample precision matrix obtained in S4;

[0014] S6: constructing a test statistic of the unit to be detected according to the improved precision estimation matrix;

[0015] S7: detecting according to the test statistic of the unit to be detected and the set radar parameters to obtain a detection result.

[0016] The beneficial effects of the present application are:

[0017] The detection method in the present application can significantly improve the detection performance of the radar system when the dimension N and the sample size K are in the same order of magnitude. When the false alarm probability is 10 -3 , and the detection probability is 0.9, the signal-to-jamming-and-noise ratio gain of the present application is improved by more than 3dB compared with the traditional method, solving the problem of low radar detection accuracy caused by poor detection performance of the adaptive detector in the large-dimensional asymptotic system (i.e. N and K are basically comparable in order of magnitude). BRIEF DESCRIPTION OF DRAWINGS

[0018] Figure 1 is a flowchart of the implementation of the present application;

[0019] Figure 2 is a schematic diagram of the detection performance curve of the present application. DETAILED DESCRIPTION

[0020] Specific implementation one: in combination Figure 1 to illustrate the present application, including:

[0021] S1: initializing radar parameters; collecting radar data and pre-processing the radar data to obtain pre-processed radar data;

[0022] S2: determining a unit to be detected and a reference unit according to the pre-processed radar data;

[0023] S3: constructing a sample covariance matrix according to the reference unit, and then inverting the sample covariance matrix to obtain a sample precision matrix;

[0024] S4: obtaining eigenvalues and eigenvectors of the sample precision matrix according to the sample precision matrix,

[0025] S5: constructing an improved precision estimation matrix according to the radar parameters set in S1 and the eigenvalues and eigenvectors of the sample precision matrix obtained in S4;

[0026] S6: constructing a test statistic of the unit to be detected according to the improved precision estimation matrix;

[0027] S7: detecting according to the inspection statistics of the unit to be detected and the set radar parameters to obtain a detection result.

[0028] By using the detection method in the application, the detection performance of the radar system can be significantly improved when the dimension N and the sample size K are in the same order of magnitude. When the false alarm probability is 10 -3 , and the detection probability is 0.9, the signal-to-jamming-and-noise ratio gain of the application is improved by more than 3dB than the traditional method, solving the problem of low radar detection accuracy caused by poor detection performance of the adaptive detector in the large dimension asymptotic system (i.e. N and K are basically comparable in order of magnitude)

[0029] Specific implementation method two: the difference between this implementation method and the specific implementation method one is that the radar parameters are initialized in S1; radar data is collected and preprocessed to obtain preprocessed radar receiving data; the specific process is as follows:

[0030] S1.1: setting radar system parameters,

[0031] S1.2: performing radar scanning on a target area according to the radar with set parameters; each pulse signal emitted by the radar is collected and processed by N a antenna array elements of the radar; radar echo data is obtained;

[0032] The radar echo data includes X distance unit radar echo data; X is a positive integer;

[0033] The directivity of a single antenna is limited, and in order to adapt to various applications, two or more single antennas working at the same frequency are fed and spatially arranged according to certain requirements to form an antenna array, also called an antenna array. The antenna radiation unit constituting the antenna array is called an antenna array element.

[0034] In radar detection, a distance unit refers to a minimum area in which the received signals are divided according to distance by a radar system. Each distance unit corresponds to a spatial interval from the radar to a certain distance (or target).

[0035] S1.3: preprocessing the radar echo data to obtain preprocessed radar data;

[0036] The other steps and parameters are the same as those in the specific implementation method one.

[0037] Specific implementation method three: the difference between this implementation method and the specific implementation method one is that

[0038] The specific process of the radar system parameters in S1.1 is as follows:

[0039] The number of radar transmitted pulses is set to N p , Np is a positive integer;

[0040] set the false alarm probability P of the radar fa ;

[0041] The other steps and parameters are the same as one of the first to third embodiments.

[0042] The fourth embodiment is different from the first to fourth embodiments in that,

[0043] The specific process of preprocessing the radar echo data in S1.3 to obtain the preprocessed radar data is as follows:

[0044] The N p ×N a data in one range cell of the radar echo data is rearranged by column to obtain an N-dimensional column vector, where N=N p ×N a ; finally, X N-dimensional column vectors are obtained, and the X N-dimensional column vectors are taken as the preprocessed radar data.

[0045] The rearrangement by column is, for example, the original data dimension is 2×3, and is represented as:

[0046]

[0047] The rearranged N=6-dimensional data is represented as

[0048]

[0049] The other steps and parameters are the same as one of the first to third embodiments.

[0050] The fifth embodiment is different from the first to fourth embodiments in that,

[0051] The specific process of determining the to-be-detected unit and the reference unit according to the preprocessed radar data in S2 is as follows:

[0052] S2.1: Determine the distance cell where the to-be-detected unit is located, and take the N-dimensional column vector corresponding to this distance cell as the to-be-detected unit y0;

[0053] S2.2: Take K / 2 distance cells before and after the to-be-detected unit in the distance cell where the to-be-detected unit is located, obtain K distance cells, and take the K N-dimensional column vectors corresponding to the selected K distance cells as the reference units [y1, y2,..., y K ],

[0054] The other steps and parameters are the same as one of the first to fourth embodiments.

[0055] Embodiment six: the difference between this embodiment and embodiments one to five is that,

[0056] The S3 constructs a sample covariance matrix according to a reference unit, and then inverses the sample covariance matrix to obtain a sample precision matrix, which is expressed by a formula as follows:

[0057]

[0058] wherein (·) H represents a conjugate transpose; represents a sample covariance matrix, represents a sample precision matrix; y i represents an N-dimensional feature vector corresponding to the i-th distance unit; i = [1, 2, 3, …, K]; represents y i a conjugate transpose; other steps and parameters are the same as one of embodiments one to five.

[0059] Embodiment seven: the difference between this embodiment and embodiments one to six is that,

[0060] In the S4, the eigenvalues and eigenvectors of the sample precision matrix are obtained according to the sample precision matrix, and the specific process is as follows:

[0061] The QR decomposition is performed on the sample precision matrix to obtain the eigenvalues of the sample precision matrix , and the eigenvectors corresponding to the eigenvalues;

[0062] The N eigenvalues of the sample precision matrix are expressed as The N eigenvectors corresponding to the N eigenvalues are expressed as wherein (·) T represents a transpose;

[0063] The QR decomposition is a method of decomposing a matrix into an orthogonal matrix Q and an upper triangular matrix R

[0064] The QR decomposition can be used to calculate the eigenvalues of a matrix, and the eigenvalues are gradually approximated by repeatedly performing the QR decomposition. It is well known to those skilled in the art.

[0065] Other steps and parameters are the same as one of embodiments one to six.

[0066] Embodiment eight: the difference between this embodiment and embodiments one to seven is that,

[0067] In the S5, an improved precision estimation matrix is constructed according to the radar parameters set in the S1 and the eigenvalues and eigenvectors of the sample precision matrix obtained in the S4; the specific process is as follows:

[0068] S5.1: Set radar parameters according to S1 to obtain the rank r of the clutter component in the radar echo; expressed by the formula as follows:

[0069] r = N p +N a -1

[0070] S5.2: According to the rank of the clutter component in the radar echo, S4, obtain the eigenvalues and eigenvectors of the sample precision matrix to construct an improved precision estimation matrix, and construct an improved precision matrix expressed by the formula as follows:

[0071]

[0072] denotes the estimated noise power; and denote the kth and jth sample eigenvalues of the sample precision matrix ordered from large to small, and k and j represent the ordered indices of the sum; denotes the improved clutter component eigenvalue.

[0073] The other steps and parameters are the same as one of the first to seventh embodiments.

[0074] The ninth embodiment is different from the first to eighth embodiments in that,

[0075] The S6 is constructed according to the improved precision estimation matrix to construct the test statistic T of the detection unit to be detected, expressed by the formula as follows:

[0076]

[0077] wherein s represents a steering vector of N x 1, s H denotes the conjugate transpose of s, y0 represents the detection unit to be detected, denotes the conjugate transpose of y0; || represents the modulus of a complex number;

[0078] The steering vector s is used to indicate the direction or target in signal processing, and is artificially set according to the actual situation. Generally, the steering vector s is defined as a unit vector in a certain direction.

[0079] The other steps and parameters are the same as one of the first to eighth embodiments.

[0080] The tenth embodiment is different from the first to ninth embodiments in that,

[0081] The S7 in accordance with the detection unit of the detection statistics T and the detection threshold τ set to determine whether there is a target in the distance unit where the detection unit is specific process:

[0082] If T≥τ, τ represents the set detection threshold, then there is a target in the distance unit where the detection unit y0 is located;

[0083] If T<τ, then there is no target in the distance unit where the detection unit y0 is located:

[0084] The detection threshold τ is set according to the false alarm probability P fa The detection threshold and the false alarm probability are converted by looking up the table, which is well known to those skilled in the art;

[0085] The other steps and parameters are the same as one of the first nine embodiments.

[0086] Combined with the simulation analysis of the first ten embodiments

[0087] Radar system configuration: in this embodiment, the radar system is configured to transmit N p = 16 pulses, each pulse is received by N a = 4 antenna elements. Therefore, the data dimension of each distance unit is N = N p × N a = 64.

[0088] Reference unit selection: select K = 128 distance unit data around the detection unit as reference unit.

[0089] False alarm probability setting: set the false alarm probability to P fa = 10 -3 .

[0090] Start the radar system, scan the target area, and collect the data set containing multiple distance units.

[0091] The 16×4 data in each distance unit is rearranged as a 64-dimensional column vector.

[0092] Select the data of a distance unit from the data set as the detection unit y0.

[0093] Select 128 distance unit data around the detection unit as reference unit [y1, y2,..., y K ]. Using 128 reference unit data, a 64×64-dimensional sample covariance matrix is obtained according to the following formula

[0094]

[0095] in(·) H This indicates the conjugate transpose.

[0096] For the sample covariance matrix Inverting the matrix yields the sample precision matrix.

[0097] Using QR decomposition, we obtain 64 eigenvalues and its corresponding eigenvectors in(·) T This indicates transpose.

[0098] According to the theoretical formula r = N p +N a -1, calculate the rank r of the clutter component in the echo. According to the following formula... Nonlinear shrinkage is performed on the 64 eigenvalues:

[0099]

[0100] Where N = 64 and K = 128.

[0101] The improved accuracy matrix estimation result is obtained according to the following formula:

[0102]

[0103] Where N = 64. Using the improved accuracy matrix estimation results, the generalized likelihood ratio test statistic T is constructed according to the following formula:

[0104]

[0105] Based on the set false alarm probability P fa =10 -3 The desired detection threshold τ is obtained through Monte Carlo simulation experiments.

[0106] Compare the test statistic T with the threshold τ. If T ≥ τ, it means that there is a target in the distance unit y0 to be detected; if T < τ, it means that there is no target in the distance unit y0 to be detected.

[0107] The detection performance curve of this embodiment is plotted on Figure 2 In the middle, the threshold τ passes through 10 5 The results were obtained from Monte Carlo simulations, with the clutter rank r set to 8. It can be seen that when N and K are of the same order of magnitude, the detection performance of this invention is significantly improved compared to existing technologies (generalized likelihood ratio detectors based on the sample precision matrix): when the false alarm probability is 10... -3 When the detection probability is 0.9, compared with the existing technology based on the sample precision matrix, the signal-to-noise ratio gain of the present invention is more than 3dB.

[0108] The above merely describes preferred embodiments of the present application, and it should be understood that the present application is not limited to the specific embodiments described above. Although the present application has been disclosed with reference to the preferred embodiments above, it should be understood that the present application is not limited to the specific embodiments described above, but any person skilled in the art can make some changes or modifications to the above disclosed technical content without departing from the technical solution of the present application, and any simple modification, equivalent replacement and improvement of the above embodiments made within the scope of the technical solution of the present application and the spirit and principles of the present application are still within the protection scope of the present application.

Claims

1. A radar adaptive detection method based on improved large dimensional precision matrix estimation, characterized in that, The method comprises the following steps: S1: setting radar parameters; collecting radar data and pre-processing the radar data to obtain pre-processed radar data; S2: determining a to-be-detected unit and a reference unit according to the pre-processed radar data; S3: constructing a sample covariance matrix according to the reference unit, and then inverting the sample covariance matrix to obtain a sample precision matrix; S4: obtaining eigenvalues and eigenvectors of the sample precision matrix according to the sample precision matrix, S5: constructing an improved precision estimation matrix according to the radar parameters set in S1 and the eigenvalues and eigenvectors of the sample precision matrix obtained in S4; the specific process is as follows: S5.1: obtaining the rank r of the clutter component in the radar echo according to the radar parameters set in S1; the formula is as follows: r = N p + N a - 1 In the formula, N p represents the number of transmitted pulses of the radar, N a represents the number of antenna elements of the radar, S5.2: Constructing the improved precision matrix according to the eigenvalues and eigenvectors of the sample precision matrix obtained from the rank of the clutter component in the radar echo, S4 This can be expressed in a formula as follows: represents an estimated noise power; represents an improved clutter component eigenvalue, and respectively represent the kth and jth sample eigenvalues of the sample precision matrix ordered from large to small, i, k and j are all ordering serial numbers. the i-th sample eigenvalue of the sorted sample precision matrix in descending order; N denotes the dimension, the i-th sample eigenvalue of the sorted sample precision matrix the i-th sample eigenvalue of the sorted sample precision matrix the corresponding eigenvector, denotes the conjugate transpose of, K denotes the number of reference cells, S6: constructing a test statistic of the to-be-detected unit according to the improved precision estimation matrix; S7: detecting according to the test statistic of the to-be-detected unit and the set radar parameters to obtain a detection result.

2. The radar adaptive detection method based on improved large-dimension precision matrix estimation according to claim 1, wherein the radar parameters are set in S1; radar data is collected and pre-processed to obtain pre-processed radar reception data; the specific process is as follows: S1.1: setting radar system parameters, obtaining radar echo data; S1.2: The radar scans the target area according to the set parameters, and each pulse signal emitted by the radar is collected and processed by N a antenna elements of the radar. the radar echo data comprises X distance unit radar echo data; X is a positive integer; S1.3: pre-processing the radar echo data to obtain pre-processed radar data.

3. The radar adaptive detection method based on improved large-dimension precision matrix estimation according to claim 2, wherein the specific process of setting radar system parameters in S1.1 is as follows:

4. The radar adaptive detection method based on improved large-dimension precision matrix estimation according to claim 3, wherein the specific process of pre-processing the radar echo data to obtain pre-processed radar data in S1.3 is as follows: X N-dimensional column vectors are finally obtained, and the X N-dimensional column vectors are taken as the pre-processed radar data. The number of radar transmission pulses is set as N p , N p is a positive integer; Setting the false alarm probability P of the radar fa .

5. The radar adaptive detection method based on improved large-dimension precision matrix estimation according to claim 4, wherein the to-be-detected unit and the reference unit are determined according to the pre-processed radar data in S2; the specific process is as follows: S2.1: determining the distance unit where the to-be-detected unit is located, and taking the N-dimensional column vector corresponding to the distance unit as the to-be-detected unit y0; N p ×N a data in one range cell of radar echo data is rearranged by column to obtain an N-dimensional column vector, where N=N p ×N a ; 6. The radar adaptive detection method based on improved large-dimension precision matrix estimation according to claim 5, wherein the sample covariance matrix is constructed according to the reference unit in S3, and then the sample covariance matrix is inverted to obtain a sample precision matrix, which is expressed by the formula as follows:

7. The radar adaptive detection method based on improved large-dimension precision matrix estimation according to claim 6, wherein the eigenvalues and eigenvectors of the sample precision matrix are obtained according to the sample precision matrix in S4; the specific process is as follows:

8. The radar adaptive detection method based on improved large-dimension precision matrix estimation according to claim 7, wherein ​ S2.2: Take K / 2 distance units before and after the distance unit where the unit to be detected is located, and obtain K distance units. Take the K N-dimensional column vectors corresponding to the K distance units as reference units, and represent the K reference units as [y1, y2,..., yK]T. K ] ​ ​ where (·) H denotes the conjugate transpose; denotes the sample covariance matrix, denotes the sample precision matrix; y i denotes the N-dimensional feature vector corresponding to the i-th distance unit; i = [1, 2, 3..., K]; denotes y i the conjugate transpose of ​ ​ For sample precision matrix The sample precision matrix is ​​obtained by performing QR decomposition. The eigenvalues ​​and their corresponding eigenvectors; The sample precision matrix The N eigenvalues ​​are represented as The N eigenvectors corresponding to the N eigenvalues ​​are represented as follows: in(·) T This indicates transpose. ​ The S6 constructs a test statistic of the to-be-detected unit according to the improved precision estimation matrix, and is expressed by a formula as follows: wherein T denotes a test statistic of the cell to be tested, s denotes a steering vector, s H denotes a conjugate transpose of s, y0denotes the cell to be tested, denotes a conjugate transpose of y0; || denotes a modulo operation.

9. The radar adaptive detection method based on improved large-dimensional precision matrix estimation according to claim 8, characterized in that, The S7 detects according to the test statistic of the to-be-detected unit and the set radar parameter, obtains a detection result, and the specific process is as follows: If T≥τ, τ represents a set detection threshold, it is indicated that there is a target in the distance unit where the to-be-detected unit y0 is located; If T<τ, it is indicated that there is no target in the distance unit where the to-be-detected unit y0 is located: The detection threshold τ of the setting is according to the false alarm probability P of the set radar fa Table lookup conversion.

Images