Ship complex HRRP estimation method and system matching the non-Gaussian characteristics of sea clutter

Abstract

The present invention discloses a method and system for estimating the complex HRRP of a ship that matches the non-Gaussian characteristics of sea clutter. The present invention comprises the following steps: obtaining original radar echo data; using a threshold detection method to obtain the distance interval where the complex HRRP of the ship is located for the original radar echo data; representing the radar echo signal within the distance interval as a vector matrix model; using the K‑S minimum distance criterion to select a composite Gaussian model that matches the statistical characteristics of the interference vector in the vector matrix model; and using the sparse optimization method corresponding to the composite Gaussian model to estimate the complex HRRP of the ship. The present invention uses a single-parameter probability model to constrain the sparsity of the complex HRRP of the ship, uses the K‑S minimum distance criterion to determine which composite Gaussian model to use to constrain the statistical characteristics of the sea clutter, and uses the A‑D test to determine the sparse parameters of the complex HRRP of the ship. q , significantly reducing the performance loss of ship complex HRRP estimation caused by clutter model mismatch.

Description

Technical Field

[0001] The invention belongs to the technical field of radar signal processing, and in particular relates to a ship complex HRRP estimation method and system that matches the non-Gaussian characteristics of sea clutter. Background Art

[0002] The complex high-resolution range profile (HRRP) is obtained by summing up all the complex echo vectors of the target scatterer within a single range unit, corresponding to the projection of the complex echo of the target scattering center on the radar line of sight. In the application of high-resolution sea radar, because the complex HRRP contains rich information about the target, many target classification and recognition technologies have been developed based on the complex HRRP.

[0003] For most of the complex HRRP of ships, some strong scattering units are mainly located in the masts, superstructures, cranes and weapons of the ship. Sparsity is also an inherent characteristic of the complex HRRP of ships. However, in the application of sea radar, since sea clutter exhibits strong non-Gaussian characteristics, it poses a serious challenge to the estimation of the complex HRRP of ships. Therefore, the problem of ship complex HRRP estimation can be attributed to estimating the sparse ship complex HRRP from the radar complex echo containing non-Gaussian sea clutter.

[0004] So far, many sparse estimation methods for recovering target signals in noise have been developed. However, these methods take into account the existence of noise, but most of them assume that the background noise is Gaussian noise. In the sparse estimation of the high-resolution sea radar scene, the sea clutter presents a long-tailed non-Gaussian characteristic. In order to solve this problem, there is also a sparse recovery iterative minimization method (SRIM) method in the prior art. This method uses the generalized Pareto distribution as the probability constraint model of the clutter, which improves the accuracy of the sparse estimation of the signal under the background of sea clutter to a certain extent. However, the generalized Pareto distribution cannot constrain the probability characteristics of sea clutter well in many cases, and other composite Gaussian models need to be used to constrain the statistical characteristics of sea clutter. Therefore, when the SRIM method is directly applied to the problem of ship complex HRRP estimation, it sometimes causes serious performance loss. Summary of the invention

[0005] In order to solve the above problems existing in the prior art, the present invention provides a method and system for estimating complex HRRP of a ship matching the non-Gaussian characteristics of sea clutter.

[0006] In a first aspect, the present invention provides a method for estimating complex HRRP of a ship matching the non-Gaussian characteristics of sea clutter, comprising:

[0007] Get raw radar echo data;

[0008] A threshold detection method is used to obtain the distance interval where the ship's complex HRRP is located on the original radar echo data;

[0009] Representing the radar echo signal within the distance interval as a vector matrix model;

[0010] A composite Gaussian model matching the statistical characteristics of the interference vector in the vector matrix model is selected using the KS minimum distance criterion;

[0011] The sparse optimization method corresponding to the composite Gaussian model is used to estimate the complex HRRP of the ship.

[0012] In a second aspect, the present invention provides a ship complex HRRP estimation system matching the non-Gaussian characteristics of sea clutter, comprising:

[0013] A radar signal receiving unit, used to obtain raw radar echo data;

[0014] A threshold detection unit, used for applying a threshold detection method to the original radar echo data to obtain a distance interval where the ship's complex HRRP is located;

[0015] A signal processing unit, used for representing the radar echo signal within the distance interval as a vector matrix model;

[0016] A model selection unit, used for selecting a composite Gaussian model that matches the statistical characteristics of the interference vector in the vector matrix model by using a KS minimum distance criterion;

[0017] The sparse estimation unit is used to estimate the complex HRRP of the ship by adopting the sparse optimization method corresponding to the composite Gaussian model.

[0018] In a third aspect, the present invention provides a computer-readable storage medium having program instructions stored thereon, wherein the program instructions implement the above method when executed.

[0019] In a fourth aspect, the present invention provides a computer program product, comprising a computer program / instruction, which implements the above method when executed by a processor.

[0020] Beneficial effects of the present invention: the present invention adopts generalized Pareto distribution and CGIG distribution to constrain the statistical characteristics of sea clutter, adopts a single-parameter probability model to constrain the sparsity of the ship's complex HRRP, adopts the KS minimum distance criterion to determine which composite Gaussian model to use to constrain the statistical characteristics of sea clutter, and adopts the AD test to determine the sparse parameter q of the ship's complex HRRP, which solves the problem of clutter model mismatch when the existing sparse optimization method is applied to the ship's complex HRRP estimation under the background of non-Gaussian sea clutter, and significantly reduces the performance loss of the ship's complex HRRP estimation caused by the clutter model mismatch. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 2 is a flow chart of a method for sparsely estimating complex HRRP of a ship matching the non-Gaussian characteristics of sea clutter provided by an embodiment of the present invention;

[0022] Figure 2 The embodiment of the present invention provides a sparse estimation method for applying a vector matrix model of a radar echo signal to match the non-Gaussian characteristics of sea clutter to realize an estimation framework diagram of a ship's complex HRRP;

[0023] FIG3 (a) and FIG3 (b) are curve diagrams showing the variation of the error and variance of the sparse estimation of complex HRRP of a ship with the signal-to-clutter ratio when the sea clutter obeys the generalized Pareto distribution, according to a sparse estimation method of complex HRRP of a ship matching the non-Gaussian characteristics of sea clutter provided by an embodiment of the present invention;

[0024] FIG3 (c) and FIG3 (d) are curve diagrams showing the variation of the error and variance of the sparse estimation of complex HRRP of a ship with shape parameters when the sea clutter obeys the generalized Pareto distribution according to a sparse estimation method of complex HRRP of a ship matching the non-Gaussian characteristics of sea clutter provided by an embodiment of the present invention;

[0025] FIG3 (e) and FIG3 (f) are curve diagrams showing the variation of the sparse parameter of the sparse estimation error and variance of the ship complex HRRP when the sea clutter obeys the generalized Pareto distribution in a ship complex HRRP sparse estimation method matching the non-Gaussian characteristics of sea clutter provided by an embodiment of the present invention;

[0026] FIG4 (a) and FIG4 (b) are curve diagrams showing the variation of the error and variance of the sparse estimation of complex HRRP of a ship with the signal-to-clutter ratio when the sea clutter obeys the CGIG distribution according to a sparse estimation method of complex HRRP of a ship matching the non-Gaussian characteristics of sea clutter provided by an embodiment of the present invention;

[0027] FIG4( c ) and FIG4( d ) are curve diagrams showing the variation of the error and variance of the sparse estimation of complex HRRP of a ship with shape parameters when the sea clutter obeys the CGIG distribution according to a sparse estimation method of complex HRRP of a ship matching the non-Gaussian characteristics of sea clutter provided by an embodiment of the present invention;

[0028] FIG4(e) and FIG4(f) are curve diagrams showing the variation of the sparse parameter of the sparse estimation error and variance of the ship complex HRRP when the sea clutter obeys the CGIG distribution in a ship complex HRRP sparse estimation method matching the non-Gaussian characteristics of sea clutter provided by an embodiment of the present invention;

[0029] FIG5 (a) is an experimental diagram of measured data at a KS distance provided by an embodiment of the present invention;

[0030] Figure 5(b) is the estimation result of the sparse estimation method matching the generalized Pareto distribution characteristics of Figure 5(a);

[0031] Figure 5(c) is the estimation result of the sparse estimation method matching the CGIG distribution characteristics in Figure 5(a);

[0032] Figure 5 (d) is the estimation result of Figure 5 (a) using the SLIM method;

[0033] FIG6 (a) is another experimental diagram of measured data at a KS distance provided by an embodiment of the present invention;

[0034] Figure 6 (b) is the estimation result of the sparse estimation method matching the generalized Pareto distribution characteristics of Figure 6 (a);

[0035] Figure 6 (c) shows the estimation result of the sparse estimation method for matching the CGIG distribution characteristics in Figure 6 (a);

[0036] Figure 6(d) is the estimation result of Figure 6(a) using the SLIM method. DETAILED DESCRIPTION

[0037] The present invention is further described in detail below with reference to specific embodiments, but the embodiments of the present invention are not limited thereto.

[0038] Embodiment 1

[0039] See also Figure 1 , Figure 1 1 is a flow chart of a method for sparsely estimating complex HRRP of a ship matching the non-Gaussian characteristics of sea clutter provided by an embodiment of the present invention, which includes:

[0040] Step 1: Get the raw radar echo data.

[0041] In this embodiment, it is assumed that the radar broadband pulse The bandwidth is , the carrier frequency is The down-converted signal of the received echo is , t The time when the radar signal is transmitted or received. right and Sampling can produce a discrete sequence and reference signal , n is the sampling point number, where is the width of the distance cell, is the speed of light. In order to reduce the range sidelobes, a window function is needed in the frequency domain, and pulse compression is achieved in the frequency domain through Discrete Fourier Transform (DFT);

[0042]

[0043] in Indicates the inverse , Represents the reference signal After performing DFT, take its complex conjugate, Represents the reference signal After performing DFT, take the square of its modulus, is the window function, and yes Length. is the point spread function to represent pulse compression.

[0044] When the high-resolution sea surveillance radar works in scanning mode, the radar echo of the ship is expressed as:

[0045]

[0046] in is the complex HRRP of the ship along the distance unit, For the For a high range resolution radar with a range cell of several meters, the time delay of a strong scatterer in a range cell is can be assumed to be an integer . , They are respectively the starting position and the ending position of the distance interval where the ship's complex HRRP is located.

[0047] Step 2: Use the threshold detection method to obtain the distance interval where the ship's complex HRRP is located on the original ship radar echo data.

[0048] In the application of target size feature extraction based on HRRP, in order to adapt to the range resolution capability of the radar and the scale characteristics of the target, the window pane of the radar receiving echo is usually discretized in units of radar range resolution units. The length of the intercepted window pane of the radar receiving echo needs to be greater than the length of the target echo. Therefore, in addition to the echo signal of the target, the one-dimensional range image of the radar also includes information such as noise and sea clutter. Threshold detection can be used to obtain the distance interval where the ship's complex HRRP is located. That is, if the radar echo power of a certain range unit is greater than a certain value, it can be considered that the range unit may be located in the distance interval where the ship's complex HRRP is located.

[0049] In some embodiments, step 2 specifically includes:

[0050] Step 2-1: The threshold used is the average power of high-resolution sea clutter, specifically:

[0051] in, , are the shape parameters and scale parameters of the K distribution fitting sea clutter determined by the moment estimation method.

[0052] Step 2-2: Remember is the radar echo power at the kth distance unit, and records all The continuous interval , are the starting and ending positions of the continuous interval respectively.

[0053] Step 2-3: The distance interval where the ship's HRRP is located It is expressed as:

[0054]

[0055] in, Representation interval Length.

[0056] Step 3: Perform vector matrix representation on the radar echo data within the distance interval.

[0057] The radar echo data is expressed as a vector matrix, that is, the radar echo data is expressed as The purpose is to express the problem of estimating the complex HRRP of a ship from radar echoes as a sparse reconstruction problem that can be solved using sparse optimization methods, where is the observation vector, is the perception matrix, is the sparse vector to be estimated, is the clutter interference vector.

[0058] In some embodiments, step 3 specifically includes:

[0059] Step 3-1: The echo data of the radar within the distance interval is expressed as:

[0060]

[0061]

[0062] in, Indicates a single pulse radar echo, which is a ship echo and high resolution sea clutter The superposition results; , , High resolution sea clutter.

[0063] Step 3-2: Convert the complex vector sequence Convert to a one-dimensional complex-valued sequence:

[0064]

[0065] in, It is a single pulse radar echo.

[0066] Step 3-3: Represent the one-dimensional complex-valued sequence as a matrix-vector form:

[0067]

[0068] in, is the radar echo signal vector; represents the vector corresponding to the complex HRRP of the ship; is the interference vector, which is modeled using composite Gaussian distribution, including generalized Pareto distribution and CGIG distribution, that is, its probability density function is:

[0069] ,

[0070] ,

[0071] in, Distribution represents the probability density function of the generalized Pareto, CGIG distribution, The size of the interference vector The nth value of , are the shape parameters and scale parameters of the composite Gaussian distribution fitting sea clutter determined by the moment estimation method, and exp is the natural exponent;

[0072] is the extended PSF vector, which is expressed as:

[0073] ,in, The length is All-zero vector of ; for Length;

[0074] is the distance unit shift matrix, which is expressed as:

[0075]

[0076] in; for The center of and Represents an all-0 column vector and an all-0 row vector of length N-1 respectively; represents the identity matrix with N-1 rows and N-1 columns, represents the identity matrix of N rows and N columns; A is the size of The perception matrix, L is the distance interval where the ship's complex HRRP is located Length.

[0077] Step 4: Use the KS minimum distance criterion to select a composite Gaussian model that matches the statistical characteristics of the interference vector in the vector matrix model. Calculate the KS minimum distance between the clutter and the generalized Pareto distribution and the CGIG distribution, and use the distribution with the smaller KS minimum distance as the composite Gaussian model of the clutter.

[0078] In some embodiments, the KS minimum distance is calculated as follows:

[0079]

[0080] in, , represents the amplitude cumulative density function of the composite Gaussian model. When it is a generalized Pareto distribution,

[0081]

[0082] When it is CGIG distribution,

[0083]

[0084] Among them, the interference vector Sort the values ​​in to get the sequence , is the nth value in the sequence.

[0085] Step 5: Use the sparse optimization method corresponding to the composite Gaussian model to estimate the complex HRRP of the ship.

[0086] See also Figure 2 , Figure 2 The embodiment of the present invention provides a framework diagram for estimating the complex HRRP of a ship by applying a sparse optimization method matching the non-Gaussian characteristics of sea clutter to a vector matrix model of a radar echo signal.

[0087] In the sparse optimization method for matching the non-Gaussian characteristics of sea clutter, since the sparse parameter q values ​​corresponding to different ships are different, it is necessary to select a suitable sparse parameter q value when sparsely estimating the vector matrix model. The value range of the sparse parameter q is [0.02 0.05:0.05:1].

[0088] In some embodiments, step five includes:

[0089] Step 5-1: Calculate the ship complex HRRP estimation results corresponding to each value in the q value interval:

[0090] In this embodiment, the estimation method is as follows:

[0091] The sparse optimization method for matching the non-Gaussian characteristics of sea clutter uses MAP to estimate the complex HRRP of the ship, namely:

[0092]

[0093] in, is the ship complex HRRP estimated using the maximum a posteriori probability; is the ship complex HRRP vector The joint probability density function of , which uses a single-parameter random distribution, is , is the sparse parameter, is the ship complex HRRP vector The lth value of ; is the joint probability density function of the interference vector, that is , when the composite Gaussian model in step 4 is selected as the generalized Pareto distribution, S is GP, for , which is the probability density function of the generalized Pareto distribution; when the composite Gaussian model is selected as the CGIG distribution, S is CGIG, for , which is the probability density function of CGIG distribution; N is the radar echo signal vector Length, for The nth value of is the row vector corresponding to the nth row of the perception matrix A.

[0094] Will , Bring in The following optimization problem can be obtained:

[0095] When the composite Gaussian model in step 4 is selected as the generalized Pareto distribution, the optimization problem is expressed as:

[0096]

[0097]

[0098] When the composite Gaussian model in step 4 is selected as CGIG distribution, the optimization problem is expressed as:

[0099]

[0100]

[0101] in, is the estimation result of the ship's complex HRRP when the sparse parameter is q; yes The lth value of .

[0102] The gradient decomposition method is used to solve the above optimization problem.

[0103] Specifically, the steps include:

[0104] 5-1-1: Re-HRRP to Ship To initialize:

[0105] In this embodiment, a matched filter is applied to initialize the algorithm, i.e.

[0106]

[0107] in, yes The lth value of is the ship complex HRRP vector Initialization vector of The perception matrix The column vector corresponding to the lth column of ; H is the conjugate transpose operator.

[0108] 5-1-2: Iterative update of ship HRRP :

[0109] In this embodiment, when the composite Gaussian model in step 4 is selected as the generalized Pareto distribution, the ship's composite HRRP The iteration formula is:

[0110]

[0111] in, , Ship complex HRRP No. m +1, no. m The vector of iterations, m is the number of iterations; is the weight matrix

[0112] , yes The lth value of , diag represents the diagonalization operation of the vector; They are respectively the prediction deviation function and the constraint function, expressed as:

[0113]

[0114] When the composite Gaussian model in step 4 is selected as CGIG distribution, the ship's complex HRRP The iteration formula is:

[0115]

[0116] in, They are adaptive gain factor, dynamic step size factor, and projection residual function, expressed as:

[0117]

[0118]

[0119]

[0120] 5-1-3: Determine the iteration termination condition:

[0121] In this embodiment, when the ship restores HRRP Iterate 20 times or satisfy Stop iteration when As the estimated value of the complex HRRP of the ship corresponding to the sparse parameter q ,in is a small positive number.

[0122] Step 5-2: Calculate the AD test statistic corresponding to each sparse parameter q. Specifically, the steps include:

[0123] Step 5-2-1: Determine the residual sequence of the ship's complex HRRP estimate .

[0124] In this embodiment, let is the residual of the ship complex HRRP estimation when the sparse parameter is q, that is ; for vector Sort the absolute values ​​of the numbers in to get the sequence , This is the nth value of the sequence.

[0125] Step 5-2-2: Calculate the AD test statistic corresponding to each sparse parameter q.

[0126] In this embodiment, the statistic of the AD test corresponding to the sparse parameter q is:

[0127]

[0128] Among them, when the composite Gaussian model in step 4 is selected as the generalized Pareto distribution, is the amplitude cumulative density function of the generalized Pareto distribution. When the composite Gaussian model in step 4 is selected as the CGIG distribution, is the cumulative density function of the amplitude of CGIG distribution.

[0129] Step 5-3: The sparse parameter q of the ship complex HRRP is determined as

[0130] .

[0131] Step 5-4: Sparse parameters The corresponding This is the final ship complex HRRP estimation vector.

[0132] Embodiment 2

[0133] Corresponding to the above-mentioned ship complex HRRP sparse estimation method matching the non-Gaussian characteristics of sea clutter, the embodiment of the present application further provides a ship complex HRRP sparse estimation system matching the non-Gaussian characteristics of sea clutter, including:

[0134] A radar signal receiving unit, used to obtain raw radar echo data;

[0135] A threshold detection unit, used for applying a threshold detection method to the original radar echo data to obtain a distance interval where the ship's complex HRRP is located;

[0136] A signal processing unit, used for representing the radar echo signal within the distance interval as a vector matrix model;

[0137] A model selection unit, used for selecting a composite Gaussian model that matches the statistical characteristics of the interference vector in the vector matrix model by using a KS minimum distance criterion;

[0138] The sparse estimation unit is used to estimate the complex HRRP of the ship by adopting the sparse optimization method corresponding to the composite Gaussian model.

[0139] Embodiment 3

[0140] Corresponding to the above-mentioned method for sparse estimation of complex HRRP of a ship that matches the non-Gaussian characteristics of sea clutter, an embodiment of the present application also provides a computer-readable storage medium on which program instructions are stored. When the program instructions are executed, a sparse estimation of complex HRRP of a ship that matches the non-Gaussian characteristics of sea clutter is implemented.

[0141] Embodiment 4

[0142] Corresponding to the above-mentioned method for sparse estimation of complex HRRP of a ship that matches the non-Gaussian characteristics of sea clutter, an embodiment of the present application also provides a computer program product, including a computer program / instruction, which, when executed by a processor, implements a sparse estimation of complex HRRP of a ship that matches the non-Gaussian characteristics of sea clutter.

[0143] In addition, although the present application is described in the context of functional modules, it should be understood that, unless otherwise specified, one or more of the functions and / or features described may be integrated into a single physical device and / or software module, or one or more functions and / or features may be implemented in a separate physical device or software module. It is also understood that a detailed discussion of the actual implementation of each module is unnecessary for understanding the present application. More specifically, in view of the properties, functions, and internal relationships of the various functional modules in the device disclosed herein, the actual implementation of the module will be understood within the conventional skills of the engineer. Therefore, those skilled in the art can implement the present application set forth in the claims without excessive experimentation using ordinary techniques. It is also understood that the specific concepts disclosed are merely illustrative and are not intended to limit the scope of the present application, which is determined by the full scope of the attached claims and their equivalents.

[0144] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for enabling an electronic device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0145] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0146] Verification example:

[0147] The beneficial effects of the invention are further illustrated below through simulation experiments.

[0148] 1. Simulation conditions:

[0149] The simulation experiment of this embodiment is carried out in an environment where the computer configuration is core i7 2.6GHZ, the memory is 8G, the WINDOWS 10 system and the computer software configuration is Matlab R2022b.

[0150] 2. Simulation content and result analysis:

[0151] The simulation experiment of this embodiment uses the sparse optimization method for matching the non-Gaussian characteristics of sea clutter proposed in this application and the existing sparse optimization method to carry out ship complex HRRP estimation experiments in simulation data and measured data. Specifically, when the composite Gaussian model is selected as the generalized Pareto distribution, the sparse estimation method for matching the non-Gaussian characteristics of sea clutter is called the ship complex HRRP sparse estimation method matching the generalized Pareto distribution clutter characteristics. When the composite Gaussian model is selected as the CGIG distribution, the sparse estimation method for matching the non-Gaussian characteristics of sea clutter is called the ship complex HRRP sparse estimation method matching the CGIG distribution clutter characteristics, so as to compare the influence of the adaptability of the complex Gaussian model to the probability characteristics of sea clutter on the sparse optimization results.

[0152] In this simulation experiment, it is assumed that the radar is an X-band high-resolution radar, which transmits a linear frequency modulation (LFM) pulse with a time width of 43.2 microseconds and a bandwidth of 200MHz. The sampling frequency of the baseband signal is 200MHz, and the pulse repetition frequency (PRF) of the radar is 1800Hz. The transmitted LFM waveform is positive frequency modulation. When the pulse is compressed, a Hamming window is added in the time domain, and pulse compression is realized in the frequency domain. The farther the position from the main lobe, the lower the level of the range side lobe. The first range side lobe is about -31.27dB. The scale parameter of the sea clutter is set to =5, and the sparsity parameter q takes values ​​in the interval [0.02 0.05:0.05:1]. 10,000 independent experiments are performed in each specified parameter set of the average signal-to-clutter ratio (SCR), sea clutter shape parameter, and sparsity parameter to calculate the mean and variance of the relative error of each method estimating 10,000 complex HRRPs of ships. This embodiment conducts two sets of simulation experiments. The simulation experiments corresponding to Figures 3 (a) to 3 (f) use data that obeys the generalized Pareto distribution as sea clutter data, and the simulation experiments corresponding to Figures 4 (a) to 4 (f) use data that conforms to the CGIG distribution as sea clutter data. This setting is to compare the influence of the adaptability of the complex Gaussian model to the probability characteristics of sea clutter on the sparse estimation results.

[0153] The scale parameter of the generalized Pareto distribution is set to 0.2; Figure 3(a) and Figure 3(b) are the shape parameters The average relative error and variance of the relative error of the complex HRRP estimation of the ship when the signal-to-noise ratio is set to 10, the sparsity parameter q is set to 0.2, and the signal-to-noise ratio is in the interval [10:2:50]. The horizontal axis is the variation range of the signal-to-noise ratio in dB, and the vertical axis is the average relative error and the variance of the relative error, respectively. Figure 3 (c) and Figure 3 (d) are the average relative error and variance of the relative error of the complex HRRP estimation of the ship when the signal-to-noise ratio is set to 20dB, the sparsity parameter is set to 0.2, and the shape parameter is in the interval [1 2:2:20]. The horizontal axis is the variation range of the shape parameter, and the vertical axis is the average relative error and the variance of the relative error, respectively. Figure 3 (e) and Figure 3 (f) are the average relative error and variance of the relative error of the complex HRRP estimation of the ship when the signal-to-noise ratio is set to 20dB, the sparsity parameter is set to 0.2, and the shape parameter is in the interval [1 2:2:20]. The horizontal axis is the variation range of the shape parameter, and the vertical axis is the average relative error and the variance of the relative error, respectively. The average relative error and variance of the relative error of the ship complex HRRP estimation when the sparse parameter q is set to 10 and takes values ​​in the interval [0.02 0.05:0.05:1]. The horizontal axis is the range of the sparse parameter, and the vertical axis is the average relative error and the variance of the relative error. In the figure, the green straight line of all pictures is the change curve of the average relative error and the variance of the relative error of the ship complex HRRP estimation using the sparse optimization method matching the clutter characteristics of the generalized Pareto distribution, the pink diamond straight line is the change curve of the average relative error and the variance of the relative error of the ship complex HRRP estimation using the sparse optimization method matching the clutter characteristics of the generalized CGIG distribution, and the red triangle line is the change curve of the average relative error and the variance of the relative error of the ship complex HRRP estimation using the existing SLIM method.

[0154] It can be seen from Figure 3 (a) to Figure 3 (f) that: First, in all cases, the sparse estimation method for ship complex HRRP estimation matching the non-Gaussian characteristics of sea clutter proposed in this application has better estimation performance than the SLIM method. In fact, since the SLIM method uses Gaussian distribution as the interference background, it will inevitably suffer performance loss when applied to ship complex HRRP estimation. However, the proposed sparse optimization method matching the non-Gaussian characteristics of sea clutter can significantly reduce the impact of the non-Gaussian characteristics of sea clutter. Second, the sparse estimation method for matching the clutter characteristics of the generalized Pareto distribution proposed in this application has better estimation performance than the sparse estimation method for matching the clutter characteristics of the CGIG distribution except in the case of high signal-to-clutter ratio. This is because the sea clutter used in the experiment obeys the generalized Pareto distribution, so the sparse estimation method for matching the clutter characteristics of the generalized Pareto distribution is more suitable for the experimental conditions and has better estimation performance. However, under the condition of high signal-to-clutter ratio, the influence of clutter on target signal estimation is weakened, so the sparse estimation method for matching the clutter characteristics of the CGIG distribution is better than the sparse estimation method for matching the clutter characteristics of the generalized Pareto distribution. Third, the sparse estimation method for ship complex HRRP estimation proposed in this application will improve its estimation performance as the signal-to-clutter ratio increases. Fourth, the sparse estimation method for ship complex HRRP estimation proposed in this application is little affected by the non-Gaussian characteristics of sea clutter, and the estimation performance remains basically unchanged as the shape parameters of sea clutter increase. Fifth, the estimation performance of the sparse estimation method for ship complex HRRP estimation proposed in the present application continues to improve as the sparse parameter q decreases, which means that the sparser the complex HRRP of the ship, the better the estimation performance of the sparse estimation method for ship complex HRRP estimation proposed in the present application.

[0155] The scale parameter of the CGIG distribution is set to 10; Figure 4(a) and Figure 4(b) are the shape parameters The average relative error and variance of the relative error of the complex HRRP estimation of the ship when the signal-to-noise ratio is set to 0.2, the sparsity parameter q is set to 0.2, and the signal-to-noise ratio is in the interval [10:2:50]. The horizontal axis is the variation range of the signal-to-noise ratio in dB, and the vertical axis is the average relative error and the variance of the relative error, respectively. Figure 4 (c) and Figure 4 (d) are the average relative error and variance of the relative error of the complex HRRP estimation of the ship when the signal-to-noise ratio is set to 20dB, the sparsity parameter is set to 0.2, and the shape parameter is in the interval [0.1 0.5:0.5:5]. The horizontal axis is the variation range of the shape parameter, and the vertical axis is the average relative error and the variance of the relative error, respectively. Figure 4 (e) and Figure 4 (f) are the average relative error and variance of the relative error of the complex HRRP estimation of the ship when the signal-to-noise ratio is set to 20dB, the shape parameter is set to 0.2, and the shape parameter is in the interval [0.1 0.5:0.5:5]. The horizontal axis is the variation range of the shape parameter, and the vertical axis is the average relative error and the variance of the relative error, respectively. The average relative error and variance of the ship complex HRRP estimation when the sparse parameter q is set to 0.2 and takes values ​​in the interval [0.020.05:0.05:1]. The horizontal axis is the range of variation of the sparse parameter, and the vertical axis is the average relative error and variance of the relative error. In the figure, the green straight line of all pictures is the variation curve of the average relative error and variance of the relative error of the ship complex HRRP estimation using the sparse estimation method matching the clutter characteristics of the CGIG distribution, the pink diamond straight line is the variation curve of the average relative error and variance of the relative error of the ship complex HRRP estimation using the sparse estimation method matching the clutter characteristics of the generalized Pareto distribution, and the red triangle line is the variation curve of the average relative error and variance of the relative error of the ship complex HRRP estimation using the existing SLIM method.

[0156] From Figures 4(a) to 4(f), we can find conclusions similar to those in Figures 3(a) to 3(f), but the difference is that the sparse estimation method for matching the clutter characteristics of the CGIG distribution proposed in the present application has better estimation performance than the sparse estimation method for matching the clutter characteristics of the generalized Pareto distribution in all cases. This is because the sea clutter used in the experiment obeys the CGIG distribution, so the sparse estimation method for the clutter characteristics of the CGIG distribution is more suitable for the experimental conditions and has better estimation performance.

[0157] Further, FIG5(a) is an amplitude diagram of radar clutter data (unit: dB). FIG5(b) is an amplitude diagram of the complex HRRP estimation result of the ship using the sparse estimation method for matching the clutter characteristics of the generalized Pareto distribution proposed in this application (unit: dB). FIG5(c) is an amplitude diagram of the complex HRRP estimation result of the ship using the sparse estimation method for matching the clutter characteristics of the CGIG distribution proposed in this application (unit: dB). FIG5(d) is an amplitude diagram of the complex HRRP estimation result of the ship using the SLIM method (unit: dB).

[0158] The sea clutter data in the radar clutter in Figures 5 (a) to 5 (d) have a KS distance of 0.0311 with the generalized Pareto distribution and a KS distance of 0.0370 with the CGIG distribution. According to the method proposed in this application, the minimum KS distance criterion is used to judge that the sparse estimation method that matches the clutter characteristics of the generalized Pareto distribution should have a better estimation result for the complex HRRP of the ship. From the actual estimation results, the sparse estimation method that matches the clutter characteristics of the CGIG distribution in Figure 5 (c) has a serious overestimation. The sparse estimation method that matches the clutter characteristics of the generalized Pareto distribution in Figure 5 (b) is significantly better, and the estimation results of Figures 5 (b) and 5 (c) are better than the estimation results of the SLIM method in Figure 5 (d), which proves the effectiveness of the method proposed in this application.

[0159] The sea clutter data in the radar clutter in Figures 6 (a) to 6 (d) have a KS distance of 0.0043 with the generalized Pareto distribution and a KS distance of 0.0042 with the CGIG distribution. According to the method proposed in this application, the minimum KS distance criterion is used to judge that the sparse estimation method matching the clutter characteristics of the CGIG distribution should have a better estimation result for the complex HRRP of the ship. From the actual estimation results, the sparse estimation method matching the clutter characteristics of the CGIG distribution in Figure 6 (c) can effectively estimate the strong scatterers of the ship, but the sparse estimation method matching the clutter characteristics of the generalized Pareto distribution in Figure 6 (b) has poor estimation performance for the strong scatterers of the ship, and the estimation results of Figures 6 (b) and 6 (c) are better than the estimation results of the SLIM method in Figure 6 (d), which proves the effectiveness of the method proposed in this application.

[0160] The above contents are further detailed descriptions of the present invention in combination with specific preferred embodiments, and it cannot be determined that the specific implementation of the present invention is limited to these descriptions. For ordinary technicians in the technical field to which the present invention belongs, several simple deductions or substitutions can be made without departing from the concept of the present invention, which should be regarded as falling within the scope of protection of the present invention.

Claims

1. The ship complex HRRP estimation method matching the non-Gaussian characteristics of sea clutter includes: Step 1: Obtain original radar echo data; Step 2: Using a threshold detection method on the original radar echo data to obtain the distance interval where the ship's complex HRRP is located; Step 3, representing the radar echo signal within the distance interval as a vector matrix model; Step 4, using the KS minimum distance criterion to select a composite Gaussian model that matches the statistical characteristics of the clutter interference vector in the vector matrix model; Step 5: Use the sparse optimization method corresponding to the composite Gaussian model to estimate the complex HRRP of the ship; The clutter interference vector in the vector matrix model is modeled using a composite Gaussian distribution, specifically including a generalized Pareto distribution and a composite Gaussian inverse Gaussian distribution; a composite Gaussian model that matches the statistical characteristics of the clutter interference vector in the vector matrix model is selected using a KS minimum distance criterion, the KS minimum distances between the clutter and the generalized Pareto distribution and the composite Gaussian inverse Gaussian distribution are calculated respectively, and the distribution with a smaller KS minimum distance is selected as the composite Gaussian model of the clutter; It also includes maximum a posteriori probability estimation and the use of gradient decomposition methods to solve optimization problems; The joint probability density function of the complex HRRP vectors of ships in the maximum a posteriori probability estimation adopts a single-parameter random distribution, which constrains the sparsity of the complex HRRP vectors of ships in the form of an exponential function, where the sparsity parameter is used to control the sparsity of the distribution; The sparse parameter is determined by finding a parameter value that minimizes the AD test statistic within a preset value range, wherein the AD test statistic is used to measure the matching degree between the estimation result and the selected composite Gaussian model; specifically: Determine the residual sequence of the ship's complex HRRP estimate Let d q is the residual error of the ship complex HRRP estimation when the sparse parameter is q. q Sort the absolute values ​​of the numbers in to get the sequence That is the nth value of the sequence; Calculate the AD test statistic for each sparse parameter q; The statistic of the AD test corresponding to the sparse parameter q is: Among them, when the composite Gaussian model in step 4 is selected as the generalized Pareto distribution, CDF is the amplitude cumulative density function of the generalized Pareto distribution; when the composite Gaussian model in step 4 is selected as the composite Gaussian inverse Gaussian distribution, CDF is the amplitude cumulative density function of the composite Gaussian inverse Gaussian distribution.

2. The method according to claim 1, characterized in that: The original radar echo data is obtained in step 1 by receiving the pulse signal transmitted by the radar and receiving the echo signal reflected from the target.

3. The method according to claim 1 or 2, characterized in that: In step 2, a threshold detection method is used to obtain the distance interval where the ship's complex HRRP is located, including calculating the threshold of the radar echo power and determining the distance unit where the radar echo power exceeds the threshold as the distance interval where the ship's complex HRRP is located.

4. The method according to claim 1, characterized in that: The vector matrix model in step 3 also includes an observation vector, a perception matrix, and a sparse vector to be estimated.

5. The ship complex HRRP estimation system matching the non-Gaussian characteristics of sea clutter is characterized by include: A radar signal receiving unit, used to obtain raw radar echo data; A threshold detection unit, used for applying a threshold detection method to the original radar echo data to obtain a distance interval where the ship's complex HRRP is located; A signal processing unit, used for representing the radar echo signal within the distance interval as a vector matrix model; A model selection unit, used for selecting a composite Gaussian model that matches the statistical characteristics of the clutter interference vector in the vector matrix model by using a KS minimum distance criterion; A sparse estimation unit, used to estimate the complex HRRP of the ship by using the sparse optimization method corresponding to the composite Gaussian model; The clutter interference vector in the vector matrix model is modeled using a composite Gaussian distribution, specifically including a generalized Pareto distribution and a composite Gaussian inverse Gaussian distribution; a composite Gaussian model that matches the statistical characteristics of the clutter interference vector in the vector matrix model is selected using a KS minimum distance criterion, the KS minimum distances between the clutter and the generalized Pareto distribution and the composite Gaussian inverse Gaussian distribution are calculated respectively, and the distribution with a smaller KS minimum distance is selected as the composite Gaussian model of the clutter; It also includes maximum a posteriori probability estimation and the use of gradient decomposition methods to solve optimization problems; The joint probability density function of the complex HRRP vectors of ships in the maximum a posteriori probability estimation adopts a single-parameter random distribution, which constrains the sparsity of the complex HRRP vectors of ships in the form of an exponential function, where the sparsity parameter is used to control the sparsity of the distribution; The sparse parameter is determined by finding a parameter value that minimizes the AD test statistic within a preset value range, wherein the AD test statistic is used to measure the matching degree between the estimation result and the selected composite Gaussian model; specifically: Determine the residual sequence of the ship's complex HRRP estimate Let d q is the residual error of the ship complex HRRP estimation when the sparse parameter is q. q Sort the absolute values ​​of the numbers in to get the sequence That is the nth value of the sequence; Calculate the AD test statistic for each sparse parameter q; The statistic of the AD test corresponding to the sparse parameter q is: Among them, when the compound Gaussian model in the model selection unit is selected as the generalized Pareto distribution, CDF is the amplitude cumulative density function of the generalized Pareto distribution; when the compound Gaussian model in the model selection unit is selected as the compound Gaussian inverse Gaussian distribution, CDF is the amplitude cumulative density function of the compound Gaussian inverse Gaussian distribution.

6. A computer-readable storage medium, characterized in that: Program instructions are stored thereon, and when the program instructions are executed, the method according to any one of claims 1 to 4 is implemented.

7. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the method according to any one of claims 1 to 4 is implemented.

Images