A high-maneuvering target motion parameter estimation and coherent accumulation detection method

Abstract

The application discloses a radar signal processing technical field and a high-maneuvering target motion parameter estimation and coherent accumulation detection method. The high-maneuvering target motion parameter estimation and coherent accumulation detection method comprises the following steps: performing pulse compression, neighborhood cross-correlation processing and extracting a self-correlation item on a radar echo of a high-maneuvering target; designing a perception matrix, reconstructing a sparse signal by using a Bayesian compressive sensing algorithm and extracting high-order motion parameters of the target; performing high-order phase compensation, correcting a linear phase, obtaining coherent accumulation and performing CFAR detection; judging the number of high-order motion parameters of the target, completing coherent accumulation detection and outputting a detection result. The high-maneuvering target motion parameter estimation and coherent accumulation detection method has high estimation precision, low calculation complexity and can be effectively applied to single-target and multi-target scenes.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of radar signal processing, in particular to a high-maneuvering target motion parameter estimation and coherent accumulation detection method. BACKGROUND

[0002] In recent years, with the rapid development of aviation technology, the detection of high-maneuvering targets with high threat has gradually become an important challenge in the field of radar detection. Currently, known high-maneuvering targets with high threat include advanced jet fighters, supersonic glide bombs, near-space hypersonic vehicles, and various tactical cruise missiles. Generally speaking, coherent accumulation can improve the signal-to-noise ratio of the echo of a high-maneuvering target and is the most effective way to achieve better detection and tracking performance. However, for a high-maneuvering target (defined here as a motion target with speed, acceleration, and jerk), it has the following characteristics: on the one hand, its motion trajectory is a complex high-order function, and the distance envelope is difficult to effectively extract; on the other hand, the echo of a high-maneuvering target contains high-order motion phases, and the motion parameters such as acceleration and jerk need to be effectively estimated to realize phase compensation and coherent accumulation.

[0003] Currently, existing high-maneuvering target motion parameter estimation and coherent accumulation detection methods mainly fall into two categories: one category is represented by the Generalized Radon Fourier Transform (GRFT) detection method, which realizes accurate compensation of high-maneuvering target motion parameters through multi-dimensional traversal search. However, this category of detection methods has the main problems of high computational complexity, long calculation time, and interference from blurred sidelobes, making it difficult to work effectively in a multi-target scenario. The other category of detection methods is the time-frequency analysis method based on Lv’s distribution (LVD) or fractional Fourier transform. The estimation accuracy of this category of detection methods is limited by the time-frequency resolution and is difficult to effectively improve the parameter estimation accuracy. In addition, this category of detection methods is easily affected by cross terms in a multi-target scenario, making it difficult to accurately extract the motion parameters of multiple targets. Both categories of motion parameter estimation and coherent accumulation detection methods have technical problems such as deficiencies and drawbacks. The above technical problems make it necessary to further improve the high-maneuvering target motion parameter estimation and coherent accumulation detection method, and therefore, there is an urgent need to design a high-maneuvering target motion parameter estimation and coherent accumulation detection method. SUMMARY

[0004] In order to solve the above problems, the purpose of the present application is to provide a high maneuvering target motion parameter estimation and coherent accumulation detection method, which solves the technical problems of high computational complexity, low estimation accuracy, and unsuitability for multi-target scenes in the existing Radon Fourier transform inspection method and the time-frequency analysis type detection method of Lue distribution or fractional Fourier transform, has high estimation accuracy, low computational complexity, and can be effectively applied to single-target and multi-target scenes.

[0005] In order to achieve the above purpose, the technical scheme of the present application is as follows:

[0006] As an aspect of the present application, a high maneuvering target motion parameter estimation and coherent accumulation detection method is provided, comprising the following steps:

[0007] S1. Pulse compression, neighborhood cross-correlation processing and extraction of autocorrelation terms are performed on the radar echo of the high maneuvering target;

[0008] S2. A perception matrix is designed, and a Bayesian compressed sensing algorithm is used to reconstruct a sparse signal to extract high-order motion parameters of the target;

[0009] S3. High-order phase compensation is performed, linear phase is corrected, coherent accumulation is obtained, and CFAR detection is performed;

[0010] S4. The number of high-order motion parameters of the target is judged, coherent accumulation detection is completed, and the detection result is output.

[0011] As a high maneuvering target motion parameter estimation and coherent accumulation detection method of the above aspect of the present application, wherein S1 comprises the following steps:

[0012] S11. Pulse compression is performed on the radar echo S r (τ′,t m ) of the high maneuvering target, and post-pulse compression echo S rc (τ′,t m ) is obtained;

[0013] S12. Neighborhood cross-correlation processing is performed on the echo data within the coherent processing time, and two-dimensional data χ(τ′,t m ) is obtained;

[0014] S13. Through non-coherent accumulation, the autocorrelation term is extracted from χ(τ′,t m ) as an observation signal for the sparse reconstruction process.

[0015] As a high maneuvering target motion parameter estimation and coherent accumulation detection method of the above aspect of the present application, wherein the radar echo S r (τ′,t m ) of the high maneuvering target in S11 is specifically as follows:

[0016]

[0017] wherein ξ 0，l represents the propagation coefficient of the lth target, λ0=c / f c represents the wavelength, and c represents the speed of light.

[0018] The expression of the pulse-echo S rc (τ′,t m ) in S11 is as follows:

[0019]

[0020] wherein B represents the bandwidth of the transmitted waveform, ξ 1，l represents the complex coefficient of the lth target after pulse compression.

[0021] As a high-maneuverability target motion parameter estimation and coherent accumulation detection method of the above aspect of the present application, the expression of the two-dimensional data χ(τ′,t m ) in S12 is as follows:

[0022]

[0023] wherein χ self (τ′,t m ) represents the autocorrelation term, and χ cross (τ′,t m ) represents the cross-correlation term.

[0024] The expression of the autocorrelation term is as follows:

[0025]

[0026] The expression of the coefficients of each order of the autocorrelation term is as follows:

[0027]

[0028]

[0029]

[0030] A 3，l = 3a 3，l T r ;

[0031] The expression of the cross-correlation term χ cross (τ′,t m ) is as follows:

[0032]

[0033] The expression of the coefficients of each order of the cross-correlation term is as follows:

[0034]

[0035]

[0036]

[0037]

[0038]

[0039] As a high maneuvering target motion parameter estimation and coherent accumulation detection method of the above aspects of the application, wherein S13 comprises the following steps:

[0040] S131. Take the absolute value of χ(τ', t m ) and do non-coherent accumulation along the slow time, and its expression is as follows:

[0041]

[0042] S132. Select the coordinate value τ'0 with the maximum absolute value in u(τ'), which is the peak position coordinate of the range envelope of the autocorrelation term χ self (τ', t m );

[0043] S133. Extract the slow time data y(t m ) = χ self (τ' = τ0, t m ) corresponding to the τ'0 coordinate from χ self (τ', t m ), as the observation signal in the sparse reconstruction process.

[0044] As a high maneuvering target motion parameter estimation and coherent accumulation detection method of the above aspects of the application, wherein S2 comprises the following steps:

[0045] S21. Design and construct the sensing matrix Φ of sparse reconstruction;

[0046] S22. Based on the sensing matrix Φ and the observation signal y(t m ), substitute the compressed sensing signal model y = Φx, and use the Bayesian compressed sensing algorithm to reconstruct a one-dimensional sparse signal vector x ∈ ℂ PQ×1 ;

[0047] S23. Convert the one-dimensional sparse signal vector x ∈ ℂ PQ×1 to a two-dimensional matrix X ∈ ℂ P×Q , wherein the elements in the matrix X satisfy

[0048] [X] p′，q′ = x q′+p′Q ;

[0049] S24. Estimate the position coordinates of L peaks from the two-dimensional matrix X∈￡ P×Q , denoted as According to the peak position, obtain the estimated values of the acceleration and jerk of the L targets, respectively

[0050]

[0051]

[0052] Output the estimated results of the motion parameters of the L targets And initialize l=1.

[0053] As a high-maneuvering target motion parameter estimation and coherent accumulation detection method of the above-mentioned aspects of the application, wherein S21 includes the following steps:

[0054] S211. Define a set representing the acceleration parameter range and a set representing the jerk parameter range;

[0055] The set of acceleration parameter ranges

[0056] The set of jerk parameter ranges

[0057] Wherein, P and Q represent the number of elements in the acceleration and jerk parameter sets respectively, and Δa2 and Δa3 represent the search step of the acceleration and jerk parameters respectively;

[0058] S212. According to the parameter search sets Ξ2, Ξ3, design the sensing matrix Φ∈￡ M×PQ of sparse reconstruction;

[0059] S213. Construct the sensing matrix Φ=(U T e W T ) T , wherein the element [Φ] m，q′+p′Q can be expressed as

[0060]

[0061] Wherein, define two matrices U∈￡ M×P and matrix W∈￡ M×Q , whose elements are respectively expressed as:

[0062]

[0063]

[0064] As one of the high-maneuvering target motion parameter estimation and coherent accumulation detection methods of the above-mentioned aspect of the present application, S3 comprises the following steps:

[0065] S31. Constructing a frequency domain compensation phase according to the estimated acceleration and jerk, and performing phase compensation on the pulse-range frequency domain radar echo s rc (f, t m ) of the target;

[0066] S32. Correcting the residual linear motion phase of the phase-compensated frequency domain echo signal by using a Keystone algorithm, and obtaining a coherent accumulation result in the "range time domain-azimuth frequency domain" by azimuth Fourier transform, and performing CFAR detection.

[0067] As one of the high-maneuvering target motion parameter estimation and coherent accumulation detection methods of the above-mentioned aspect of the present application, the expression of the pulse-range frequency domain of the high-maneuvering target pulse compression echo in S31 is as follows:

[0068]

[0069] Using the motion parameter estimation value of the lth target Constructing a compensation function h l (f, t m ), the expression of which is as follows:

[0070]

[0071] The expression of the compensated signal is s' rc (f, t m ) = s rc (f, i m )·h l (f, t m ).

[0072] As one of the high-maneuvering target motion parameter estimation and coherent accumulation detection methods of the above-mentioned aspect of the present application, S4 comprises the following steps:

[0073] S41. If l < L is satisfied, setting l = l + 1, and continuing to repeat the S3 step for the next target;

[0074] S42. If l < L is not satisfied, directly outputting the detection result of all targets.

[0075] By adopting the above technical solution, the present application has the following advantages:

[0076] The application provides a high-maneuvering target motion parameter estimation and coherent accumulation detection method, which utilizes neighborhood cross-correlation processing to correct envelope distance migration of the high-maneuvering target, does not need high-dimensional parameter searching, and has low calculation complexity and high calculation efficiency; in addition, the application utilizes a Bayesian compressive sensing method to effectively improve the precision of high-maneuvering target motion parameter estimation through sparse time-frequency representation, and a sparse reconstruction method is not affected by cross terms, and can still maintain good estimation performance in a multi-target scene; in summary, the application significantly improves the existing high-maneuvering target coherent accumulation detection and motion parameter estimation technology, can be widely applied to radar applications such as air search and space monitoring, can be effectively applied to motion parameter estimation and coherent accumulation of high-maneuvering targets, can effectively adapt to single-target and multi-target scenes, and has great application prospects in practical fields such as radar space monitoring and radar target detection. BRIEF DESCRIPTION OF DRAWINGS

[0077] Figure 1 A flowchart of the high-maneuvering target motion parameter estimation and coherent accumulation detection method of the application;

[0078] Fig. 2 is a schematic diagram of echo and neighborhood cross-correlation processing of a single high-maneuvering target according to the application;

[0079] Fig. 3 is a comparison diagram of motion parameter estimation performance of the application and the prior art in a single-target scene;

[0080] Figure 4 Fig. 4 is a comparison diagram of coherent accumulation detection probability of the application and the prior art in a single-target scene;

[0081] Fig. 5 is a comparison diagram of motion parameter estimation performance of the application and the prior art in a multi-target scene;

[0082] Fig. 6 is an effect diagram of super-resolution of the application in a multi-target scene. DETAILED DESCRIPTION

[0083] The technical solutions of the application are specifically described below in combination with the accompanying drawings of the specification. It should be noted that, in this document, relationship terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or sequence between the entities or operations. Moreover, the terms "include", "contain" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed or inherent to such process, method, article or device.

[0084] Suppose a single-transmitting and single-receiving radar, whose transmitting signal is a linear frequency modulation (LFM) waveform, and the expression of the transmitting signal is as follows:

[0085]

[0086] where f denotes the carrier frequency, τ denotes the fast time, t c m = mT r , m = 1, 2,..., M denotes the slow time, i.e., the azimuth time,

[0087] T r denotes the pulse repetition time, M denotes the number of coherent integration pulses, T p denotes the pulse width, μ denotes the chirp rate of the LFM waveform, denotes the window function.

[0088] Suppose there are L high-maneuvering targets in the observation domain, and their motion trajectories varying with the slow time are modeled as

[0089]

[0090] where r 0，l denotes the initial distance between the radar platform and the target, a 1，l , a 2，l , a 3，l denote the radial velocity, acceleration and jerk of the target, respectively.

[0091] The echoes of the L high-maneuvering targets after down-sampling can be expressed as

[0092]

[0093] where ξ 0，l denotes the propagation coefficient of the lth target, λ0= c / f c denotes the wavelength, and c denotes the speed of light.

[0094] The received signal after pulse compression can be expressed as

[0095]

[0096] where B denotes the bandwidth of the transmitted waveform, ξ 1，l denotes the complex coefficient of the lth target after pulse compression.

[0097] As shown in the above formula, the range envelope and the slow-time phase of the high-maneuvering target are both third-order polynomial functions varying with the slow time, which increases the difficulty of extracting the range envelope and the phase of the high-maneuvering target from the two-dimensional radar echoes to realize coherent integration and parameter estimation.

[0098] It is declared that the special meanings of the mathematical symbols used in the following text are as follows: a bold lowercase letter, such as x, denotes a vector, a bold uppercase letter, such as X, denotes a matrix, x n ​denotes the n-th element of the vector x, [x] m，n denotes the element of the m-th row, n-th column of the matrix X.

[0099] According to the above signal model, a high-maneuvering target motion parameter estimation and coherent accumulation detection method, specifically comprising the following steps as shown in Figure 1

[0100] S1. Pulse compression, adjacent cross correlation processing (ACCF) and extracting autocorrelation terms of the radar echo of the high-maneuvering target are performed;

[0101] S1 specifically comprises the following steps:

[0102] S11. Pulse compression is performed on the radar echo S r (τ′t m ) of the high-maneuvering target, to obtain the pulse-compressed echo S rc (τ′,t m );

[0103] The expression of the pulse-compressed echo S rc (τ′,t m ) in S11 is as follows:

[0104]

[0105] S12. Adjacent cross correlation processing is performed on the echo data within the coherent processing time, to obtain two-dimensional data χ(τ′,t m );

[0106] In order to extract the range profile of the high-maneuvering target from the two-dimensional echo data, adjacent cross correlation processing (ACCF) is performed on the multi-pulse echo data within the coherent processing time, that is, the adjacent two pulses are convolved, and the expression is as follows:

[0107]

[0108] wherein χ self (τ′,t m ) represents the autocorrelation term, and χ cross (τ′,t m ) represents the cross correlation term.

[0109] The expression of the autocorrelation term is

[0110]

[0111] The expression of the autocorrelation term of each order is as follows:

[0112] ​

[0113]

[0114]

[0115] A 3，l = 3a 3，l T r ;

[0116] The expression of the cross-correlation term χ cross (τ', t m ) is as follows:

[0117]

[0118] The expression of the coefficients of each order of the cross-correlation term is as follows:

[0119]

[0120]

[0121]

[0122]

[0123]

[0124] After the ACCF processing of the pulse compression echo of the high-maneuvering target in the coherent processing time, the envelope of the autocorrelation term is located in the same range cell, while the envelope of the cross-correlation term still exists range migration. That is, for the autocorrelation term, the high-order range migration is corrected, and therefore the envelope of the autocorrelation term is extracted for coherent accumulation and parameter estimation.

[0125] S13. Extract the autocorrelation term from χ(τ', t m ) as the observation signal of the sparse reconstruction process by non-coherent accumulation.

[0126] S13 includes the following steps:

[0127] S131. Take the absolute value of χ(τ', t m ) and perform non-coherent accumulation along the slow time, and the expression is as follows:

[0128]

[0129] S132. Select the coordinate value τ'0 with the maximum absolute value in u(τ') as the peak position coordinate of the range envelope of the autocorrelation term χ self (τ', t m );

[0130] S133. From χ self (τ', tm extract the slow-time data y(t m ) = χ self (τ' = τ0, t m ) as the observation signal in the sparse reconstruction process.

[0131] S2. Design a sensing matrix to reconstruct the sparse signal and extract the target high-order motion parameters by using a Bayesian compressive sensing algorithm;

[0132] S21. Design and construct a sensing matrix Φ for sparse reconstruction.

[0133] S21 specifically includes the following steps:

[0134] S211. Define a set of acceleration parameter ranges represented by a set of acceleration parameter ranges and a set of jerk parameter ranges represented by a set of jerk parameter ranges

[0135] where P and Q represent the number of elements in the acceleration and jerk parameter sets, respectively, and Δa2 and Δa3 represent the search step size, i.e., the estimation accuracy, of the acceleration and jerk parameters, respectively.

[0136] S212. Design a sensing matrix Φ ∈ ℂ M×PQ , also known as a dictionary matrix, based on the parameter search sets Ξ2 and Ξ3.

[0137] S213. Construct the sensing matrix Φ = (U T eW T ) T where the elements [Φ] m，q′+p′Q can be represented as

[0138]

[0139] where two matrices U ∈ ℂ M×P and a matrix W ∈ ℂ M×Q are defined, and their elements are represented as follows:

[0140]

[0141]

[0142] S22. Based on the sensing matrix Φ and the observation signal y(t m ) in step S133, substitute the compressive sensing signal model y = Φx, and use a Bayesian compressive sensing (BCS) algorithm to reconstruct a one-dimensional sparse signal vector x ∈ ℂ PQ×1 .

[0143] S23. Reconstruct the one-dimensional sparse signal vector x∈￡ PQ×1 Convert to a two-dimensional matrix X∈￡ P×Q , where the elements in the matrix X satisfy

[0144] [X] p′，q′ =x q′+p′Q ;

[0145] S24. From the two-dimensional matrix X∈￡ P×Q The position coordinates of L peaks are estimated and expressed as The estimated values ​​of acceleration and jerk of L targets are obtained according to the peak positions, which are:

[0146]

[0147]

[0148] Output the motion parameter estimation results of L targets And initialize l=1.

[0149] S3. Perform high-order phase compensation, correct linear phase, obtain coherent accumulation and perform CFAR detection;

[0150] S3 includes the following steps:

[0151] S31. Construct frequency domain compensation phase according to the acceleration and jerk estimated in step S24, and calculate the radar echo s in the pulse-range frequency domain. rc (f, t m ) for phase compensation;

[0152] The pulse-range frequency domain expression of the pulse pressure echo of a high-maneuvering target in S31 is as follows:

[0153]

[0154] Using the motion parameter estimate of the lth target Construct compensation function h l (f, t m ), which is expressed as follows:

[0155]

[0156] The signal expression after compensation is s′ rc (f, t m )=s rc (f, t m )·h l (f, t m ).

[0157] S32. For the high-order motion phase compensated frequency domain echo signal s' rc (f, t m ), the remaining linear motion phase is further corrected by using a Keystone algorithm, and the final coherent accumulation result is obtained in the "range time domain-azimuth frequency domain" through azimuth Fourier transform, and CFAR detection is performed.

[0158] S4. Determine the number of target high-order motion parameters, complete coherent accumulation detection, and output the detection result.

[0159] S4 includes the following steps:

[0160] S41. Determine whether l < L is satisfied, if satisfied, set l = l + 1 and return to step S3 for coherent accumulation detection of the next target;

[0161] S42. If l < L is not satisfied, directly output the detection result of all targets.

[0162] The present application will be further described by the following experimental embodiment, assuming that the system parameters and target parameters in a single target scene are as shown in Table 1:

[0163] Table 1: System parameters and target parameters of the present example simulation

[0164] System parameters Value (unit) Target parameter Value (unit) Center frequency f c ]]> 10 GHz initial distance r0 100 km Bandwidth B 10 MHz speed a1 1000 m / s Pulse width T p ]]> 20 μs acceleration a2 52 m / s 2 ]]> Pulse repetition interval T r ]]> 1.82 ms acceleration a3 18 m / s 3 ]]> Pulse repetition frequency f p ]]> 550 Hz Signal-to-noise ratio after pulse 5 dB Number of coherent pulses M 513

[0165] The system parameters set in the single target scene and the motion parameters of the high-maneuvering target are shown in Table 1, and the pulse compression echo of the high-maneuvering target is obtained by simulation using the above parameters, as shown in Figure 2(a). As can be seen from Figure 2(a), the distance envelope of the target has obvious range migration; the ACCF processing of the pulse compression echo of the high-maneuvering target is performed, and the processing result shown in Figure 2(b) is obtained. Further, the one-dimensional non-coherent accumulation result shown in Figure 2(c) is obtained by using the method of step S13 of the present application to extract the autocorrelation term from two-dimensional data as the observation signal for sparse reconstruction process. As can be seen from Figure 2(c), the autocorrelation term envelope is located in the distance unit number 680.

[0166] The envelope data of the ACCF autocorrelation term is extracted from this position, and the existing method and the method of the present application are used to estimate the motion parameters of the high-maneuvering target, and the estimation result is shown in Figure 3:

[0167] Figure 3(a) is an estimation result obtained by using the RFT method, from which it can be seen that the RFT method cannot effectively estimate the motion parameters of a high-maneuvering target; Figure 3(b) is an estimation result obtained by using the GRFT method, from which it can be seen that the method can estimate the motion parameters of a high-maneuvering target through the position of a maximum peak, but the method has high computational complexity, long time consumption, and the calculation time reaches 84.65s, and there are obvious fuzzy sidelobes in the estimation result; Figure 3(c) is an estimation result obtained by using the cyclic ACCF method, from which it can be seen that the method cannot effectively estimate the high-order motion parameters of a target under the condition of 5dB signal-to-noise ratio after pulse compression; Figure 3(d) is an estimation result obtained by using the ACCF-LVD method, from which it can be seen that the method can estimate the motion parameters of a high-maneuvering target through the position of a peak; Figure 3(e) is a motion parameter of a high-maneuvering target estimated by using the method; and Figure 3(f) is a coherent accumulation result obtained after compensating echo data by using the motion parameter estimation value obtained by using the method, from which it can be seen that the coherent accumulation result obtained by using the method has a narrower main lobe width than the ACCF-LVD method in Figure 3(d).

[0168] The above algorithms are quantitatively analyzed under different signal-to-noise ratios, and the quantitative analysis result is specifically as shown in Table 1. Figure 4 Figure 4 It can be seen from Table 1 that the method has obvious signal-to-noise ratio advantages compared with the cyclic ACCF, ACCF-LVD and RFT methods, and has a signal-to-noise ratio improvement of about 5dB compared with the ACCF-LVD method under the low signal-to-noise ratio environment of 0dB to 5dB after pulse compression; compared with the GRFT method, the method has obvious computational efficiency advantages, and the method takes only 0.41s under the same hardware conditions, which is much smaller than the GRFT method (takes 84.56s).

[0169] The method will be further described below through a multi-target scene embodiment, and the target motion parameters under the multi-target scene are specifically as shown in Table 2, and the signal-to-noise ratio is set to 5dB after pulse compression.

[0170] Table 2: Simulation system parameters and target parameters under a multi-target scene

[0171]

[0172] ​Fig. 5 is a pulse echo and motion parameter estimation result in a multi-target scene, wherein Fig. 5(a) is a pulse echo of multiple high-maneuvering targets; Fig. 5(b) is a target motion parameter estimated by using the GRFT method, from which it can be seen that the result only shows one target and the sidelobe corresponding to the target, because the GRFT method cannot effectively distinguish different targets and the sidelobe corresponding to each target in a multi-target scene due to the influence of the sidelobe, and therefore cannot effectively adapt to a multi-target scene; Fig. 5(c) and Fig. 5(d) are three-dimensional display and two-dimensional display of the multi-target motion parameter estimation result obtained by using the ACCF-LVD method, from which it can be seen that the ACCF-LVD method is affected by the cross term of the traditional time-frequency transform and cannot effectively estimate the motion parameter of the multi-target.

[0173] Table 3: Simulation system parameters and target parameters of the present example

[0174]

[0175]

[0176] In order to verify that the method has higher estimation resolution, the high-maneuvering target motion parameters shown in Table 3 are set. Fig. 6(a) and Fig. 6(b) are three-dimensional display and two-dimensional display of the multi-target motion parameter estimation result obtained by using the ACCF-LVD method, from which it can be seen that the method cannot effectively distinguish the two high-maneuvering targets and only one peak value is in the estimation result. Fig. 6(c) and Fig. 6(d) are three-dimensional display and two-dimensional display of the multi-target motion parameter estimation result obtained by using the high-maneuvering target motion parameter estimation and coherent accumulation detection method, from which it can be seen that the detection method breaks through the limitation of the traditional time-frequency resolution and effectively estimates the two different high-maneuvering targets.

[0177] In summary, the high-maneuvering target motion parameter estimation and coherent accumulation detection method has higher calculation efficiency compared with the traditional GRFT method, and can effectively adapt to a multi-target scene; in addition, it has higher estimation accuracy compared with the traditional time-frequency transform method (ACCF-LVD and cyclic ACCF) and significantly improves the estimation performance.

[0178] Finally, it should be pointed out that although the present application has been described with reference to the current specific embodiments, those skilled in the art should recognize that the above embodiments are only used to illustrate the present application and are not used as a limitation to the present application, and various equivalent changes or replacements can be made without departing from the concept of the present application, therefore, any changes or modifications to the above embodiments within the scope of the spirit of the present application will fall within the scope of the claims of the present application.

Claims

1. A method for motion parameter estimation and coherent accumulation detection of a highly maneuverable target, characterized in that: The following steps are involved: S1. Perform pulse compression and neighborhood cross-correlation processing on radar echoes of highly maneuverable targets and extract autocorrelation terms; S2. Design a sensing matrix and use the Bayesian compressed sensing algorithm to reconstruct sparse signals and extract high-order motion parameters of the target; S3. Perform high-order phase compensation, correct linear phase, obtain coherent accumulation and perform CFAR detection; S4. Determine the number of target high-order motion parameters, complete coherent accumulation detection and output the detection results.

2. The method for motion parameter estimation and coherent integration detection of a high-maneuverability target according to claim 1, characterized in that: Said S1 comprises the following steps: S11. Radar echo S for high maneuvering targets r (τ,t m ) to perform pulse compression and obtain the post-pulse compression echo S rc (τ,t m ); S12. Perform neighborhood cross-correlation processing on the echo data within the coherent processing time to obtain two-dimensional data χ(τ′, t m ); S13. Through non-coherent accumulation, from χ(τ′, t m ) to extract the autocorrelation term as the observation signal in the sparse reconstruction process.

3. The method for motion parameter estimation and coherent integration detection of a high-maneuverability target according to claim 2, characterized in that: The radar echo of the high maneuverability target in S11 is specifically as follows: r (τ,t m )Down: where ξ 0,l represents the propagation coefficient of the lth target, λ0=c / f c represents wavelength, c represents the speed of light; The pulse pressure echo S in S11 rc (τ,t m ) is as follows: Where B represents the bandwidth of the transmitted waveform, ξ 1,l represents the complex coefficient of the lth target pulse after compression.

4. The method for motion parameter estimation and coherent integration detection of a high-maneuverability target according to claim 3, characterized in that: The two-dimensional data χ(τ′,t m ) is as follows: where χ self (τ′,t m ) represents the autocorrelation term, χ cross (τ′,t m ) represents the cross-correlation term; The expression of the autocorrelation term is as follows: The expressions of the coefficients of each order of the autocorrelation term are as follows: A 3,l =3a 3,l T r ; Cross-correlation term χ cross (τ′,t m ) is as follows: The expressions of the coefficients of each order of the cross-correlation terms are as follows:

5. The method for motion parameter estimation and coherent accumulation detection of a high-maneuverability target according to claim 4, characterized in that: The S13 includes the following steps: S131. Change χ(τ′,t m ) takes its absolute value and performs non-coherent accumulation along the slow time. Its expression is as follows: S132. Select the coordinate value τ0′ with the largest absolute value in u(τ′), which is the autocorrelation term χ self (τ′,t m )’s peak position coordinates of the distance envelope; S133. From χ self (τ′,t m ) to extract the slow time data y(t m )=χ self (τ′=τ0,t m ), as the observation signal in the sparse reconstruction process.

6. The method for motion parameter estimation and coherent accumulation detection of a high-maneuverability target according to claim 5, characterized in that: The S2 comprises the following steps: S21. Design and construct the sparse reconstruction perception matrix Φ; S22. Based on the perception matrix Φ and the observation signal y(t m ), substitute the compressed sensing signal model y = Φx, and use the Bayesian compressed sensing algorithm to reconstruct the one-dimensional sparse signal vector S23. One-dimensional sparse signal vector Convert to a two-dimensional matrix The elements in the matrix X satisfy [X] p′,q′ =x q′+p′Q ; S24. From the two-dimensional matrix The position coordinates of L peaks are estimated and expressed as The estimated values ​​of acceleration and jerk of L targets are obtained according to the peak positions, which are: Output the motion parameter estimation results of L targets And initialize l=1.

7. The method for motion parameter estimation and coherent accumulation detection of a high-maneuverability target according to claim 6, characterized in that: The S21 includes the following steps: S211. Define a set representing the acceleration parameter range and a set representing the jerk parameter range; A collection of acceleration parameter ranges A collection of jerk parameter ranges Where P and Q represent the number of elements in the acceleration and jerk parameter sets, respectively. Δa2 and Δa3 represent the search steps for acceleration and jerk parameters, respectively. S212. Designing a sparsely reconstructed perception matrix based on the parameter search set Ξ2,Ξ3 S213. Construct the perception matrix Φ = (U T e W T ) T , where the element [Φ] m,q′+p′Q It can be expressed as Among them, define two matrices and matrix Its elements are represented as follows:

8. The method for motion parameter estimation and coherent integration detection of a high-maneuverability target according to claim 7, characterized in that: The S3 includes the following steps: S31. Construct frequency domain compensation phase based on the estimated acceleration and jerk, and calculate the radar echo s in the pulse-range frequency domain. rc (f,t m ) for phase compensation; S32. Use the Keystone algorithm to correct the remaining linear motion phase of the phase-compensated frequency-domain echo signal. Azimuth Fourier transform is then performed to obtain coherent accumulation results in the range-time domain and azimuth-frequency domain, and CFAR detection is performed.

9. The method for motion parameter estimation and coherent accumulation detection of a high-maneuverability target according to claim 8, characterized in that: The pulse-range frequency domain expression of the pulse pressure echo of the high-maneuvering target in S31 is as follows: Using the motion parameter estimate of the lth target Construct compensation function h l (f,t m ), which is expressed as follows: The signal expression after compensation is s′ rc (f,t m )=s rc (f,t m )·h l (f,t m ).

10. The method for motion parameter estimation and coherent integration detection of a high-maneuverability target according to claim 1, characterized in that: The S4 comprises the following steps: S41. If l < L, set l = l + 1 and repeat step S3 for the next target. S42. If l<L is not satisfied, directly output the detection results of all targets.

Images