Unmanned aerial vehicle group detection and resolution method and device based on KT and DeANM

Abstract

The invention discloses an unmanned aerial vehicle group detection and resolution method and device based on KT and DeANM. The method comprises the following steps: carrying out spatial-dimension digital beam forming conversion on echo signals to obtain three-dimensional echo signals; searching the acceleration and the velocity fuzzy number of the target according to the three-dimensional echo signal, and compensating the preliminarily filtered three-dimensional echo signal based on a keystone transformation algorithm; performing coarse estimation on the distance, the speed and the angle of the target according to the compensated three-dimensional echo signal; extracting a distance dimension and an angle dimension of the compensated three-dimensional echo signal according to the distance coarse estimation value, the speed coarse estimation value and the angle coarse estimation value, and performing inverse Fourier transform to obtain a two-dimensional echo signal; and performing decoupling atom norm minimization processing on the two-dimensional echo signal to obtain a Toeplitz matrix of the target, and performing Vandermonde decomposition on the Toeplitz matrix of the target to obtain a distance super-resolution estimated value and an angle super-resolution estimated value of the target. According to the invention, high-precision detection and resolution of the unmanned aerial vehicle group can be realized.

Description

Technical Field

[0001] The present invention belongs to the technical field of radar signal processing, and particularly relates to a method and device for detecting and distinguishing an unmanned aerial vehicle (UAV) swarm based on Keystone (KT) and Decoupled Atomic Norm Minimization (DeANM). Background Art

[0002] UAV swarms are characterized by high speed, strong mobility, and small size. They are applied in various scenarios, such as commercial performances, cargo transportation, agricultural protection, and military operations, etc. Therefore, UAV swarm control systems will have a broader application prospect. However, there is also a possibility that swarm UAVs may be used maliciously. In critical areas, even if only one swarm UAV breaks through the defense, the danger will be huge. Therefore, accurate detection and positioning are very important. However, compared with a single UAV which is a typical weak target with characteristics such as small radar cross-section, long distance, and time-varying motion, UAV swarms exhibit a new characteristic, namely high density. This characteristic poses a severe challenge to radar sensors. Therefore, how to accurately detect and position UAV swarms under low signal-to-noise ratio is an important part for ensuring safety, improving resource utilization efficiency, and environmental monitoring, etc.

[0003] Current methods for detecting UAV swarms either use long-time coherent accumulation technology to improve the signal-to-noise ratio and Doppler resolution of targets in echo signals, or adopt sparse methods to search for targets.

[0004] The long-time coherent accumulation technology can improve the signal-to-noise ratio and Doppler resolution of targets by increasing the coherent processing interval. The Keystone (KT) transform and the Radon-Fourier transform are typical long-time accumulation methods. The KT transform has a lower computational complexity compared with the latter. However, most of the time, the KT transform will have the problem of Doppler frequency coupling, resulting in energy diffusion.

[0005] Common sparse methods can be roughly divided into in-grid sparse methods, off-grid sparse methods, and meshless sparse methods. However, both grid-based and off-grid sparse methods will have the problem of dictionary mismatch. To solve this problem, some research has proposed to implement the process of reconstructing sparse signals in a continuous parameter space and use the total variation norm for signal recovery processing. However, for methods based on this idea, such as the atomic norm minimization method, its computational complexity is relatively high. To reduce the computational burden, a Decoupled Atomic Norm Minimization (DeANM) obtained by introducing a new atomic set into the atomic norm minimization method is proposed.

[0006] However, although the KT transformation, a long-term accumulation technique, can improve the signal-to-noise ratio (SNR) of the target, its Doppler resolution is relatively low, and it is unable to accurately locate the specific position of a single drone in a drone swarm. Although the DeANM, a sparse super-resolution method, can separate the drones in a drone swarm and obtain accurate information about individual drones, it has extremely high requirements for the SNR of the echo signal.

[0007] Therefore, the current methods for detecting and resolving drone swarms cannot balance detection accuracy and resolution accuracy. Summary of the Invention

[0008] Embodiments of the present invention provide a method and device for detecting and resolving drone swarms based on KT and DeANM, which can solve the problem that the current methods for detecting and resolving drone swarms cannot balance detection accuracy and resolution accuracy.

[0009] In a first aspect, a method for detecting and resolving drone swarms based on KT and DeANM provided by an embodiment of the present invention includes:

[0010] Performing a digital beamforming transformation in the spatial dimension on the echo signal to obtain a three-dimensional echo signal, where the three-dimensional echo signal includes information about the target in the range time domain, azimuth slow time domain, and spatial frequency domain;

[0011] Searching for the acceleration and velocity ambiguity numbers of the target according to the three-dimensional echo signal, and based on the keystone transform algorithm, compensating the preliminarily filtered three-dimensional echo signal according to the estimated acceleration value and the estimated velocity ambiguity number to obtain a compensated three-dimensional echo signal, where the preliminarily filtered three-dimensional echo signal is obtained from the three-dimensional echo signal;

[0012] Coarsely estimating the range, velocity, and angle of the target according to the compensated three-dimensional echo signal to obtain a coarse range estimate value, a coarse velocity estimate value, and a coarse angle estimate value of the target;

[0013] Extracting the range dimension and angle dimension of the compensated three-dimensional echo signal according to the coarse range estimate value, the coarse velocity estimate value, and the coarse angle estimate value, and performing an inverse Fourier transform to obtain a two-dimensional echo signal, where the two-dimensional echo signal includes information about the target in the range dimension and angle dimension;

[0014] Performing a decoupled atomic norm minimization process on the two-dimensional echo signal to obtain a Toeplitz matrix of the target, and performing a Vandermonde decomposition on the Toeplitz matrix of the target to obtain a super-resolution range estimate value and a super-resolution angle estimate value of the target.

[0015] In a second aspect, an embodiment of the present invention provides a device for detecting and resolving drone swarms based on KT and DeANM, including:

[0016] The first transformation unit is used to perform digital beamforming transformation in the spatial dimension on the echo signal to obtain a three-dimensional echo signal, where the three-dimensional echo signal includes information of the target in the range time domain, azimuth slow time domain, and spatial frequency domain;

[0017] The compensation unit is used to search for the acceleration and velocity ambiguity numbers of the target according to the three-dimensional echo signal, and based on the keystone transform algorithm, compensate the preliminarily filtered three-dimensional echo signal according to the acceleration estimation value and velocity ambiguity number estimation value to obtain a compensated three-dimensional echo signal, where the preliminarily filtered three-dimensional echo signal is obtained from the three-dimensional echo signal;

[0018] The coarse estimation unit is used to coarsely estimate the distance, velocity, and angle of the target according to the compensated three-dimensional echo signal to obtain a coarse distance estimation value, a coarse velocity estimation value, and a coarse angle estimation value of the target;

[0019] The second transformation unit is used to extract the range dimension and angle dimension of the compensated three-dimensional echo signal according to the coarse distance estimation value, coarse velocity estimation value, and coarse angle estimation value, and perform an inverse Fourier transform to obtain a two-dimensional echo signal, where the two-dimensional echo signal includes information of the target in the range dimension and angle dimension;

[0020] The fine estimation unit is used to perform decoupled atomic norm minimization processing on the two-dimensional echo signal to obtain the Toeplitz matrix of the target, and perform Vandermonde decomposition on the Toeplitz matrix of the target to obtain a super-resolution distance estimation value and a super-resolution angle estimation value of the target.

[0021] The beneficial effects of the embodiment of the present invention compared with the prior art are as follows: According to the method provided by the present invention, compensating the echo signal through the keystone transform algorithm can improve the signal-to-noise ratio of the echo signal, and performing atomic norm minimization processing can improve the resolution of the target in the echo signal, so as to achieve high-precision detection and high-resolution resolution of small targets in a group of targets, and has good robustness; at the same time, by transforming the echo signal into a three-dimensional echo signal in space-range-frequency and then performing subsequent processing, compared with the current two-dimensional super-resolution in space-range, it can further improve the resolution performance of cluster targets and improve the resolution. Description of the Drawings

[0022] Figure 1 The implementation flowchart of a method for detecting and resolving a drone swarm based on KT and DeANM provided by an embodiment of the invention;

[0023] Figure 2 A schematic diagram of the geometric position of the target and the radar;

[0024] Figure 3A waveform schematic diagram of a single-pulse single-channel of an echo signal provided by an embodiment of the present invention;

[0025] Figure 4 A schematic diagram of speed and acceleration after parameter search provided by an embodiment of the present invention;

[0026] Figure 5 A schematic diagram of a three-dimensional echo signal after keystone transform provided by an embodiment of the present invention;

[0027] Figure 6 A range-Doppler schematic diagram after constant false alarm detection provided by an embodiment of the present invention;

[0028] Figure 7 A structural schematic diagram of a UAV swarm detection and resolution device based on KT and DeANM provided by an embodiment of the present invention. Detailed implementation manners

[0029] In the following description, for the purpose of illustration rather than limitation, specific details such as specific system architectures and technologies are set forth in order to provide a thorough understanding of the embodiments of the present invention. However, those skilled in the art should understand that the present invention may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present invention with unnecessary details.

[0030] It should be understood that when used in the specification and appended claims of the present invention, the term "comprising" indicates the presence of the described features, wholes, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their combinations.

[0031] It should also be understood that the term "and / or" as used in the specification and appended claims of the present invention refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0032] As used in the specification and appended claims of the present invention, the term "if" may be construed as "when" or "once" or "in response to determining" or "in response to detecting" depending on the context. Similarly, the phrase "if determined" or "if detecting [the described condition or event]" may be construed as meaning "once determined" or "in response to determining" or "once detecting [the described condition or event]" or "in response to detecting [the described condition or event]" depending on the context.

[0033] In addition, in the description of the specification and the appended claims of the present invention, the terms "first", "second", "third", etc. are only used for distinguishing descriptions and should not be construed as indicating or implying relative importance.

[0034] The reference to "one embodiment" or "some embodiments" etc. described in the specification of the present invention means that a specific feature, structure, or characteristic described in connection with that embodiment is included in one or more embodiments of the present invention. Thus, the statements "in one embodiment", "in some embodiments", "in other some embodiments", "in still other embodiments", etc. that appear in different places in this specification do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "comprising", "including", "having" and their variants all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0035] The present invention will be further described in detail below in conjunction with specific embodiments, but the implementation manners of the present invention are not limited thereto.

[0036] The method for detecting and distinguishing an unmanned aerial vehicle (UAV) swarm based on KT and DeANM provided by an embodiment of the present invention can be applied to electronic devices such as mobile terminals, personal laptop computers, supercomputers, etc. The embodiment of the present invention does not impose any restrictions on the specific types of the electronic devices.

[0037] Figure 1 The flowchart of implementing a method for detecting and distinguishing an unmanned aerial vehicle (UAV) swarm based on KT and DeANM provided by an embodiment of the present invention is shown. By way of example and not limitation, this method can be applied to the above-mentioned electronic devices. This method may include steps S101 - S107, and each step will be described below.

[0038] S101, perform a digital beamforming transformation in the spatial dimension on the echo signal to obtain a three-dimensional echo signal.

[0039] Exemplarily, the three-dimensional echo signal may include information of a target (such as a UAV) in the range time domain, azimuth slow time domain, and spatial frequency domain.

[0040] In one possible implementation manner, the echo signal may be obtained after a target reflects the transmitted signal.

[0041] In an example, if the radar is an N-element digital array radar, the transmitted signal thereof is a pulse linear frequency modulation signal, and M pulses are transmitted within one correlation processing interval, then the transmitted signal may satisfy the following formula:

[0042]

[0043] where is the transmitted signal, is a rectangular window function, where j is the imaginary unit, B is the modulation bandwidth, is the fast time, and t m = mT r , where m is a positive integer less than or equal to M - 1, and T r is the pulse repetition period, and T P is the pulse width, and f c is the carrier frequency, is the total time.

[0044] In one example, referring to Figure 2 , if the target is in the line of sight direction of the radar, due to the low radar resolution, there may be K targets in each beam, then the echo signal (referring to the waveform shown in Figure 3 ) can satisfy the following formula:

[0045]

[0046] where, is the echo signal, l is a positive integer less than or equal to N - 1, used to represent the index of the transmitting antenna element, and A 0 is the amplitude of the echo signal, referring to Figure 2 , R k0 , v k0 , a k0 , θ k0 are respectively the distance, velocity, acceleration, and azimuth angle of the k-th target in the beam from the initial position of the radar, and k is a positive integer less than or equal to K. c is the speed of light, and n r0 is Gaussian white noise with a mean of 0 and a constant variance, and d is the element spacing.

[0047] In one possible implementation, the digital beamforming operation in the spatial dimension can be performed on the echo signal based on the steering vector of the echo signal to obtain the three-dimensional echo signal.

[0048] In one example, the steering vector of the echo signal can satisfy the following formula:

[0049]

[0050] where, a(θ) is the steering vector of the echo signal in the θ direction, and θ = zδ θ (z = 1,..., Z), z is the index of the grid number, and δ θ is the angular resolution, Z is the total number of grids divided, and λ represents the wavelength.

[0051] In one example, the three-dimensional echo signal can satisfy the following formula:

[0052]

[0053] Among them, is a three-dimensional echo signal, and A DBF is the amplitude of the three-dimensional echo signal. Similarly, it represents Gaussian white noise with a mean of 0 and a constant variance, and sinc(x) = sin(πx) / πx.

[0054] S102. Search for the acceleration and velocity ambiguity numbers of the target based on the three-dimensional echo signal to obtain the estimated value of the target's acceleration and the estimated value of the velocity ambiguity number.

[0055] In a possible implementation, the first operation can be performed on the three-dimensional echo signal twice to obtain a preliminarily filtered three-dimensional echo signal, and then the preliminarily filtered three-dimensional echo signal is searched to obtain the estimated value of the target's acceleration and the estimated value of the velocity ambiguity number.

[0056] For example, referring to Figure 4 the two peaks searched in, the values of their X, Y, and Z coordinates are the estimated value of the acceleration, the velocity ambiguity number, and the amplitude of the signal respectively.

[0057] Exemplarily, the first operation may include range-Doppler compression processing and range fast Fourier transform in the execution order.

[0058] In an example, the three-dimensional echo signal Y DBF (f, t m , θ) obtained after the first execution of the first operation can satisfy the following formula:

[0059]

[0060] Among them, f represents the range frequency, and f r is the pulse repetition frequency, and n amb is the velocity ambiguity number of the target.

[0061] Similarly, the preliminarily filtered three-dimensional echo signal can satisfy the following formula:

[0062]

[0063] Among them, A PC is the amplitude of the preliminarily filtered three-dimensional echo signal, and n PC (f, t m , z) is Gaussian noise.

[0064] S103. Based on the keystone transform algorithm, compensate the preliminarily filtered three-dimensional echo signal according to the estimated value of the acceleration and the estimated value of the velocity ambiguity number to obtain the compensated three-dimensional echo signal.

[0065] In one example, a two-dimensional matched filtering function can be constructed in the range time domain and slow time dimension based on the estimated acceleration value of the target and the estimated velocity ambiguity number, and then the wedge transform ((f + f c )t m = f c t′ m ) is performed on the preliminarily filtered three-dimensional signal in the range frequency domain. By multiplying this two-dimensional matched filtering function with the three-dimensional echo signal after the wedge transform (see Figure 5 ), the three-dimensional echo signal is compensated to obtain the compensated three-dimensional echo signal.

[0066] Exemplarily, the two-dimensional matched filtering function can satisfy the following formula:

[0067]

[0068] where is the two-dimensional matched filtering function, are respectively the estimated acceleration value of the target and the estimated velocity ambiguity number, and t′ m is the slow time after scale transformation.

[0069] Exemplarily, the compensated three-dimensional echo signal can satisfy the following formula:

[0070]

[0071] where Y DK (f, t′ m , θ) is the compensated three-dimensional echo signal, A DK is the amplitude of the compensated three-dimensional echo signal, and n DK (f, t m , z) also represents Gaussian noise.

[0072] By performing operations such as wedge transform, velocity ambiguity number and azimuth quadratic term estimation and compensation on the echo signal in the range frequency domain, the range migration and Doppler frequency migration of the target can be compensated.

[0073] S104. Coarse estimate the distance, velocity and angle of the target based on the compensated three-dimensional echo signal to obtain the coarse estimated distance value, coarse estimated velocity value and coarse estimated angle value of the target.

[0074] In one possible implementation, the inverse Fourier transform and Fourier transform can be respectively performed on the compensated three-dimensional echo signal in the range frequency domain and slow time domain to obtain the transformed three-dimensional echo signal; then constant false alarm detection is performed on the transformed three-dimensional echo signal in the range dimension, velocity dimension and angle dimension to obtain the coarse estimated distance value, coarse estimated velocity value and coarse estimated angle value of the target.

[0075] For example, refer to Figure 6 , which shows the waveform obtained by performing constant false alarm rate detection. The X, Y, and Z coordinates of this waveform respectively correspond to the rough estimated distance value, rough estimated speed value, and amplitude of the target.

[0076] In one example, the transformed three-dimensional echo signal can satisfy the following formula:

[0077]

[0078] where is the transformed three-dimensional echo signal, and A F is the amplitude of the transformed three-dimensional echo signal. also represents Gaussian noise, and f m and T M are respectively the Doppler dimension frequency variable and the coherent integration time.

[0079] S105. Extract the range dimension and angle dimension of the compensated three-dimensional echo signal according to the rough estimated distance value, rough estimated speed value, and rough estimated angle value, and perform an inverse Fourier transform to obtain a two-dimensional echo signal in the range dimension and angle dimension.

[0080] In one example, the two-dimensional echo signal can satisfy the following formula:

[0081]

[0082] where S n-l (n, l) is the two-dimensional echo signal, and A n-l , e n-l respectively represent the amplitude and noise in the range, speed, and angle domains. is the range frequency domain sampling index, N r is the number of range frequency domain sampling points, l is the spatial domain sampling index, N is the number of spatial domain sampling points, and F s is the sampling frequency.

[0083] S106. Perform decoupled atomic norm minimization processing on the two-dimensional echo signal to obtain the Toeplitz matrix of the target.

[0084] In one possible implementation, the two-dimensional echo signal can be rewritten based on the normalized frequencies of the target along the range dimension and angle dimension to obtain the rewritten two-dimensional echo signal. Then, based on the atomic norm minimization model, perform decoupled atomic norm minimization processing on the two-dimensional echo signal to obtain the Toeplitz matrix of the target.

[0085] In one example, the rewritten two-dimensional echo signal can satisfy the following formula:

[0086]

[0087] Among them, S is the rewritten two-dimensional echo signal, and f Rk and f θk are the normalized frequencies in the angular dimension and the normalized frequency in the range dimension respectively. S, y, and e n are all one-dimensional vectors, e n is the noise vector, and G k is the received signal amplitude.

[0088] In one example, the decoupled atomic norm minimization of the target can be transformed into a semi-definite programming (SDP) problem through the atomic norm minimization model.

[0089] Exemplarily, the atomic norm minimization model can satisfy the following formula:

[0090]

[0091] Among them, T(u R ) and T(u θ ) are the optimized Toeplitz matrices corresponding to the distance and the angle respectively, where the first row sequences are u R and u θ . Tr() represents the trace of the matrix, and λ 1 is the regularization coefficient, which is used to balance the signal sparsity and data fidelity.

[0092] S107. Perform Vandermonde decomposition on the Toeplitz matrix of the target to obtain the range super-resolution estimation value and the angle super-resolution estimation value of the target.

[0093] In one example, the matrix pencil or subspace class method can be used to perform Vandermonde decomposition on the Toeplitz matrix of the target.

[0094] Exemplarily, the range super-resolution estimation value and the angle super-resolution estimation value of the target can satisfy the following formula:

[0095]

[0096] Among them, are the range super-resolution estimation value and the angle super-resolution estimation value of the target respectively, are the estimated values of the normalized frequencies in the angular dimension and the range dimension respectively.

[0097] According to the method provided by the present invention, compensating the echo signal through the keystone transform algorithm can improve the signal-to-noise ratio of the echo signal, and the resolution of the target in the echo signal can be improved through atomic norm minimization processing, so that high-precision detection and high-resolution resolution of small targets in a group of targets can be achieved, and it has good robustness; at the same time, by transforming the echo signal into a three-dimensional echo signal in space-range-frequency and then performing subsequent processing, compared with the current two-dimensional super-resolution only in space-range, the resolution performance of cluster targets can be further improved, and the resolution can be increased.

[0098] Figure 7 The following shows a schematic structural diagram of a drone swarm detection and resolution device based on KT and DeANM provided by an embodiment of the present invention. By way of example and not limitation, the device may include a first transformation unit 710, a compensation unit 720, a rough estimation unit 730, a second transformation unit 740, and a fine estimation unit 750.

[0099] Exemplarily, the first transformation unit is used to perform a digital beamforming transformation in the spatial dimension on the echo signal to obtain a three-dimensional echo signal, where the three-dimensional echo signal includes information of the target in the range time domain, azimuth slow time domain, and spatial frequency domain; the compensation unit is used to search for the acceleration and velocity ambiguity numbers of the target according to the three-dimensional echo signal, and based on the keystone transform algorithm, compensate the preliminarily filtered three-dimensional echo signal according to the acceleration estimation value and the velocity ambiguity number estimation value to obtain a compensated three-dimensional echo signal, where the preliminarily filtered three-dimensional echo signal is obtained according to the three-dimensional echo signal; the rough estimation unit is used to roughly estimate the distance, velocity, and angle of the target according to the compensated three-dimensional echo signal to obtain a rough distance estimation value, a rough velocity estimation value, and a rough angle estimation value of the target; the second transformation unit is used to perform an inverse Fourier transform on the distance dimension and the angle dimension of the compensated three-dimensional echo signal according to the rough distance estimation value, the rough velocity estimation value, and the rough angle estimation value to obtain a two-dimensional echo signal, where the two-dimensional echo signal includes information of the target in the distance dimension and the angle dimension; the fine estimation unit is used to perform decoupled atomic norm minimization processing on the two-dimensional echo signal to obtain the Toeplitz matrix of the target, and perform Vandermonde decomposition on the Toeplitz matrix of the target to obtain a distance super-resolution estimation value and an angle super-resolution estimation value of the target.

[0100] In one example, the three-dimensional echo signal may satisfy the following formula:

[0101]

[0102] where is the three-dimensional echo signal, and A DBF is the amplitude of the three-dimensional echo signal, denotes Gaussian white noise with a mean of 0 and a constant variance, sinc(x) = sin(πx) / πx, and K is the total number of targets present within a radar beam. is the fast time, t m = mT r , where m is a positive integer less than or equal to M - 1, and T r is the pulse repetition period, T P is the pulse width, f c is the carrier frequency, and c is the speed of light. R k0 , ν k0 , a k0 , θ k0 are respectively the distance, velocity, acceleration, and azimuth angle of the k-th target within a radar beam from the radar's initial position.

[0103] In one example, the two-dimensional echo signal can satisfy the following formula:

[0104]

[0105] where S n-l (n, l) is the two-dimensional echo signal, and A n-l , e n-l respectively represent the amplitudes and noises in the distance, velocity, and angle domains. is the distance frequency domain sampling index, N r is the number of distance frequency domain sampling points, l is the spatial domain sampling index, N is the number of spatial domain sampling points, and F s is the sampling frequency.

[0106] In one example, the fine estimation unit 750 can specifically be used for:

[0107] Performing decoupled atomic norm minimization processing on the two-dimensional echo signal based on the atomic norm minimization model to obtain the Toeplitz matrix of the target;

[0108] The atomic norm minimization model satisfies the following formula:

[0109]

[0110] where T(u R ), T(u θ ) are respectively the optimal Toeplitz matrices corresponding to the distance and angle with the first row sequence being u R and u θ , Tr() represents the trace of the matrix, λ 1 is the regularization coefficient, S is the rewritten two-dimensional echo signal, and y is the noise-free signal.

[0111] In one example, the distance super-resolution estimate and the angle super-resolution estimate of the target can satisfy the following formula:

[0112]

[0113] Wherein, are the distance super-resolution estimate and the angle super-resolution estimate of the target respectively, are the estimates of the normalized frequencies of the angle-distance dimension of the target respectively.

[0114] According to the device provided by the present invention, compensating the echo signal through the keystone transform algorithm can improve the signal-to-noise ratio of the echo signal, and minimizing the atomic norm can improve the resolution of the target in the echo signal, so as to achieve high-precision detection and high-resolution resolution of small targets in a group of targets, and has good robustness; at the same time, by transforming the echo signal into a three-dimensional echo signal in space-distance-frequency and then performing subsequent processing, compared with the current two-dimensional super-resolution in space-distance, it can further improve the resolution performance of cluster targets and improve the resolution.

[0115] In the above embodiments, the descriptions of the respective embodiments have their own emphases. For parts not detailed or recorded in a certain embodiment, reference may be made to the relevant descriptions of other embodiments.

Claims

1. A drone swarm detection and identification method based on KT and DeANM, characterized in that: include: Performing a spatial-dimensional digital beamforming transformation on the echo signal to obtain a three-dimensional echo signal, wherein the three-dimensional echo signal includes information of the target in the range time domain, the azimuth slow time domain, and the spatial frequency domain; Searching for the acceleration and velocity fuzzy number of the target according to the three-dimensional echo signal, and compensating the three-dimensional echo signal after preliminary filtering according to the acceleration estimation value and the velocity fuzzy number estimation value based on the keystone transformation algorithm to obtain a compensated three-dimensional echo signal, wherein the three-dimensional echo signal after preliminary filtering is obtained according to the three-dimensional echo signal; The distance, speed and angle of the target are roughly estimated according to the compensated three-dimensional echo signal to obtain a rough distance estimation value, a rough speed estimation value and a rough angle estimation value of the target; Extracting the distance dimension and the angle dimension of the compensated three-dimensional echo signal according to the rough distance estimation value, the rough speed estimation value and the rough angle estimation value, and performing inverse Fourier transform to obtain a two-dimensional echo signal, wherein the two-dimensional echo signal includes information of the target in the distance dimension and the angle dimension; The two-dimensional echo signal is subjected to decoupled atomic norm minimization processing to obtain the Toeplitz matrix of the target, and the Toeplitz matrix of the target is subjected to Vandermonde decomposition to obtain a distance super-resolution estimation value and an angle super-resolution estimation value of the target.

2. The method according to claim 1, characterized in that The three-dimensional echo signal satisfies the following formula: in, is the three-dimensional echo signal, A DBF is the amplitude of the three-dimensional echo signal, represents Gaussian white noise with a mean of 0 and a constant variance, sin c(x) = sin(πx) / πx, K is the total number of targets in one radar beam, For fast time, t m =mT r , m is a positive integer less than or equal to M-1, T r is the pulse repetition period, T P is the pulse width, f c is the carrier frequency, c is the speed of light, R k0 、v k0 、a k0 ,θ k0 are respectively the initial distance, velocity, acceleration and azimuth between the kth target in a beam of the radar and the radar, k is a positive integer less than or equal to K, and z is the index of the divided grid number.

3. The method according to claim 1, characterized in that The two-dimensional echo signal satisfies the following formula: Among them, S n-l (n, l) is the two-dimensional echo signal, A n-l 、e n-l represent the amplitude and noise in the distance, velocity and angle domains respectively, is the distance frequency domain sampling index, N r is the number of sampling points in the distance frequency domain, l is the spatial domain sampling index, F s is the sampling frequency.

4. The method according to claim 1, characterized in that: The step of performing decoupled atomic norm minimization processing on the two-dimensional echo signal to obtain the Toeplitz matrix of the target includes: Performing decoupled atomic norm minimization processing on the two-dimensional echo signal based on an atomic norm minimization model to obtain a Toeplitz matrix of the target; The atomic norm minimization model satisfies the following formula: Among them, T(u R )、T(u θ ) are the distance and angle respectively. The first row of the corresponding sequence is u R and u θ The optimal Toeplitz matrix, Tr() represents the trace of the matrix, λ1 is the regularization coefficient, S is the rewritten two-dimensional echo signal, and y is the noise-free signal.

5. The method according to claim 1, characterized in that The distance super-resolution estimation value and the angle super-resolution estimation value of the target satisfy the following formula: in, are respectively the distance super-resolution estimation value and the angle super-resolution estimation value of the target, are the estimates of the normalized frequencies of the angle dimension and the distance dimension of the target, respectively.

6. A drone swarm detection and identification device based on KT and DeANM, characterized in that: include: A first transformation unit, the first transformation unit is used to perform a spatial-dimensional digital beamforming transformation on the echo signal to obtain a three-dimensional echo signal, wherein the three-dimensional echo signal includes information of the target in a range time domain, an azimuth slow time domain, and a spatial frequency domain; A compensation unit, the compensation unit is used to search for the acceleration and velocity ambiguity number of the target according to the three-dimensional echo signal, and based on the keystone transformation algorithm, compensate the three-dimensional echo signal after preliminary filtering according to the acceleration estimation value and the velocity ambiguity number estimation value to obtain a compensated three-dimensional echo signal, wherein the three-dimensional echo signal after preliminary filtering is obtained according to the three-dimensional echo signal; A rough estimation unit, the rough estimation unit is used to roughly estimate the distance, speed and angle of the target according to the compensated three-dimensional echo signal to obtain a rough distance estimation value, a rough speed estimation value and a rough angle estimation value of the target; a second transformation unit, the second transformation unit being used to extract the distance dimension and the angle dimension of the compensated three-dimensional echo signal according to the rough distance estimation value, the rough speed estimation value and the rough angle estimation value, and perform inverse Fourier transformation to obtain a two-dimensional echo signal, wherein the two-dimensional echo signal includes information of the target in the distance dimension and the angle dimension; A precision estimation unit, the precision estimation unit is used to perform decoupled atomic norm minimization processing on the two-dimensional echo signal to obtain the Toeplitz matrix of the target, and perform Vandermonde decomposition on the Toeplitz matrix of the target to obtain a distance super-resolution estimation value and an angle super-resolution estimation value of the target.

7. The device according to claim 6, characterized in that The three-dimensional echo signal satisfies the following formula: in, is the three-dimensional echo signal, A DBF is the amplitude of the three-dimensional echo signal, represents Gaussian white noise with a mean of 0 and a constant variance, sin c(x) = sin(πx) / πx, K is the total number of targets in one radar beam, For fast time, t m =mT r , m is a positive integer less than or equal to M-1, T r is the pulse repetition period, T P is the pulse width, f c is the carrier frequency, c is the speed of light, R k0 、v k0 、a k0 ,θ k0 are respectively the initial distance, velocity, acceleration and azimuth between the kth target in a beam of the radar and the radar, k is a positive integer less than or equal to K, and z is the index of the divided grid number.

8. The device according to claim 6, characterized in that The two-dimensional echo signal satisfies the following formula: Among them, S n-l (n, l) is the two-dimensional echo signal, A n-l 、e n-l represent the amplitude and noise in the distance, velocity and angle domains respectively, is the distance frequency domain sampling index, N r is the number of sampling points in the distance frequency domain, l is the spatial domain sampling index, F s is the sampling frequency.

9. The device according to claim 6, characterized in that The precise estimation unit is specifically used for: Performing decoupled atomic norm minimization processing on the two-dimensional echo signal based on an atomic norm minimization model to obtain a Toeplitz matrix of the target; The atomic norm minimization model satisfies the following formula: Among them, T(u R )、T(u θ ) are the distance and angle respectively. The first row of the corresponding sequence is u R and u θ The optimal Toeplitz matrix, Tr() represents the trace of the matrix, λ1 is the regularization coefficient, S is the rewritten two-dimensional echo signal, and y is the noise-free signal.

10. The device according to claim 6, characterized in that The distance super-resolution estimation value and the angle super-resolution estimation value of the target satisfy the following formula: in, are respectively the distance super-resolution estimation value and the angle super-resolution estimation value of the target, are the estimates of the normalized frequencies of the angle dimension and the distance dimension of the target, respectively.

Images