A method, device, equipment and medium for identifying and classifying a UAV

Abstract

The application discloses a kind of unmanned aerial vehicle identification classification method, device, equipment and medium, its method includes: LFMCW radar is collected with the environment clutter when there is no unmanned aerial vehicle and the detection echo when there is unmanned aerial vehicle;Echo simulation is carried out to each secondary detection point of target unmanned aerial vehicle, and the corresponding theoretical echo is obtained;The secondary detection point is the detection point corresponding to non-rotor;Environment clutter and theoretical echo are combined to generate dictionary set by discretization;Based on dictionary set, OMP algorithm is used to suppress clutter to generate suppressed echo based on detection echo;The rotating speed of rotor and blade length are estimated by extracting the Doppler feature of rotor from suppressed echo;According to the rotating speed of rotor and blade length, target unmanned aerial vehicle is identified and classified;The application can solve the technical problem that target unmanned aerial vehicle cannot be identified due to the influence of clutter in complex environment or low-altitude environment.

Description

TECHNICAL FIELD

[0001] The application relates to a UAV identification and classification method, device, equipment and medium, and belongs to the technical field of UAVs. BACKGROUND

[0002] UAVs are widely used in the world today, and have made great achievements in both military and civilian fields. However, there are numerous black flight incidents in civilian field, and the jamming of UAVs in military field is particularly important. Being able to classify and identify unknown models of UAVs at a long distance can not only save spectrum resources, but also avoid electromagnetic interference in a certain space.

[0003] The key characteristic parameters for long-distance identification of UAVs often include blade rotation speed, blade length and the like. From the characteristic parameters, the means for identifying UAVs becomes more and more important. Conventional radar systems face the challenge of high resolution when distinguishing small UAVs. As a kind of small target micro-motion dynamics effect, micro-Doppler has important significance in estimating the rotation speed of UAV rotor blades and the length of blades.

[0004] Linear frequency modulated continuous wave (LFMCW) radar transmits electromagnetic waves with a specific waveform to the target, extracts and processes the frequency information of the echo signal, and then detects the motion characteristics of the target, which has broad application demand in the field of micro-motion target detection and estimation.

[0005] The most commonly used parameter estimation means is the empirical mode decomposition method, which has good detection effect in open and spacious areas, but in complex or low-altitude environments, the influence of clutter leads to the inability to identify the target. SUMMARY

[0006] The purpose of the application is to overcome the deficiencies in the prior art, and provide a UAV identification and classification method, device, equipment and medium, which uses the orthogonal matching pursuit method (OMP) to solve the technical problem that the influence of clutter in complex or low-altitude environments leads to the inability to identify the target.

[0007] To achieve the above purpose, the application adopts the following technical scheme:

[0008] In a first aspect, the application provides a UAV identification and classification method, comprising:

[0009] obtaining the environmental clutter when the UAV exists and the detection echo when the UAV exists collected by the LFMCW radar;

[0010] Echo simulations were performed on each secondary detection point of the target UAV to obtain the corresponding theoretical echoes; the secondary detection points are the detection points corresponding to the non-rotor.

[0011] The environmental clutter and theoretical echo are discretized and combined with matrices to generate a dictionary set;

[0012] Based on the dictionary set, the OMP algorithm is used to suppress clutter in the probe echo and generate a suppressed echo.

[0013] The Doppler features of the rotor were extracted from the suppressed echo, and the rotor speed and blade length were estimated.

[0014] Target drones are identified and classified based on rotor speed and blade length.

[0015] Optionally, the theoretical echo S ri (t) is:

[0016]

[0017] In the formula, S ri (t) represents the theoretical echo from the secondary detection point i, A ri Let f be the amplitude of the echo signal from the radar receiver to the non-detection point i, j be the imaginary unit, and f be the amplitude of the echo signal from the radar receiver to the non-detection point i. c Let τ be the carrier frequency of the signal, c be the speed of light, K be the frequency modulation slope of the linear modulation signal, t be the time interval, and τ be the frequency of light. i τ is the echo delay of the radar receiver to non-detection point i. i =2R i (t) / c, R i (t) represents the distance from the radar receiver to the non-detection point i at time t, and I represents the number of detection points.

[0018] Optionally, the discretization and matrix combination of environmental clutter and theoretical echo to generate a dictionary set includes:

[0019] Environmental clutter S r ′ (t) and theoretical echo S ri (t) is discretized into a vector S r ′ S ri :

[0020] S r ′ =[S r ′ (1),S r ′ (2),…S r ′ (n)…,S r ′ (N)]T

[0021] S ri =[S ri (1),S ri (2),…S r ′ i (n)…,S ri (N)] T i = 1, 2, ..., I

[0022] In the formula, S r ′ (n), S r ′ i (n) represents the values ​​of environmental clutter and the theoretical echo of the secondary detection point i at sampling point n, and N is the number of sampling points;

[0023] Vector S r ′ S ri As atoms, they are combined in a matrix to generate a dictionary set D:

[0024] D = S r1 ∪S r2 ∪,…,∪S rI ∪S r ′

[0025] In the formula, I represents the number of sub-detection points.

[0026] Optionally, the step of using the OMP algorithm based on a dictionary set to suppress clutter in the probe echo and generate suppressed echo includes:

[0027] Detecting echo S r (t) is discretized into a vector S r :

[0028] S r =[S r (1),S r (2),…S r (n)…,S r (N)] T

[0029] In the formula, S r (n) represents the value of the probe echo at sampling point n, and N is the number of sampling points;

[0030] Create an empty matrix D new Initialize residual r = S r , Sequence number k = 1, Frequency p = 1;

[0031] Repeat steps S1-S4 until p>I+1:

[0032] S1. Calculate the relationship between each atom and vector S in dictionary set D. r Contribution α q :

[0033] α q = <D q ,S r >

[0034] In the formula, D q Let α be the q-th atom in dictionary set D, where q = 1, 2, ..., I, I+1; q For atom D q With matrix S r Contribution <D q ,S r >To calculate D q S r The dot product of vectors;

[0035] S2, Obtaining Contribution α q The atom corresponding to the maximum value of is denoted as D. m m = 1, 2, ..., I, I+1; let it be an empty matrix D new Add the k-th column vector to the empty matrix D new In the middle, the atom D m Delete from dictionary set D;

[0036] S3, based on the empty matrix D new Vector S r When the residual r is minimized using the least squares method, the atom D m With vector S r Relevance λ m :

[0037]

[0038] S4. Let the sequence number k = k + 1, and the number of iterations p = p + 1;

[0039] According to vector S r Atom D m Relevance λ m Calculate the suppressed echo

[0040]

[0041] In the formula, ||·|2 is the second norm.

[0042] Optionally, the step of extracting the Doppler features of the rotor from the suppressed echo and estimating the rotor speed and blade length includes:

[0043] Time-frequency joint analysis was performed on the suppressed echo to obtain the Doppler time-frequency diagram of the rotor.

[0044] Count the number of frequency peaks in the Doppler time-frequency plot of the rotor, and estimate the rotor speed ω based on the number of frequency peaks:

[0045]

[0046] In the formula, For the number of frequency peaks, For the number of rotors, The number of blades for each rotor;

[0047] Obtain the Doppler frequency shift of the rotor's Doppler time-frequency plot, and estimate the rotor blade length L by combining this with the rotor's rotational speed.

[0048]

[0049] In the formula, f c Where c is the signal carrier frequency, f is the speed of light, and f is the signal carrier frequency. d This is the Doppler frequency shift.

[0050] Optionally, the identification and classification of the target UAV based on the rotor's rotational speed and blade length includes:

[0051] Obtain the rotational speed and blade length of the rotor of a classic drone;

[0052] The target UAV and a classic UAV are compared based on the rotor speed and blade length to obtain the comparison index J(x). a ,x b ):

[0053]

[0054] In the formula, x a ,x b For the target UAV a and the classic UAV b; x a,μ μ b,μ Let μ be the μ-th characteristic parameter of the a-th target UAV and the b-th classic UAV. When μ = 1, the characteristic parameter is the rotational speed of the rotor. When μ = 2, the characteristic parameter is the blade length of the rotor.

[0055] Take the comparison index J(x) a ,x b The smallest x a x b Determine the target drone x a And classic drones x b They belong to the same category.

[0056] Secondly, the present invention provides a drone identification and classification device, the device comprising:

[0057] The echo acquisition module is used to acquire environmental clutter when there are no UAVs and detection echoes when there are UAVs, collected by the LFMCW radar.

[0058] The echo simulation module is used to perform echo simulation on each secondary detection point of the target UAV and obtain the corresponding theoretical echo; the secondary detection points are the detection points corresponding to the non-rotor.

[0059] The dictionary module is used to discretize and combine environmental clutter and theoretical echoes into matrices to generate a dictionary set.

[0060] The echo suppression module is used to suppress clutter in the probe echo based on the dictionary set and the OMP algorithm to generate suppressed echo;

[0061] The rotor estimation module is used to extract the Doppler features of the rotor from the suppressed echo and estimate the rotor speed and blade length.

[0062] The identification and classification module is used to identify and classify target drones based on the rotor's rotation speed and blade length.

[0063] Thirdly, the present invention provides an electronic device, including a processor and a storage medium;

[0064] The storage medium is used to store instructions;

[0065] The processor is configured to operate according to the instructions to perform the steps according to the method described above.

[0066] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method.

[0067] Compared with the prior art, the beneficial effects achieved by the present invention are as follows:

[0068] This invention provides a method, apparatus, device, and medium for identifying and classifying unmanned aerial vehicles (UAVs). The method employs Orthogonal Matching Pursuit (OMP) to separate and reduce noise from environmental clutter and secondary detection point echoes in the probe echo, thereby improving the accuracy of rotor parameter estimation. Based on Doppler features, the rotational speed and blade length of the rotor are estimated, and then the target UAV is identified and classified. This solves the technical problem that the inability to identify target UAVs due to clutter in complex or low-altitude environments is a challenge. The apparatus, device, and medium can achieve the same technical effect using the above method. Attached Figure Description

[0069] Figure 1 This is a flowchart of the drone identification and classification method provided in Embodiment 1 of the present invention;

[0070] Figure 2 This is a flowchart of the OMP algorithm used to suppress clutter in the probe echo and generate the suppressed echo, provided in Embodiment 1 of the present invention.

[0071] Figure 3 This is a Doppler time-frequency diagram of a rotor provided in Embodiment 1 of the present invention. Detailed Implementation

[0072] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and should not be used to limit the scope of protection of the present invention.

[0073] Example 1:

[0074] like Figure 1 As shown, this embodiment of the invention provides a method for identifying and classifying unmanned aerial vehicles (UAVs), including the following steps:

[0075] 1. Acquire environmental clutter when no UAVs are present and detection echoes when UAVs are present, collected by LFMCW radar;

[0076] This embodiment uses the AD9361 radio frequency (RF) agile transceiver and a broadband horn antenna to transmit and receive continuous wave signals.

[0077] 2. Perform echo simulation on each secondary detection point of the target UAV to obtain the corresponding theoretical echo; the secondary detection points are the detection points corresponding to the non-rotor.

[0078] Theoretical echo S ri (t) is:

[0079]

[0080] In the formula, S ri (t) represents the theoretical echo from the secondary detection point i, A ri Let f be the amplitude of the echo signal from the radar receiver to the non-detection point i, j be the imaginary unit, and f be the amplitude of the echo signal from the radar receiver to the non-detection point i. c Let τ be the carrier frequency of the signal, c be the speed of light, K be the frequency modulation slope of the linear modulation signal, t be the time interval, and τ be the frequency of light. i τ is the echo delay of the radar receiver to non-detection point i. i =2R i (t) / c, R i (t) represents the distance from the radar receiver to the non-detection point i at time t, and I represents the number of detection points.

[0081] 3. Discretize and combine environmental clutter and theoretical echoes using matrix operations to generate a dictionary set;

[0082] Generating dictionary sets specifically includes:

[0083] Environmental clutter S r ′ (t) and theoretical echo S ri (t) is discretized into a vector S r ′ S ri :

[0084] S r ′ =[S r ′ (1),S r ′ (2),…S r ′ (n)…,S r ′ (N)] T

[0085] S ri =[S ri (1),S ri (2),…S r ′ i (n)…,S ri (N)] T i = 1, 2, ..., I

[0086] In the formula, S r ′ (n), S r ′ i (n) represents the values ​​of environmental clutter and the theoretical echo of the secondary detection point i at sampling point n, and N is the number of sampling points;

[0087] Vector S r ′ S ri As atoms, they are combined in a matrix to generate a dictionary set D:

[0088] D = S r1 ∪S r2 ∪,…,∪S rI ∪S r ′

[0089] In the formula, I represents the number of sub-detection points.

[0090] 4. Based on the dictionary set, the OMP algorithm is used to suppress clutter in the probe echo and generate suppressed echo;

[0091] like Figure 2 As shown, the generation of suppressed echo specifically includes:

[0092] Detecting echo S r (t) is discretized into a vector S r :

[0093] S r =[S r (1),S r (2),…S r (n)…,S r (N)] T

[0094] In the formula, S r (n) represents the value of the probe echo at sampling point n, and N is the number of sampling points;

[0095] Create an empty matrix D new Initialize residual r = S r , Sequence number k = 1, Frequency p = 1;

[0096] Repeat steps S1-S4 until p>I+1:

[0097] S1. Calculate the relationship between each atom and vector S in dictionary set D. r Contribution α q :

[0098] α q = <D q ,S r >

[0099] In the formula, D q Let α be the q-th atom in dictionary set D, where q = 1, 2, ..., I, I+1; q For atom D q With matrix S r Contribution <D q ,S r >To calculate D q S r The dot product of vectors;

[0100] S2, Obtaining Contribution α q The atom corresponding to the maximum value of is denoted as D. m m = 1, 2, ..., I, I+1; let it be an empty matrix D new Add the k-th column vector to the empty matrix D new In the middle, the atom D m Delete from dictionary set D;

[0101] S3, based on the empty matrix D new Vector S r When the residual r is minimized using the least squares method, the atom D m With vector Sr Relevance λ m :

[0102]

[0103] S4. Let the sequence number k = k + 1, and the number of iterations p = p + 1;

[0104] According to vector S r Atom D m Relevance λ m Calculate the suppressed echo

[0105]

[0106] In the formula, ||·|2 is the second norm.

[0107] 5. Extract the Doppler features of the rotor from the suppressed echo and estimate the rotor speed and blade length;

[0108] Estimating the rotor's rotational speed and blade length specifically includes:

[0109] Time-frequency joint analysis was performed on the suppressed echo to obtain the Doppler time-frequency diagram of the rotor.

[0110] Count the number of frequency peaks in the Doppler time-frequency plot of the rotor, and estimate the rotor speed ω based on the number of frequency peaks:

[0111]

[0112] In the formula, For the number of frequency peaks, For the number of rotors, The number of blades for each rotor;

[0113] Obtain the Doppler frequency shift of the rotor's Doppler time-frequency plot, and estimate the rotor blade length L by combining this with the rotor's rotational speed.

[0114]

[0115] In the formula, f c Where c is the signal carrier frequency, f is the speed of light, and f is the signal carrier frequency. d This is the Doppler frequency shift.

[0116] like Figure 3 As shown, a Doppler time-frequency diagram of a rotor is provided. The rotor includes two blades, and four flashes (number of frequency peaks) of blades 1 and 2 are clearly visible in the diagram, with ω = 4 rad / s.

[0117] 6. Identify and classify target UAVs based on rotor rotation speed and blade length;

[0118] This embodiment provides a recognition and classification method based on the K-means clustering algorithm, including:

[0119] Obtain the rotational speed and blade length of the rotor of a classic drone;

[0120] The target UAV and a classic UAV are compared based on the rotor speed and blade length to obtain the comparison index J(x). a ,x b ):

[0121]

[0122] In the formula, x a ,x b For the target UAV a and the classic UAV b; x a,μ μ b,μ Let μ be the μ-th characteristic parameter of the a-th target UAV and the b-th classic UAV. When μ = 1, the characteristic parameter is the rotational speed of the rotor. When μ = 2, the characteristic parameter is the blade length of the rotor.

[0123] Take the comparison index J(x) a ,x b The smallest x a x b Determine the target drone x a And classic drones x b They belong to the same category.

[0124] Example 2:

[0125] This invention provides a drone identification and classification device, the device comprising:

[0126] The echo acquisition module is used to acquire environmental clutter when there are no UAVs and detection echoes when there are UAVs, collected by the LFMCW radar.

[0127] The echo simulation module is used to simulate the echoes of each secondary detection point of the target UAV and obtain the corresponding theoretical echoes; the secondary detection points are the detection points corresponding to the non-rotor.

[0128] The dictionary module is used to discretize and combine environmental clutter and theoretical echoes into matrices to generate a dictionary set.

[0129] The echo suppression module is used to suppress clutter in the probe echo based on the dictionary set and the OMP algorithm to generate suppressed echo;

[0130] The rotor estimation module is used to extract the Doppler features of the rotor from the suppressed echo and estimate the rotor speed and blade length.

[0131] The identification and classification module is used to identify and classify target drones based on the rotor's rotation speed and blade length.

[0132] Example 3:

[0133] Based on Embodiment 1, this embodiment of the invention provides an electronic device, including a processor and a storage medium;

[0134] Storage media are used to store instructions;

[0135] The processor is used to perform operations according to instructions to execute the steps according to the method described above.

[0136] Example 4:

[0137] Based on Embodiment 1, this embodiment of the invention provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the above-described method.

[0138] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0139] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0140] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0141] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0142] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for identifying and classifying unmanned aerial vehicles, the method comprising: The method comprises the following steps: acquiring environment clutter when there is no UAV and detection echo when there is UAV collected by LFMCW radar; echo simulation is performed on each sub-detection point of the target UAV to acquire corresponding theoretical echo; the sub-detection point is a detection point corresponding to a non-rotor; discretization and matrix combination are performed on the environment clutter and the theoretical echo to generate a dictionary set; OMP algorithm is adopted based on the dictionary set to perform clutter suppression on the detection echo to generate suppressed echo; Doppler features of the rotor are extracted from the suppressed echo, and the rotating speed and the blade length of the rotor are estimated; the target UAV is identified and classified according to the rotating speed and the blade length of the rotor. The method comprises the following steps: The probe echo S r (t) is discretized into a vector S r : S r = [S r (1), S r (2),... S r (n),..., S r (N)] T In the formula, S r (n) is the value of the detected echo at sampling point n, and N is the number of sampling points. Create an empty matrix D new , initialize the residual r = S r , the sequence number k = 1, the number of times p = 1; repeating steps S1-S4 until p>I+1: S1, compute the contribution degree a of each atom in the dictionary set D to the vector S r q :​ a q = <D q , S r > where D q is the qth atom in the dictionary set D, q = 1, 2, …, I, I + 1; a q is the contribution of the atom D q to the matrix S r ; <D q , S r > is the vector inner product of D q and S r ; and I is the number of secondary probe points. S2, obtaining the contribution degree a q The atom corresponding to the maximum value of the contribution degree a is denoted as D m , m = 1, 2, …, I, I + 1; the kth column vector of the null matrix D new is added to the null matrix D new , and the atom D m is deleted from the dictionary set D. S3. According to the empty matrix D new , the vector S r The atomic D m correlation degree λ r of the vector S m : S4, let the serial number k=k+1, and the number of times p=p+1; According to the vector S r , the atom D m , the correlation λ m Calculate the suppression of the echo wherein ‖·‖2 is a two-norm. 2.The method of claim 1, wherein, The theoretical echo S ri (t) is: where S ri (t) is the theoretical echo of the secondary detection point i, A ri is the echo signal amplitude of the radar receiver to the non-detection point i, j is the imaginary unit, f c is the signal carrier frequency, c is the speed of light, K is the frequency modulation slope of the linear modulation signal, t is the time, τ i is the echo time delay of the radar receiver to the non-detection point i, τ i = 2R i (t) / c, R i (t) is the distance from the radar receiver to the non-detection point i at time t, I is the number of secondary detection points. 3.The method of claim 1, wherein, The method comprises the following steps: discretize the environmental clutter S'(t) r (t) and the theoretical echo S ri (t) into a vector S' r , S ri : S' r = [S' r (1), S' r (2), … S' r (n), …, S' r (N)] T S ri = [S ri (1), S ri (2), … S′ ri (n), …, S ri (N)] T ; i = 1, 2, …, I where S' (n) is the value of the theoretical echo of the secondary probe point i at sampling point n, and N is the number of sampling points. r (n), S' (n) ri (n) is the value of the theoretical echo of the secondary probe point i at sampling point n, and N is the number of sampling points. The vector S' is generated by the following equation: r , S ri The dictionary set D is generated by matrix combination as atoms. D = S r1 ∪S r2 ∪,…,∪S rI ∪S′ r wherein I is the number of sub-detection points. 4.The method of claim 1, wherein, The method comprises the following steps: time-frequency joint analysis is performed on the suppressed echo to acquire Doppler time-frequency diagram of the rotor; the frequency peak value number in the Doppler time-frequency diagram of the rotor is counted, and the rotating speed ω of the rotor is estimated according to the frequency peak value number: wherein is the number of frequency peaks, is the number of rotors, is the number of blades per rotor; the Doppler frequency shift of the Doppler time-frequency diagram of the rotor is acquired, and the blade length L of the rotor is estimated in combination with the rotating speed of the rotor. In the formula, f c Where c is the signal carrier frequency, f is the speed of light, and f is the signal carrier frequency. d This is the Doppler frequency shift. 5.The method of claim 1, wherein, The method comprises the following steps: the rotating speed and the blade length of the rotor of the classic UAV are acquired; Based on the rotating speed of the rotor and the length of the blade, the target UAV and the classic UAV are compared to obtain a comparison index J(x a ,x b ): In the formula, x a , x b is the a-th target unmanned aerial vehicle and the b-th classical unmanned aerial vehicle; x a,μ , μ b,μ is the μ-th characteristic parameter of the a-th target unmanned aerial vehicle and the b-th classical unmanned aerial vehicle, when μ=1, the characteristic parameter is the rotating speed of the rotor, when μ=2, the characteristic parameter is the length of the blade of the rotor; Take the comparison index J(x) a ,x b The smallest x a x b Determine the target drone x a And classic drones x b They belong to the same category.

6. An unmanned aerial vehicle identification and classification apparatus, comprising: The device comprises: an echo acquisition module, configured to acquire environment clutter when there is no UAV and detection echo when there is UAV collected by LFMCW radar; an echo simulation module, configured to perform echo simulation on each sub-detection point of the target UAV to acquire corresponding theoretical echo; the sub-detection point is a detection point corresponding to a non-rotor; a dictionary set module, configured to perform discretization and matrix combination on the environment clutter and the theoretical echo to generate a dictionary set; an echo suppression module, configured to perform clutter suppression on the detection echo based on the dictionary set by adopting OMP algorithm to generate suppressed echo, comprising: The probe echoes S r (t) are discretized into a vector S r : S r = [S r (1), S r (2),... S r (n),..., S r (N)] T wherein r(n) is the value of the detection echo at the sampling point n, and N is the number of sampling points; Create an empty matrix D new , initialize the residual r = S r , sequence number k = 1, number of times p = 1; repeating steps S1-S4 until p>I+1: S1, compute the contribution degree a of each atom in the dictionary set D to the vector S r q :​ a q = <D q , S r > where D q is the qth atom in the dictionary set D, q = 1, 2,..., I, I + 1; a q is the contribution of the atom D q to the matrix S r ; <D q , S r > is the vector inner product of D q and S r ; and I is the number of secondary probe points. S2, obtaining the contribution degree a q The atom corresponding to the maximum value of the contribution degree a is denoted as D m , m = 1, 2, …, I, I + 1; and the kth column vector of the empty matrix D new is added to the empty matrix D new , and the atom D m is deleted from the dictionary set S. S3. According to the empty matrix D new , the vector S r The atomic D m correlation degree λ r of the vector S m : S4, let the serial number k=k+1, and the number of times p=p+1; According to the vector S r , the atom D m , the correlation λ m Computing the suppression of the post-echo wherein ‖·‖2 is a two-norm. an rotor estimation module, configured to extract Doppler features of the rotor from the suppressed echo and estimate the rotating speed and the blade length of the rotor; an identification and classification module, configured to identify and classify the target UAV according to the rotating speed and the blade length of the rotor.

7. An electronic device, comprising: The device comprises a processor and a storage medium; the storage medium is configured to store instructions; the processor is configured to operate according to the instructions to perform the steps of the method according to any one of claims 1-5.

8. A computer-readable storage medium having stored thereon a computer program, characterized in that, The program is executed by the processor to implement the steps of the method according to any one of claims 1-5.

Images