An UAV target inversion method, system, device and medium under strong sea clutter background based on joint filtering in time-frequency-Doppler domain and time-frequency ridge line analysis

Abstract

An unmanned aerial vehicle (UAV) target inversion method, system, device and medium based on joint filtering in time-frequency-Doppler domain and time-frequency ridge analysis under strong sea clutter background. For a linear frequency modulation signal system radar, a radar echo model of a multi-rotor UAV is established, and the radar echo of the multi-rotor UAV is calculated. The measured sea surface radar echo is added to the radar echo of the multi-rotor UAV to form a complete UAV radar echo under strong sea clutter background, that is, the total radar echo. The total radar echo is subjected to joint filtering in time-frequency-Doppler domain to obtain the filtered total power spectral density Pl. The time-frequency ridge extraction is performed on the filtered total power spectral density Pl to verify the fitting degree. The system, device and medium are used to carry and implement the method. Compared with the traditional single-domain UAV detection, the present invention can better suppress sea clutter, has a wider application range, and can accurately detect and identify UAV targets in a high sea state environment.

Description

Technical Field

[0001] The present invention belongs to the technical field of ocean data assimilation, and particularly relates to a method, system, device and medium for unmanned aerial vehicle (UAV) target inversion under strong sea clutter background based on joint filtering in time-frequency-Doppler domain and time-frequency ridge line analysis. Background Art

[0002] As one of the typical representatives of low, slow and small targets, the detection and identification of UAVs are seriously affected by various environmental clutters and noises. Moreover, the characteristics of low flight altitude, slow flight speed and small radar cross section (RCS) of UAVs also lead to the fact that UAVs are easily blocked by other objects or submerged by clutters, making it difficult to be detected by radar. Therefore, establishing an accurate and realistic UAV radar echo modulation model in the field of electromagnetic scattering is the premise and foundation of the research. For sea surface target detection, sea clutter is one of the main factors restricting radar performance. Its complex waveform results in the lack of a unified model that can accurately describe sea clutter, and its non-stationarity and non-Gaussianity make suppressing sea clutter a difficult problem.

[0003] When using radar to detect low, slow and small targets on the sea surface, due to its slow speed, the Doppler frequency of the target is easily submerged by sea clutter, and the overall energy of the sea clutter background is relatively strong, resulting in a low signal-to-clutter ratio / signal-to-noise ratio during detection, unable to effectively detect the target, seriously reducing the radar performance and interfering with the normal operation of the radar. And when suppressing sea clutter only in a single domain such as the time domain or the frequency domain, the effect is often not good, and it is unable to effectively suppress and detect the echo of the submerged low, slow and small target.

[0004] The literature "Research on UAV Recognition Algorithm Based on Micro-Doppler" proposed a joint time-frequency algorithm based on short-time Fourier transform and pseudo-Wigner-Ville distribution (Guo Jia, North China University of Science and Technology, Tangshan 063000). By jointly calculating the short-time Fourier transform with poor time-frequency aggregation and no cross terms and the pseudo-Wigner-Ville distribution with good time-frequency aggregation and fewer cross terms, the cross terms are suppressed, and at the same time, a high time-frequency resolution is maintained, so as to effectively detect UAVs. However, this method only detects UAV targets in a single domain, is greatly limited, and has a narrow application range. Summary of the Invention

[0005] To overcome the deficiencies of the above-mentioned existing technologies, the purpose of the present invention is to provide an unmanned aerial vehicle (UAV) target inversion method, system, device, and medium based on joint filtering in the time-frequency-Doppler domain and time-frequency ridge analysis under a strong sea clutter background. By processing radar target echo data in the time domain, frequency domain, and Doppler domain, the echo energy of the target is enhanced. Based on the ratio of the clutter to the mean value of the target echo power spectral density, a filtering factor calculation model is established to perform time-frequency domain filtering. Combining with the time-frequency ridge extraction method, the inversion results of the UAV's motion speed and distance under a strong sea clutter background are obtained. Compared with traditional single-data-domain (time domain / frequency domain) UAV detection, the present invention can better suppress sea clutter, has a wider application range, and realizes UAV target detection and recognition under high sea state conditions.

[0006] To achieve the above purpose, the technical solution adopted by the present invention is as follows:

[0007] An unmanned aerial vehicle (UAV) target inversion method based on joint filtering in the time-frequency-Doppler domain and time-frequency ridge analysis under a strong sea clutter background, comprising the following steps:

[0008] Step 1: For a linear frequency modulation (LFM) signal system radar, establish a radar echo model of a multi-rotor UAV, calculate the radar echo of the multi-rotor UAV, and add the measured sea surface radar echo to the radar echo of the multi-rotor UAV to form a complete UAV radar echo under a strong sea clutter background, that is, the total radar echo;

[0009] Step 2: Perform joint filtering in the time-frequency-Doppler domain on the total radar echo obtained in Step 1 to obtain the filtered total power spectral density P l ;

[0010] Step 3: Extract the time-frequency ridge from the filtered total power spectral density P obtained in Step 2 l to verify the fitting degree.

[0011] The specific method of Step 1 is as follows:

[0012] Step 1.1: The time-domain expression of the LFM signal as the transmitted signal is as follows:

[0013]

[0014] where, A0 is the initial amplitude of the transmitted signal, f0 is the carrier frequency, B is the frequency modulation bandwidth, T c is the frequency modulation period, k = B / T c is the frequency modulation slope, and φ0 is the initial phase;

[0015] The time-domain expression of the target echo signal is:

[0016]

[0017] where, τ is the time delay;

[0018] Step 1.2: Subtract the phase of the transmitted signal from the phase of the target echo signal in Step 1.1, and the obtained baseband signal phase is:

[0019] P b (t) = 2π(f0τ(t) + ktτ(t) - kτ 2 (t) / 2)

[0020] Step 1.3: Further obtain the beat signal from the baseband signal phase in Step 1.2 as:

[0021]

[0022] Step 1.4: Express the distance from the scatter point Q on a certain blade of a certain rotor of the UAV to the radar as:

[0023] R Q = R0 + vt + l Q cosβcos(θ0 + Ω m t - α)

[0024] where, R0 is the initial slant range of the target, v is the UAV speed, l Q is the distance from the scatter point Q to the blade rotation center, β is the pitch angle, θ0 is the initial blade rotation angle, Ω m is the angular velocity of the m-th rotor, and α is the azimuth angle;

[0025] Step 1.5: Substitute the distance R from the scatter point Q on a certain blade of a certain rotor of the UAV obtained in Step 1.4 Q into the echo time delay τ = 2R Q / c, where c is the speed of light, and combine with the beat signal obtained in Step 1.3, and integrate over the entire blade length L to obtain the radar echo expression of the multi-rotor UAV:

[0026]

[0027] where, M is the number of rotors, N is the number of blades on one rotor, σ mirco is the scattering coefficient of the blade, and σ main is the scattering coefficient of the UAV body.

[0028] The specific method of Step 2 is as follows:

[0029] Step 2.1: Perform clutter suppression on the total radar echo in the slow-time dimension of the time domain, filter out the stationary clutter, and retain the moving targets to achieve the indication of moving targets; the impulse response of the corresponding filter for moving target indication is:

[0030]

[0031] Among them, J is the number of pulse cancellations; the output signal is:

[0032] y1(t) = r s (t) * h J (q)

[0033] Among them, r s (t) is the total radar echo;

[0034] Step 2.2: Set the impulse response function of the matched filter:

[0035] h(t) = k0s * (t0 - t)

[0036] Among them, k0 is a non-zero constant, and s(t) is the target signal; perform matched filtering on the output signal y1(t) obtained in Step 2.1, and the output signal is:

[0037]

[0038] Step 2.3: Perform short-time Fourier transform on the output signal y2(t) obtained in Step 2.2:

[0039] STFT(T, F) = ∫y2(r)w(r - T)e -j2πFr dr

[0040] Among them, w(·) is the window function, and at this time the power spectral density of the transformed signal is:

[0041] P(T, F) = k(T)|STFT(T, F)| 2

[0042]

[0043] Among them, k(T) is the transformation coefficient to compensate for the influence of the window function, and f s is the sampling rate during short-time Fourier transform;

[0044] Step 2.4: Find the extreme value of the power spectral density calculated in Step 2.3, and select the mean value of the power spectral density at non-extreme frequencies in different time dimensions as the mean value of the clutter power spectral density, that is:

[0045] P1 = P(T1, F1) / L T

[0046] Among them, L T is the time length, and select the mean value of the power spectral density at the maximum frequency in different time dimensions as the mean value of the target power spectral density, that is:

[0047] P2 = max(P(T,F)) / L T

[0048] Step 2.5: Take the ratio of the mean power spectral density of the clutter obtained in Step 2.4 to the mean power spectral density of the target as the time-frequency filtering factor γ, and establish a filtering factor calculation model, that is:

[0049] γ = P1 / P2

[0050] Perform time-frequency filtering on the total power spectral density, and the specific expression is:

[0051] P l = {P[P(T,F) < γ * max(P(T,F))] = 0}

[0052] Up to this point, the joint filtering in the time-frequency-Doppler domain of the total radar echo is achieved.

[0053] The specific method of Step 3 is as follows:

[0054] [[ID=!]]Take the ratio of the mean power spectral density of the clutter obtained in Step 2.4 to the mean power spectral density of the target as the time-frequency filtering factor, perform ridge extraction on the time-frequency diagram, compare the ridge of the UAV time-frequency diagram with the ridge of the UAV time-frequency diagram under the sea clutter background, and calculate its mean fitting diagram.

[0055] The present invention also provides an unmanned aerial vehicle (UAV) target inversion system under strong sea clutter background based on joint filtering in the time-frequency-Doppler domain and time-frequency ridge analysis, including:

[0056] A UAV radar echo calculation module, which is used to establish a radar echo model of a multi-rotor UAV for a linear frequency modulation signal system radar and calculate the radar echo of the multi-rotor UAV;

[0057] A total radar echo filtering module, which is used to combine the UAV radar echo with the measured sea surface radar echo to form a complete UAV radar echo under strong sea clutter background, that is, the total radar echo, and perform joint filtering on the total radar echo in the time-frequency-Doppler domain to obtain the filtered total power spectral density P l ;

[0058] A time-frequency ridge extraction module, which is used to perform time-frequency ridge extraction on the filtered total power spectral density P l to verify the fitting degree.

[0059] The present invention also provides an unmanned aerial vehicle (UAV) target inversion device under strong sea clutter background based on joint filtering in the time-frequency-Doppler domain and time-frequency ridge analysis, including:

[0060] **Note**: There seems to be an issue with the text in ID=25 where the Chinese and English translations don't match in terms of content. The English translation above is based on the correct understanding of the Chinese text in ID=25. If this is an error in the original text, it should be corrected for a more accurate translation.Memory: It stores a computer program for the above-mentioned method for inverting UAV targets under strong sea clutter background based on joint filtering in time-frequency-Doppler domain and time-frequency ridge line analysis, and is a computer-readable device;

[0061] Processor: When executing the computer program, it is used to implement the method for inverting UAV targets under strong sea clutter background based on joint filtering in time-frequency-Doppler domain and time-frequency ridge line analysis.

[0062] The present invention also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, it can implement the method for inverting UAV targets under strong sea clutter background based on joint filtering in time-frequency-Doppler domain and time-frequency ridge line analysis.

[0063] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0064] 1. The present invention proposes a multi-dimensional method for inverting the motion state of UAV targets under strong sea clutter background. By performing joint filtering on signals in time-frequency-Doppler domain, it improves the limitations of traditional filtering methods in detecting UAVs under strong sea clutter conditions, enhances the suppression effect on sea clutter, and improves the distinguishability between UAV targets and sea clutter in multi-dimensional fields.

[0065] 2. The present invention uses the ratio of the mean power spectral density of clutter to the mean power spectral density of the target as a time-frequency filtering factor to perform time-frequency filtering and ridge line extraction, enhancing the filtering effect on sea clutter. The mean fitting of the time-frequency ridge line envelope verifies the accuracy of this method for inverting the motion state of UAVs.

[0066] In summary, the present invention effectively suppresses sea clutter by performing multi-dimensional collaborative filtering on radar target echo data in time domain-frequency domain-Doppler domain, significantly enhances the echo energy of the target, and establishes a filtering factor calculation model through the ratio of the mean power spectral density of clutter to the target echo. It conducts time-frequency domain filtering and combines the method of ridge line extraction to verify that this method is relatively accurate in inverting the motion speed and distance of UAVs under strong sea clutter background. Compared with traditional single data domain (time domain / frequency domain) UAV detection, this method can better suppress sea clutter, has a wider application range, and can accurately detect and identify UAV targets, providing a good simulation platform for UAV detection research in strong clutter environments. Brief Description of the Drawings

[0067] Figure 1 It is a flow chart of the implementation method of the present invention.

[0068] Figure 2It is the electromagnetic modeling simulation diagram of the UAV of the present invention; among them, (a) is the time-domain echo diagram of the UAV radar, (b) is the time-domain echo diagram of the filtered UAV radar, (c) is the time-frequency diagram of the UAV, (d) is the time-frequency diagram of the filtered UAV, (e) is the Doppler spectrum of the UAV, and (f) is the Doppler spectrum of the filtered UAV.

[0069] Figure 3 It is the combined diagram of the simulated UAV echo and the measured sea surface data of the present invention; among them, (a) is the pulse-distance diagram of the UAV under the sea clutter background, (b) is the pulse-distance diagram of the filtered UAV under the sea clutter background, (c) is the time-frequency diagram of the UAV under the sea clutter background, (d) is the time-frequency diagram of the filtered UAV under the sea clutter background, (e) is the Doppler spectrum of the UAV under the sea clutter background, and (f) is the Doppler spectrum of the filtered UAV under the sea clutter background.

[0070] Figure 4 It is the comparison diagram of the time-frequency ridge envelope of the UAV of the present invention and the time-frequency ridge envelope of the UAV under the sea clutter background.

[0071] Figure 5 It is the comparison diagram of the mean fitting of the time-frequency ridge envelope of the UAV of the present invention and the mean fitting of the time-frequency ridge envelope of the UAV under the sea clutter background; among them, (a) is the upper envelope comparison diagram and (b) is the lower envelope comparison diagram. Specific embodiments

[0072] The technical solutions adopted by the present invention will be further described below in conjunction with the accompanying drawings and specific embodiments.

[0073] Based on the radar target echo theory under the linear frequency modulation signal system and the characteristics of multi-rotor UAVs, the present invention conducts electromagnetic modeling of the UAV radar echo; combines the simulated radar echo of the UAV with the measured sea surface data and performs joint filtering in the time-frequency-Doppler domain; uses the ratio of the mean power spectral density of clutter to the mean power spectral density of the target as the time-frequency filtering factor to extract the time-frequency ridge. Compare the time-frequency ridge of the UAV time-frequency diagram with the time-frequency ridge of the UAV under the sea clutter background, calculate its mean fitting, and invert and verify the moving speed and distance of the UAV target. Compared with the traditional single-domain UAV detection, the present invention can better suppress sea clutter, has a wider application range, and can accurately detect and identify UAV targets in a high sea state environment.

[0074] As Figure 1 shown, a method for inverting UAV targets under strong sea clutter background based on joint filtering in the time-frequency-Doppler domain and time-frequency ridge analysis includes the following steps:

[0075] Step 1: Based on the electromagnetic scattering theory, for a linear frequency modulation (LFM) signal system radar, establish a radar echo model of a multi-rotor unmanned aerial vehicle (UAV), calculate the radar echo of the multi-rotor UAV, and add the measured sea surface radar echo to the radar echo of the multi-rotor UAV to form a complete radar echo of the UAV under a strong sea clutter background, that is, the total radar echo;

[0076] The specific method of Step 1 is as follows:

[0077] Step 1.1: The time-domain expression of the LFM signal as the transmitted signal is as follows:

[0078]

[0079] where, A0 is the initial amplitude of the transmitted signal, f0 is the carrier frequency, B is the frequency modulation bandwidth, T c is the frequency modulation period, k = B / T c is the frequency modulation slope, and φ0 is the initial phase;

[0080] The time-domain expression of the target echo signal is:

[0081]

[0082] where, τ is the time delay;

[0083] Step 1.2: Subtract the phase of the transmitted signal in Step 1.1 from the phase of the target echo signal, and the phase of the baseband signal obtained is:

[0084] P b (t) = 2π(f0τ(t) + ktτ(t) - kτ 2 (t) / 2)

[0085] Step 1.3: Further obtain the beat signal from the phase of the baseband signal in Step 1.2 as:

[0086]

[0087] Step 1.4: Express the distance from the scattering point Q on a certain blade of a certain rotor of the UAV to the radar as:

[0088] R Q = R0 + vt + l Q cosβcos(θ0 + Ω m t - α)

[0089] where, R0 is the initial slant range of the target, v is the speed of the UAV, l Q is the distance from the scattering point Q to the blade rotation center, β is the pitch angle, θ0 is the initial blade rotation angle, Ω m is the angular velocity of the m-th rotor, and α is the azimuth angle;

[0090] Step 1.5: Substitute the distance R between the scatter point Q on a certain blade of a certain rotor of the drone obtained in Step 1.4 into the echo time delay τ = 2R Q / c, where c is the speed of light, and combine with the beat signal obtained in Step 1.3, and integrate over the entire blade length L to obtain the radar echo expression of the multi-rotor drone: Q where M is the number of rotors, N is the number of blades on one rotor, σ

[0091]

[0092] is the scattering coefficient of the blade, and σ mirco is the scattering coefficient of the drone body. main

[0093] The radar echo expression of the multi-rotor drone contains the micro-motion term of the blades on the drone rotors, as well as the distance and speed terms of the drone body. Moreover, the scattering coefficient of the drone body is 100 - 300 times that of its blades, which causes the main body scattering of the drone to cover the micro-Doppler caused by the rotation of its blades to a certain extent. As shown in (a) of Figure 2 , (c) of Figure 2 , and (e) of Figure 2 , only the echo generated by the drone body close to zero frequency can be seen in the time-frequency diagram and the Doppler spectrum; while after the joint filtering operation in the time-frequency-Doppler domain, as shown in (b) of Figure 2 , (d) of Figure 2 , and (f) of Figure 2 , the Doppler broadening generated by the rotation of the drone rotor blades can be seen in both the time-frequency diagram and the Doppler spectrum, and the micro-Doppler of the blades can be reappeared.

[0094] Analysis of (a) to (f) of Figure 2 shows that the period when the signal peak appears in the radar time-domain echo is consistent with the preset period of the drone blade rotation, the Doppler broadening in the drone time-frequency diagram is consistent with the Doppler broadening brought by the rotation of the rotor blades of the preset length, and the Doppler frequency shift in the Doppler spectrum is consistent with the frequency shift caused by the movement speed of the drone body. Figure 2

[0095] Step 2: Perform joint filtering in the time-frequency-Doppler domain on the total radar echo obtained in Step 1 to obtain the filtered total power spectral density P l ;

[0096] The specific method of Step 2 is as follows:

[0097] Step 2.1: Perform clutter suppression on the total radar echo in the slow-time dimension of the time domain, filter out stationary clutter, and retain moving targets to achieve the indication of moving targets; the impulse response of the corresponding filter for moving target indication is:

[0098]

[0099] where J is the number of pulse cancellations; the output signal is:

[0100] y1(t) = r s (t) * h J (q)

[0101] where r s (t) is the total radar echo;

[0102] Step 2.2: To make the signal-to-noise ratio of the filter output reach the maximum value at a certain specific moment, set the impulse response function of the matched filter:

[0103] h(t) = k0s * (t0 - t)

[0104] where k0 is a non-zero constant and s(t) is the target signal; perform matched filtering on the output signal y1(t) obtained in Step 2.1, and the output signal is:

[0105]

[0106] Step 2.3: To analyze the UAV radar echo with non-stationary signal characteristics, perform short-time Fourier transform on the output signal y2(t) obtained in Step 2.2:

[0107] STFT(T,F) = ∫y2(r)w(r - T)e -j2πFr dr

[0108] where w(·) is the window function, and the power spectral density of the transformed signal at this time is:

[0109] P(T,F) = k(T)|STFT(T,F)| 2

[0110]

[0111] where k(T) is the transformation coefficient to compensate for the influence of the window function, and f s is the sampling rate during short-time Fourier transform;

[0112] Step 2.4: Find the extreme values of the power spectral density calculated in Step 2.3, and select the mean value of the power spectral density at non-extreme frequencies in different time dimensions as the mean value of the clutter power spectral density, that is:

[0113] P1=P(T1,F1) / L T

[0114] Among them, L T The power spectrum density at the maximum frequency is selected to obtain the average value in different time dimensions as the target power spectrum density mean, that is:

[0115] P2=max(P(T,F)) / L T

[0116] Step 2.5: Take the ratio of the mean power spectral density of the clutter obtained in step 2.4 to the mean power spectral density of the target as the time-frequency filter factor γ, and establish a filter factor calculation model, namely:

[0117] γ=P1 / P2

[0118] Perform time-frequency filtering on the total power spectrum density. The specific expression is:

[0119] P l ={P[P(T,F)<γ*max(P(T,F))]=0}

[0120] At this point, the time-frequency-Doppler domain joint filtering of the total radar echo is achieved.

[0121] The measured sea surface data comes from a public dataset published in the Journal of Radar Studies and was obtained through a sea exploration test conducted by the Information Fusion Institute of the Naval Aviation University. This test used an X-band solid-state power amplifier surveillance / navigation radar, deployed on Yangma Island in Yantai, to detect the sea surface and surface targets (such as channel buoys and ships).

[0122] Figure 3 (a) Figure 3 (c) and Figure 3 (e) is the pulse-range diagram, time-frequency diagram, and Doppler spectrum obtained when the UAV data under strong sea clutter background is not filtered. Figure 3 (b) Figure 3 (d) and Figure 3 (f) is the pulse-range diagram, time-frequency diagram, and Doppler spectrum after joint filtering in the time-frequency-Doppler domain. The sea clutter data is level 3-4 sea conditions. Figure 3 (a) Figure 3 (c) and Figure 3 In (e), because clutter is not suppressed, a buoy and an island echo can be seen in the sea clutter data. Specifically, they are manifested as strong clutter at range bins 1762 and 2100-2800 in the pulse-range diagram, and strong clutter close to zero frequency in the time-frequency diagram and Doppler spectrum. The buoy is located 4.84 km away, and the island is located 6 km behind.

[0123] And Figure 3 in (b) of Figure 3 in (d) of Figure 3 and in (f) of

[0124] By analyzing Figure 3 it can be obtained that if the UAV data in the strong sea clutter background is not processed, only partial UAV target echoes can be seen in the Doppler domain, and their energy intensity is similar to that of the clutter, making it difficult to distinguish. After performing joint time-frequency-Doppler domain filtering on it, the echo energy of the UAV target is more concentrated, making it easier to distinguish from the clutter, and the UAV target can be clearly confirmed in both the pulse-distance diagram and the time-frequency diagram.

[0125] Step 3: Perform time-frequency ridge extraction on the filtered total power spectral density P l obtained in Step 2 to verify the fitting degree.

[0126] The specific method of Step 3 is as follows:

[0127] Take the ratio of the mean power spectral density of the clutter obtained in Step 2.4 to the mean power spectral density of the target as the time-frequency filtering factor, perform ridge extraction on the time-frequency diagram, compare the UAV time-frequency diagram ridge with the UAV time-frequency diagram ridge under the sea clutter background, and calculate its mean fitting diagram.

[0128] As Figure 4 shown, the upper and lower envelope trends of the two are the same, and the difference only fluctuates up and down around 10 Hz. Further analyze the upper and lower envelopes in the form of mean fitting. As Figure 5 shown, their trends are basically fitted, meeting the expectations, verifying that the filtering method of the present invention has a good suppression effect on sea clutter and an accurate inversion of UAV targets.

[0129] Therefore, the present invention proposes a UAV target inversion method based on joint time-frequency-Doppler domain filtering and time-frequency ridge analysis, which performs joint time-frequency-Doppler domain filtering on the signal, has a better suppression effect on sea clutter, can more easily distinguish the clutter from the target echo, and enables the UAV target echo to be better presented in the multi-dimensional domain.

[0130] The present invention combines the simulated UAV radar echo data with the measured sea clutter data to simulate the situation of a UAV flying over the sea surface, and performs multi-domain joint filtering operations such as moving target indication filtering in the time domain dimension, matched filtering in the range-Doppler dimension, and time-frequency filtering in the time-frequency dimension on the total echo data to suppress sea clutter, so as to detect UAV targets in multiple dimensions such as the time domain, frequency domain, time-frequency domain, and Doppler domain. At the same time, information such as the blade length, rotation speed, distance, and moving speed of the UAV is calculated from the processed data.

[0131] The present invention also provides a UAV target inversion system under a strong sea clutter background based on joint filtering in the time-frequency-Doppler domain and time-frequency ridge line analysis, including:

[0132] A UAV radar echo calculation module, which is used to implement, for the linear frequency modulation signal system radar in step 1, establish a multi-rotor UAV radar echo model and calculate the multi-rotor UAV radar echo;

[0133] A total radar echo filtering module, which is used to implement, in step 2, combine the UAV radar echo obtained in step 1 with the measured sea surface radar echo to form a complete UAV radar echo under a strong sea clutter background, that is, the total radar echo, and perform joint filtering in the time-frequency-Doppler domain on the total radar echo to obtain the filtered total power spectral density P l ;

[0134] A time-frequency ridge line extraction module, which is used to implement, in step 3, extract the time-frequency ridge line from the filtered total power spectral density P obtained in step 2 l to verify the fitting degree.

[0135] The present invention also provides a UAV target inversion device under a strong sea clutter background based on joint filtering in the time-frequency-Doppler domain and time-frequency ridge line analysis, including:

[0136] A memory: storing a computer program for the above-mentioned UAV target inversion method under a strong sea clutter background based on joint filtering in the time-frequency-Doppler domain and time-frequency ridge line analysis, which is a computer-readable device;

[0137] A processor: used to implement the above-mentioned UAV target inversion method under a strong sea clutter background based on joint filtering in the time-frequency-Doppler domain and time-frequency ridge line analysis when executing the computer program.

[0138] The present invention also provides a computer-readable storage medium, which stores a computer program, and when the computer program is executed by a processor, it can implement the above-mentioned UAV target inversion method under a strong sea clutter background based on joint filtering in the time-frequency-Doppler domain and time-frequency ridge line analysis.

Claims

1. A method for inverting UAV targets in strong sea clutter background based on time-frequency-Doppler domain joint filtering and time-frequency ridge analysis, characterized in that: The following steps are involved: Step 1: For linear frequency modulation signal radar, a multi-rotor UAV radar echo model is established and the multi-rotor UAV radar echo is calculated. The measured sea surface radar echo is added to the multi-rotor UAV radar echo to form a complete UAV radar echo under strong sea clutter background, that is, the total radar echo; Step 2: Perform time-frequency-Doppler domain joint filtering on the total radar echo obtained in step 1 to obtain the filtered total power spectrum density P l ; Step 3: The total power spectrum density P after filtering obtained in step 2 l Time-frequency ridge extraction was performed to verify the goodness of fit; The specific method of step 2 is: Step 2.1: Perform clutter suppression on the total radar echo in the slow time dimension of the time domain, filter out the stationary clutter, and retain the moving target to achieve the indication of the moving target; the corresponding filter impulse response of the moving target indication is: Where J is the number of pulse cancellations; the output signal is: y1(t)=r s (t)*h J (q) Among them, r s (t) is the total radar echo; Step 2.2: Set the impulse response function of the matched filter: h(t)=k0s * (t0-t) Where k0 is a constant not equal to 0, s(t) is the target signal; the output signal y1(t) obtained in step 2.1 is subjected to matched filtering, and the output signal is: Step 2.3: Perform a short-time Fourier transform on the output signal y2(t) obtained in step 2.2: STFT(T,F)=∫y2(r)w(r-T)e -j2πFr dr Where w(·) is the window function, and the power spectrum density of the transformed signal is: P(T,F)=k(T)|STFT(T,F)| 2 Among them, k(T) is the transformation coefficient to compensate for the influence of the window function, f s is the sampling rate when performing short-time Fourier transform; Step 2.4: Find the extreme value of the power spectrum density calculated in step 2.3, and take the average of the power spectrum density at the non-extreme frequency in different time dimensions as the mean power spectrum density of the clutter, that is: P1=P(T1,F1) / L T Among them, L T The power spectrum density at the maximum frequency is selected to obtain the average value in different time dimensions as the target power spectrum density mean, that is: P2=max(P(T,F)) / L T Step 2.5: Take the ratio of the mean power spectral density of the clutter obtained in step 2.4 to the mean power spectral density of the target as the time-frequency filter factor γ, and establish a filter factor calculation model, namely: γ=P1 / P2 Perform time-frequency filtering on the total power spectrum density. The specific expression is: P l ={P[P(T,F)<γ*max(P(T,F))]=0} At this point, the time-frequency-Doppler domain joint filtering of the total radar echo is achieved.

2. The method for inverting UAV targets in strong sea clutter background based on time-frequency-Doppler domain joint filtering and time-frequency ridge analysis according to claim 1 is characterized in that: The specific method of step 1 is: Step 1.1: The time domain expression of the linear frequency modulation signal as the transmitted signal is as follows: Among them, A0 is the initial amplitude of the transmitted signal, f0 is the carrier frequency, B is the frequency modulation bandwidth, T c is the frequency modulation period, k=B / T c is the frequency modulation slope, φ0 is the initial phase; The time domain expression of the target echo signal is: Where, τ is the time delay; Step 1.2: Subtract the phase of the transmitted signal from the target echo signal in step 1.

1. The resulting baseband signal phase is: P b (t)=2π(f0τ(t)+ktτ(t)-kt 2 (t) / 2) Step 1.3: The beat signal is further obtained from the baseband signal phase in step 1.2: Step 1.4: Express the distance from the scattering point Q of a blade on a certain rotor of the drone to the radar as: R Q =R0+vt+l Q cosβcos(θ0+Ω m t-a) Among them, R0 is the initial slant distance of the target, v is the speed of the UAV, l Q is the distance between the scattering point Q and the blade rotation center, β is the pitch angle, θ0 is the initial rotation angle of the blade, Ω m is the angular velocity of the mth rotor, α is the azimuth angle; Step 1.5: The distance R from the scattering point Q of a blade on a certain rotor of the UAV obtained in step 1.4 to the radar Q Substitute the echo delay τ = 2R Q / c, where c is the speed of light. Combined with the beat signal obtained in step 1.3, it is integrated over the entire blade length L to obtain the radar echo expression of the multi-rotor drone: Where M is the number of rotors, N is the number of blades on a rotor, and σ mirco is the scattering coefficient of the blade, σ main is the scattering coefficient of the UAV body.

3. The method for inverting UAV targets in strong sea clutter background based on time-frequency-Doppler domain joint filtering and time-frequency ridge analysis according to claim 1 is characterized in that: The specific method of step 3 is: The ratio of the mean power spectral density of the clutter obtained in step 2.4 to the mean power spectral density of the target is used as the time-frequency filtering factor. The ridges of the time-frequency graph are extracted. The ridges of the drone time-frequency graph are compared with the ridges of the drone time-frequency graph under the sea clutter background, and their mean fitting graphs are calculated.

4. A UAV target inversion system in a strong sea clutter background based on the method according to any one of claims 1 to 3, based on time-frequency-Doppler domain joint filtering and time-frequency ridge analysis, characterized in that: include: The UAV radar echo calculation module is used to establish a multi-rotor UAV radar echo model for linear frequency modulation signal system radar and calculate the multi-rotor UAV radar echo; The total radar echo filtering module is used to combine the UAV radar echo with the measured sea surface radar echo to form a complete UAV radar echo under the background of strong sea clutter, that is, the total radar echo, and perform time-frequency-Doppler domain joint filtering on the total radar echo to obtain the total power spectrum density P after filtering. l ; The time-frequency ridge extraction module is used to extract the total power spectrum density P after filtering. l Time-frequency ridge extraction was performed to verify the goodness of fit.

5. A UAV target inversion device based on time-frequency-Doppler domain joint filtering and time-frequency ridge analysis in strong sea clutter background, characterized by: include: Memory: a computer-readable device storing a computer program for a method for inverting UAV targets in a strong sea clutter background based on time-frequency-Doppler domain joint filtering and time-frequency ridge analysis as described in any one of claims 1 to 3; Processor: used to implement the UAV target inversion method in a strong sea clutter background based on time-frequency-Doppler domain joint filtering and time-frequency ridge analysis as described in any one of claims 1-3 when executing the computer program.

6. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, which, when executed by a processor, can implement a method for inverting UAV targets in a strong sea clutter background based on time-frequency-Doppler domain joint filtering and time-frequency ridge analysis as described in any one of claims 1 to 3.

Images