An unmanned aerial vehicle ISAR imaging method based on improved PGCPF parameter estimation

Abstract

The application discloses an unmanned aerial vehicle ISAR imaging method based on improved PGCPF parameter estimation, and relates to the technical field of inverse synthetic aperture radar imaging. The method comprises the following steps: performing distance compression, slant compensation and translational compensation on received echo signals to obtain processed data; performing reconstruction on the processed data through a weighted PGCPF optimization algorithm based on differential evolution to obtain reconstructed signals; and finally performing two-dimensional ISAR imaging on the reconstructed signals. The application introduces a weight factor in a PGCPF estimation multi-component linear frequency modulated (LFM) signal method, improves the accuracy of estimated quadratic phase term and linear phase term coefficients, reduces error accumulation in subsequent estimation, and optimizes parameter selection in the estimation process through a differential evolution algorithm. The application takes the ISAR image entropy as an evaluation function, finds the best parameter combination, and obtains the optimal echo signal. The reconstructed echo signal of the estimated multi-component linear frequency modulated signal can effectively improve the imaging effect of a rotor unmanned aerial vehicle in non-stationary motion and improve the imaging quality.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of inverse synthetic aperture radar imaging technology, and particularly relates to a UAV ISAR imaging method based on improved PGCPF parameter estimation. BACKGROUND

[0002] Inverse synthetic aperture radar (ISAR) is a radar system for two-dimensional imaging of moving targets, which can obtain high-resolution images of targets at all times and all distances, and is widely used in imaging of various space targets. For targets such as UAVs, the moving rotor parts will produce a micro-Doppler effect superimposed in the radar echo of the UAV, resulting in the widening of the echo frequency band. Moreover, although the movement speed of the UAV is relatively slow, its movement mode is often non-stationary, resulting in that the Doppler frequency of each scatterer echo signal is usually time-varying, and the phase term of the radar echo develops to a high order. This makes the ISAR image obtained by the traditional range-doppler (RD) algorithm appear defocusing and trailing, which is difficult to identify.

[0003] For a target moving non-stationarily, the ISAR echo signal of the same range cell can be modeled as a multi-component linear frequency modulated (LFM) signal form, and the present application uses product generalized cubic phase function (PGCPF) estimation to extract this multi-component LFM signal. However, since the PGCPF estimation method may accumulate errors due to insufficient accuracy in estimating high-order phase term coefficients, it means that the initial small error may gradually accumulate with subsequent estimation or with the addition of more data points, resulting in a large deviation of the final estimation result. In order to reduce this influence, it is necessary to improve the estimation accuracy of PGCPF. In addition, the estimation accuracy is also affected by some variable parameters, and the selection of these parameters can directly affect the final estimation result and the final imaging quality. SUMMARY

[0004] In view of the above deficiencies in the prior art, the UAV ISAR imaging algorithm based on improved PGCPF parameter estimation provided by the present application can improve the estimation accuracy of the multi-component LFM signal, thereby effectively realizing high-resolution ISAR imaging of the rotor UAV moving non-stationarily.

[0005] In order to achieve the above-mentioned application purposes, the technical scheme adopted by the present application is as follows: a UAV ISAR imaging method based on improved PGCPF parameter estimation, which comprises the following steps:

[0006] S1, distance compression, dechirp compensation and planar compensation are performed on the received echo signal to obtain processed data;

[0007] S2, the processed data is reconstructed by a weighted PGCPF optimization algorithm based on differential evolution to obtain a reconstructed signal;

[0008] S3, two-dimensional ISAR imaging is performed on the reconstructed signal.

[0009] Further, the specific method of step S2 is as follows:

[0010] S2-1, the rotor unmanned aerial vehicle target radar echo signal with non-stationary motion is modeled as a multi-component linear frequency modulation signal, and the discrete form of the echo of the lth distance unit is l (n) is as follows:

[0011]

[0012] Where, B k , a k,0 , a k,1 , a k,2 respectively represent the amplitude, initial phase, center frequency and frequency modulation of the kth scattering point, K represents the total number of discrete points, n(t) represents the micro-Doppler interference generated by the rotor, which is a sinusoidal frequency modulation signal added to the echo signal, but the signal energy is less than the main echo energy;

[0013] S2-2, the multi-component LFM signal representing the echo is calculated, and the definition expression of PGCPF is:

[0014]

[0015] s * (-n+m)s * (-n-m)exp(-jβnm 2 )

[0016] Where, LFM represents a linear frequency modulation signal, PGCPF represents a product type generalized cubic phase function, m is the frequency domain sampling point number of PGCPF, s(n) is the input echo signal, N is the length of the signal, β is the frequency index, L is the time offset number, s * (n) represents the conjugate echo signal; initialize the parameters, let l=1, k=1;

[0017] S2-3, weight factors w1 and w2 are introduced, and the quadratic phase term coefficient is estimated according to the following formula

[0018]

[0019] where β1= argmax β |PGCPF(w1β)|

[0020] β2= argmax β |PGCPF(w1(β2-m / 2)+w2β)|

[0021] S2-4, compensate the quadratic phase term, calculate the coefficient of the linear phase term

[0022] S2-5, compensate the linear phase term again, calculate the amplitude of the LFM signal component and the initial phase coefficient

[0023] S2-6, subtract this LFM signal component from the original echo data, let k=k+1, return to S2-3, estimate the next LFM signal component, until k reaches the total number of scattering points, obtain the multi-component LFM signal of the distance unit;

[0024] S2-7, let l=l+1, return to S2-3, until the multi-component LFM signal of all distance units is estimated, reconstruct the echo signal by using the estimated signal;

[0025] S2-8, introduce the differential evolution algorithm to optimize the weight factors w1, w2 and the frequency domain resolution m of PGCPF;

[0026] Let the population size be NP, the population space L of the Gth iteration G As follows:

[0027]

[0028] X i,G represents a set of population individuals, respectively represent the population individuals about w1, w2 and m;

[0029] S2-9, randomly select three individual sets X r1,G ,X r2,G ,X r3,G , perform differential processing, let F v be a mutation factor, then the mutation individual set V i,G is:

[0030] V i,G = X r1,G +F v (X r2,G -X r3,G )

[0031] S2-10, the individuals in the population and the mutation individuals are crossed according to the probability CR to obtain trial individuals Wherein:

[0032]

[0033] Wherein, rand j (j,1) represents that j takes a random number in the interval 0 to 1;

[0034] S2-11, the current individual and the trial individual are brought into the inverse synthetic aperture radar echo to calculate the image entropy value, and the better one is selected as the next generation individual:

[0035]

[0036] Wherein, E(·) represents the function of calculating the image Shannon entropy;

[0037] S2-12, return to step S2-3, and continue to reconstruct the echo signal with the optimized parameters;

[0038] S2-13, the algorithm is ended until the iteration number G reaches the maximum value, or the convergence accuracy is satisfied, and the inverse synthetic aperture radar image with the optimal entropy is obtained.

[0039] Further, the specific method of step 2-4 is:

[0040]

[0041] Wherein,

[0042]

[0043] Further, the specific method of step 2-5 is:

[0044]

[0045] The unmanned aerial vehicle ISAR imaging algorithm based on the improved PGCPF parameter estimation has the advantages that the unmanned aerial vehicle ISAR imaging algorithm based on the improved PGCPF parameter estimation can obtain good imaging effect for the rotor unmanned aerial vehicle in non-stationary motion, can effectively remove the micro-Doppler interference generated by the rotor near the target signal, improve target recognition degree, and is favorable for high-resolution imaging of the unmanned aerial vehicle inverse synthetic aperture radar. BRIEF DESCRIPTION OF DRAWINGS

[0046] Figure 1 It is the specific flow chart of the algorithm proposed in the present application;

[0047] Figure 2 It is a target rotor unmanned aerial vehicle simulation model;

[0048] Figure 3 It is the imaging result graph of the fast minimum entropy RD;

[0049] Figure 4An imaging result map of the PGCPF estimation method.

[0050] Figure 5 An imaging result map of the algorithm provided by the present application. DETAILED DESCRIPTION

[0051] The specific embodiments of the present application are described below to facilitate the understanding of the present application for those skilled in the art, but it should be clear that the present application is not limited to the scope of the specific embodiments, and for those skilled in the art, as long as various changes are within the spirit and scope of the present application defined and determined by the appended claims, all the inventions utilizing the concept of the present application are within the scope of protection.

[0052] As shown in Figure 1 , an unmanned aerial vehicle ISAR imaging algorithm based on improved PGCPF parameter estimation includes the following steps:

[0053] S1, compressing the received echo signal in distance, compensating for slope and moving compensation to obtain processed data;

[0054] S2, reconstructing the processed data by a weighted PGCPF optimization algorithm based on differential evolution to obtain a reconstructed signal;

[0055] S3, performing two-dimensional ISAR imaging on the reconstructed signal.

[0056] The sub-steps of step S2 are as follows:

[0057] S2-1, model the rotor unmanned aerial vehicle target radar echo signal in non-stationary motion as a multi-component linear frequency modulation signal, and the discrete form of the echo of the lth distance unit is as follows:

[0058]

[0059] Where B k , a k,0 , a k,1 , a k,2 represent the amplitude, initial phase, center frequency and frequency modulation of the kth scattering point, respectively, and n(t) represents the micro-Doppler interference generated by the rotor, which is a sinusoidal frequency modulation signal added to the echo signal, but the signal energy is less than the main echo energy.

[0060] S2-2, estimate the multi-component LFM signal representing the echo by using the product type generalized cubic phase function, and the definition expression of PGCPF is

[0061]

[0062] s * (-n+m)s* (-n-m)exp(-jβnm 2 )

[0063] where m is the number of frequency domain sampling points of PGCPF, s(n) is the input echo signal, N is the length of the signal, and β is the frequency index. Initialize the parameters, let l = 1 and k = 1.

[0064] S2-3, introduce weight factors w1 and w2, respectively calculate PGCPF, and estimate the quadratic phase term coefficient according to the following formula

[0065] β1=argmax β |PGCPF(w1β)|

[0066] β2=argmax β |PGCPF(w1(β2-m / 2)+w2β)|

[0067]

[0068] S2-4, compensate the quadratic phase term, estimate the linear phase term coefficient according to the following formula

[0069]

[0070] S2-5, further compensate the linear phase term, estimate the amplitude and initial phase coefficient according to the following formula

[0071]

[0072] S2-6, subtract this LFM signal component from the original echo data, let k = k + 1, return to S2-3, estimate the next LFM signal component, until k reaches the total number of scattering points, and obtain the multi-component LFM signal of the distance unit;

[0073] S2-7, let l = l + 1, return to S2-3, until the multi-component LFM signal estimation of all distance units is completed, and reconstruct the echo signal using the estimated signal.

[0074] S2-8, introduce the differential evolution algorithm to optimize the weight factors w1, w2 and the frequency domain resolution m of PGCPF, set the population size as NP, and the population space of the Gth iteration is as follows:

[0075]

[0076] S2-9, randomly select three individuals X r1,G ,X r2,G ,X​r3,G Perform differential processing and set F v is the mutation factor, then the mutation individual is:

[0077] V i,G =X r1,G +F v (X r2,G -X r3,G )

[0078] S2-10, the individuals in the population and the mutant individuals are crossed according to the probability CR to obtain the test individuals in:

[0079]

[0080] S2-11. Bring the current individual and the test individual into the ISAR echo to calculate the image entropy value, and select the better one as the next generation individual:

[0081]

[0082] S2-12: Return to step S2-3 and continue to reconstruct the echo signal using the optimized parameters.

[0083] S2-13, until the number of iterations G reaches the maximum value, or the algorithm ends when the convergence accuracy is met, and the ISAR image with the optimal entropy is obtained.

[0084] In one embodiment of the present invention, in order to verify the performance of the proposed algorithm, we conducted a Matlab simulation experiment under the following conditions: the center frequency of the FM signal is 35 GHz, the bandwidth is 10 GHz, and the pulse repetition frequency is 833 Hz. Figure 2 The nine points in the middle of the simulation model represent the main body of the drone, the four points represent the center of each rotor, and the four points around it represent the two rotor blades. The initial speed of the drone is 5m / s and the acceleration is 0.5m / s. 2 , the rotor rotation speed is 40πrad / s. The differential evolution algorithm parameters are set as: NP=15, F v =0.4, CR = 0.5. Under the same simulation conditions, the target model was simulated using fast minimum entropy RD imaging, PGCPF estimation method imaging, and UAV ISAR imaging algorithm based on improved PGCPF parameter estimation.

[0085] from Figure 3 From the imaging results, it can be seen that the fast minimum entropy RD imaging method will produce blurred stripes for the rotor UAV with non-stationary motion. This is because the rotor generates micro-Doppler interference, and most of the scattering points also have defocusing problems due to the non-stationary motion. Figure 4 and Figure 5The imaging results of the PGCPF estimation algorithm and the algorithm proposed in this paper are shown. Because the ISAR echo signal is reconstructed, there are no blurring bands caused by micro-Doppler interference, and the scattering points are better focused. However, because the estimation accuracy of the traditional PGCPF is not as good as that of the algorithm proposed in this paper, some noise points caused by micro-Doppler interference still appear in the image, and the focusing effect of the scattering points is also poor.

[0086] Table 1 compares the results of different algorithms under the same simulation conditions. Image entropy is used to measure the quality of imaging results.

[0087] Algorithm Fast Minimum Entropy Algorithm PGCPF Estimation Algorithm Invention Algorithm Entropy Value 9.2050 5.4492 4.6384

[0088] As can be seen from the data in Table 1, the imaging results of the algorithm proposed in this paper have an image entropy approximately 0.8108 lower than that of the imaging results of the traditional PGCPF estimation method. When the image entropy is lower, the focusing effect is better and the image quality is better. Therefore, the imaging effect of the algorithm proposed in this paper is superior to the other two algorithms, and it can effectively reduce the defocusing problem caused by the non-stationary motion of the drone and the influence of the micro-Doppler effect generated by the rotor.

[0089] In summary, the UAV ISAR imaging algorithm based on the improved PGCPF parameter estimation in the present invention can effectively realize high-resolution ISAR imaging of rotorcraft UAVs performing non-stationary motion, and the estimation accuracy is also higher than that of the traditional PGCPF estimation method, and the image focusing effect is better.

Claims

1. An unmanned aerial vehicle ISAR imaging method based on improved PGCPF parameter estimation, the method comprising the following steps: S1. Distance compression, slant compensation and planar motion compensation are performed on the received echo signal to obtain processed data; S2. The processed data is reconstructed by a weighted PGCPF optimization algorithm based on differential evolution to obtain a reconstructed signal; S2-1, the rotor unmanned aerial vehicle target radar echo signal of non-stationary motion is modeled into multiple components of linear frequency modulation signal, the discrete form s of echo of the lth distance unit l (n) as follows: where B k , a k,0 , a k,1 , a k,2 represent the amplitude, initial phase, center frequency and frequency modulation of the kth scattering point respectively, K represents the total number of discrete points, and n(t) represents the micro-Doppler interference generated by the rotor, which is a sinusoidal frequency modulation signal attached to the echo signal, but the signal energy is less than the main echo energy; S2-2. A multi-component LFM signal representing the echo is calculated, and the definition expression of PGCPF is: wherein LFM represents a linear frequency modulation signal, PGCPF represents a product type generalized cubic phase function, m is a frequency domain sampling point number of the PGCPF, s(n) is an input echo signal, N is a length of the signal, β is a frequency index, L is a time offset number, s * (n) represents a conjugate echo signal; initializing parameters, letting l = 1 and k = 1; S2-3, introduce weight factors w1 and w2, and estimate the quadratic phase term coefficient according to the following formula where β1= argmax β |PGCPF(w1β)|; β2= argmax β |PGCPF(w1(β2-m / 2)+w2β)|; S2-4, compensate the second phase term, calculate the coefficient of the first phase term S2-5, compensating the phase term once more, calculating the amplitude of the LFM signal component and the initial phase coefficient S2-6. The LFM signal component is subtracted from the original echo data, k=k+1 is set, and S2-3 is returned to estimate the next LFM signal component until k reaches the total number of scattering points, and the multi-component LFM signal of the distance unit is obtained; S2-7. l=l+1 is set, and S2-3 is returned until the multi-component LFM signals of all distance units are estimated, and the echo signal is reconstructed by using the estimated signals; S2-8. A differential evolution algorithm is introduced to optimize the weight factors w1, w2 and m; Let NP be the population size, and L be the population space at the Gth iteration G As follows: X i,G represents a set of population individuals, respectively represents a population individual with respect to w1, w2, and m. S2-9, randomly select 3 individual sets X in the population r1,G , r2,G , r3,G Differential processing is performed, and let F be a mutation factor, then the mutation individual set V is: v i,G :​ V i,G = X r1,G + F v ( X r2,G - X r3,G ); S2-10, the individuals in the population and the varied individuals are crossed according to a probability CR to obtain trial individuals wherein: where rand j (0,1) represents a random number in the interval of 0 to 1; S2-11, bring the current individual and the trial individual into the inverse synthetic aperture radar echo to calculate the image entropy value, and select the better one as the next generation individual: Wherein, E(·) represents a function of calculating image Shannon entropy; S2-12. Step S2-3 is returned to continue reconstructing the echo signal with the optimized parameters; S2-13. The algorithm ends when the iteration number G reaches the maximum value or the convergence accuracy is satisfied, and an entropy-optimal inverse synthetic aperture radar image is obtained; S3. Two-dimensional ISAR imaging is performed on the reconstructed signal.

2. The UAV ISAR imaging method based on improved PGCPF parameter estimation of claim 1, wherein, The specific method of step S2-4 is: wherein 3. The UAV ISAR imaging method based on improved PGCPF parameter estimation of claim 1, wherein, The specific method of step S2-5 is:

Images