PRI agile radar system maneuvering target accumulation detection method
By improving the local quadratic interpolation method and two-dimensional parameter search, the problem of coherent accumulation and detection of maneuvering targets in the PRI agile radar system is solved, achieving efficient and accurate target detection and parameter estimation, which is applicable to different radar anti-jamming scenarios.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHANGHAI NORMAL UNIVERSITY
- Filing Date
- 2026-04-07
- Publication Date
- 2026-06-30
AI Technical Summary
In existing PRI agile radar systems, the coherent accumulation and detection of maneuvering targets suffer from high computational complexity and insufficient robustness, especially under non-uniform sampling conditions, making real-time processing and accurate detection difficult.
An improved local quadratic interpolation method is used to reconstruct the non-uniform sampling signal. Combined with the two-dimensional parameter search of Doppler ambiguity number and acceleration, motion compensation and coherence accumulation are performed. The non-uniform sampling signal is reconstructed into a uniform sampling signal through the local quadratic interpolation method. Motion correction is performed using Keystone transform and matched filter, and finally the target focused image is output.
It achieves accurate coherent accumulation and detection of maneuvering targets, reduces computational complexity, meets the real-time processing requirements of radar systems, is applicable to different PRI agility modes, and improves the robustness and accuracy of detection.
Smart Images
Figure CN122307578A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of radar signal processing technology, and in particular to a method for accumulating detection of moving targets in a PRI agile radar system. Background Technology
[0002] In recent decades, with the rapid development of high-threat military equipment technologies such as drones, stealth aircraft, and ballistic missiles, the detection, tracking, and parameter estimation of small targets with high dynamics and long range have become a key research focus and have received widespread attention. Coherent accumulation is a crucial signal processing technique that effectively improves the signal-to-noise ratio (SNR) and detection performance of radar signals by extending the observation time. During long-term observation, the motion of high-speed maneuvering targets introduces complex range migration (RM) and Doppler frequency migration (DFM), causing energy to diffuse in both the range and Doppler dimensions, severely affecting the accumulation effect.
[0003] For uniformly sampled signals, traditional signal processing methods based on Fast Fourier Transform (FFT) can effectively compensate for the energy diffusion drawback introduced by long-term coherent accumulation. However, modern radars often employ PRI-agile sampling modes to improve their resistance to electronic interference. In this mode, the target echo signal exhibits non-uniform sampling characteristics in the slow time dimension. Since traditional FFT-based signal processing methods strictly rely on the premise of uniform sampling, they cannot correctly focus energy under the PRI-agile system, resulting in severe broadening of the target Doppler spectrum and sidelobe elevation, thus causing coherent accumulation to fail.
[0004] Under the PRI agile system, existing technologies for detecting uniformly moving targets mainly employ improved Fourier transform (FT)-based methods, including Non-Uniform Discrete Fourier Transform (NUDFT), Non-Uniform Fast Fourier Transform (NUFFT), and Stochastic PRI Radon-Fourier Transform (RPRI-RFT). For detecting maneuvering targets, while existing technologies have proposed methods such as Non-Uniform Radon Fractional Fourier Transform (R-NUFRFT), these methods typically involve multi-dimensional parameter searches, resulting in high computational complexity and difficulty meeting the real-time processing requirements of radar systems. Furthermore, although there are detection methods that combine inverse range-frequency transform and fractional Fourier transform (RFRT-FRFT) with NUDFT, this method involves nonlinear transformations during the range-frequency inversion process, easily introducing cross-term interference and degrading target detection performance.
[0005] To address the aforementioned non-uniform sampling problem, existing technologies have proposed a signal reconstruction method based on Sinc interpolation. This method utilizes Sinc interpolation theory to reconstruct the non-uniform sampling sequence into a uniform grid, and then combines it with traditional two-dimensional frequency domain matched filtering techniques for motion compensation and target detection. Although this method can theoretically achieve relatively accurate target focusing and motion parameter estimation, in practical applications, it suffers from high computational complexity and insufficient robustness. Specifically, Sinc interpolation is essentially a global algorithm; the calculation of each output sampling point requires traversing all input sampling points for weighted summation, leading to high computational complexity. Especially when the radar pulse count is large, the computation time of this method increases significantly, severely restricting the real-time performance of target detection. Let the range gate number be... N The number of azimuth gate (pulse) units is M The computational complexity of the Sinc interpolation reconstruction algorithm is as high as [missing information]. O ( NM 2 The computational burden is relatively heavy. Furthermore, the global nature of Sinc interpolation reconstruction and its inherent low-pass filtering characteristics make it quite sensitive to the distribution pattern of sampling points. In the PRI periodic non-uniform agile sampling mode, Sinc interpolation may introduce reconstruction errors or "ghosting" phenomena, resulting in insufficient robustness. When using the Keystone transform (KT) to correct linear RM, this reconstruction error, combined with the approximation error inherent in the KT itself, can easily lead to target focusing failure.
[0006] A search revealed Chinese Patent Publication No. CN115575917A, which discloses a method, apparatus, and device for rapid detection of maneuvering targets using a random PRI radar. This scheme reconstructs a non-uniform slow-time signal into a uniformly sampled signal through a non-uniform sampling scale transformation and performs range travel correction. Then, it estimates the target's higher-order motion parameters through a parametric sub-path partitioning search strategy, constructs a phase compensation function to eliminate Doppler travel, and ultimately achieves maneuvering target detection. However, the non-uniform sampling scale transformation used in this method is essentially a variable substitution, and its transformation accuracy depends on the statistical distribution characteristics of the random time bias. When the non-uniform sampling mode is complex, the reconstruction accuracy is difficult to guarantee. Furthermore, while the parametric sub-path partitioning search reduces the computational load to some extent, it still requires dimensionality reduction search within a two-dimensional parameter space. When the target acceleration is large or the search range is wide, it is difficult to balance computational efficiency and estimation accuracy in terms of the number of sub-paths extracted and the search step size, potentially leading to a local optimum deviating from the global optimum.
[0007] Therefore, how to achieve coherent accumulation and detection of maneuvering targets under the condition of slow-time non-uniform sampling in PRI agile radar is a technical problem that needs to be solved. Summary of the Invention
[0008] The purpose of this invention is to overcome the shortcomings of the existing technology and provide a method for accumulating detection of maneuvering targets in a PRI agile radar system.
[0009] The objective of this invention can be achieved through the following technical solutions: According to a first aspect of the present invention, a method for accumulating detection of maneuvering targets in a PRI agile radar system is provided, comprising: The echo signal of the PRI agile radar is acquired, and the echo signal is subjected to range compression and range-dimensional Fourier transform to obtain the range frequency domain-azimuth time domain signal. Construct a two-dimensional parameter search space containing Doppler blur number and acceleration; For each set of parameters in the two-dimensional parameter search space, an improved local quadratic interpolation method is used to reconstruct the non-uniform slow-time sampled range-frequency domain-azimuth time domain signal into a uniformly sampled signal. Motion compensation and coherent accumulation are then performed on the uniformly sampled signal based on the current set of parameters to obtain the energy accumulation result under the current parameter combination. The improved local quadratic interpolation method constructs a quadratic polynomial in each sub-interval and determines the interpolation coefficients using a three-point template of the local neighborhood. Determine the global maximum value among the energy accumulation results corresponding to all parameter combinations, and take the Doppler ambiguity number and acceleration corresponding to the global maximum value as the optimal estimate; The optimal estimate is used to reconstruct, perform motion compensation and coherent accumulation on the range-frequency domain-azimuth-time domain signal, and output a focused image of the target.
[0010] As a preferred technical solution, the improved local quadratic interpolation method in the sub-interval formed by adjacent sampling points... Constructing a quadratic polynomial ,in, This is the resampling time; Using a local three-point template and their corresponding signal values To determine the interpolation coefficients so that the computation at each reconstruction point depends only on the three local input sampling points; where, the coefficients , , , , , y m Sampling points t m The function value at that location.
[0011] As a preferred technical solution, before performing interpolation reconstruction using the improved local quadratic interpolation method, phase compensation is performed on the range frequency domain-azimuth time domain signal to pre-compensate the phase corresponding to the Doppler ambiguity number currently being searched; After interpolation and reconstruction to a uniform time grid, inverse phase compensation is performed to restore the phase characteristics of the signal.
[0012] As a preferred technical solution, the phase compensation and inverse phase compensation use the same Doppler ambiguity number so that the phase pre-compensation before interpolation reconstruction matches the phase recovery after interpolation reconstruction.
[0013] As a preferred technical solution, motion compensation of the uniformly sampled signal specifically includes: Construct a range-frequency domain-azimuth-time domain matched filter that matches the Doppler ambiguity number and acceleration of the current search, the range-frequency domain-azimuth-time domain matched filter being used to simultaneously compensate for the second phase caused by Doppler ambiguity and acceleration; The uniformly sampled signal is multiplied by the matched filter to obtain the phase-compensated signal; The phase-compensated signal is coherently accumulated to obtain the energy accumulation result.
[0014] As a preferred technical solution, the expression for the range-frequency domain-azimuth-time domain matched filter is: , in, The Doppler blur number for the current search. a The acceleration of the current search. For distance frequency, For carrier frequency, The reference pulse repetition frequency, λ For wavelength, The reconstruction time is on a uniform slow time grid.
[0015] As a preferred technical solution, the coherent accumulation includes: The phase-compensated signal is subjected to range-dimensional inverse Fourier transform and azimuth-dimensional Fourier transform to obtain the target focused image under the current parameter combination, and the energy peak value is extracted from the target focused image as the energy accumulation result.
[0016] As a preferred technical solution, before performing motion compensation and coherent accumulation on the uniformly sampled signal, the method further includes: The uniformly sampled signal is subjected to Keystone transform to correct for linear distance migration.
[0017] As a preferred technical solution, determining the global maximum value among the energy accumulation results corresponding to all parameter combinations specifically includes: The energy accumulation results corresponding to each set of parameters are used to construct a two-dimensional energy accumulation matrix. The maximum value in the two-dimensional energy accumulation matrix is searched, and the Doppler ambiguity number and acceleration corresponding to the maximum value are the optimal estimates.
[0018] As a preferred technical solution, the computational complexity of the improved local quadratic interpolation method is O(NM), where N is the number of distance gate units and M is the number of pulses.
[0019] Compared with the prior art, the present invention has the following advantages: 1. This invention employs improved local quadratic interpolation to reconstruct non-uniform signals into uniform ones. Combined with two-dimensional parameter search of Doppler ambiguity number and acceleration, as well as motion compensation, it effectively overcomes the energy focusing failure problem caused by non-uniform sampling and achieves accurate coherent accumulation and detection of maneuvering targets.
[0020] 2. This invention employs an improved local quadratic interpolation method for signal reconstruction. The calculation of each reconstruction point depends only on a few local input sampling points, resulting in a computational complexity of O(NM), compared to the O(NM) of the traditional Sinc interpolation method. 2 The processing time has been increased by more than ten times, effectively reducing the computation time and meeting the real-time processing requirements of the radar system.
[0021] 3. This invention constructs a two-dimensional parameter search space containing Doppler ambiguity number and acceleration, organically integrating parameter search with interpolation reconstruction, motion compensation, and coherent accumulation, thereby achieving precise focusing of maneuvering targets and simultaneously and accurately estimating the target's Doppler ambiguity number and acceleration parameters, realizing integrated processing of detection and parameter estimation.
[0022] 4. The present invention is compatible with both PRI random agile and PRI periodic non-uniform agile sampling modes, which has a wider range of applications and can meet the needs of mobile target detection in different radar anti-jamming scenarios. Attached Figure Description
[0023] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a comparison chart of the reconstruction effects of the interpolation method in the embodiments of the present invention; wherein, Figure 2 (a) is a diagram showing the reconstruction result of standard piecewise quadratic spline interpolation; Figure 2 (b) is a diagram showing the reconstruction result of the improved local quadratic interpolation according to the present invention; Figure 3 This is a simulation result diagram of maneuvering target detection in the PRI random agility mode in an embodiment of the present invention; wherein, Figure 3 (a) is the original result after distance compression; Figure 3 (b) is the focusing result of the traditional NUDFT method; Figure 3 (c) and (d) show the focusing results of the method of the present invention on target 1 and target 2, respectively; Figure 3(e) and (f) are the parameter estimation results of targets 1 and 2 by the method of the present invention; Figure 3 (g) and (h) show the focusing results of traditional two-dimensional frequency domain matched filtering for targets 1 and 2; Figure 4 The figure shows the simulation results of maneuvering target detection under the PRI periodic non-uniform agile mode; among them, Figure 4 (a) is the original result after distance compression; Figure 4 (b) is a focusing result diagram of the method of the present invention; Figure 4 (c) is a graph showing the parameter estimation results of the method of the present invention; Figure 4 (d) shows the focusing result of traditional two-dimensional frequency domain matched filtering; Figure 5 These are simulation diagrams comparing the robustness of maneuvering target detection under different reconstruction methods in embodiments of the present invention; wherein, Figure 5 (a) is a focusing result diagram based on Sinc interpolation reconstruction; Figure 5 (b) is the focused spectrum reconstructed based on Sinc interpolation; Figure 5 (c) is a focusing result diagram of the improved local quadratic interpolation reconstruction based on the present invention; Figure 5 (d) is the focused spectrum diagram of the improved local quadratic interpolation reconstruction based on the present invention. Detailed Implementation
[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.
[0025] Example 1: like Figure 1 As shown, this invention proposes a method for accumulating detection of maneuvering targets in a PRI agile radar system, the method comprising: Step S1: Echo signal preprocessing.
[0026] The radar transmits a PRI-agile pulse train, receiving a non-uniformly sampled echo signal. Assuming the radar transmits a linear frequency modulated signal, the received baseband echo signal can be expressed as: , in, , The instantaneous slant distance of the target. The sampling time is non-uniform (caused by PRI agility). t To save time, T p The pulse width. μ To adjust the frequency, f c For carrier frequency, c At the speed of light, σ s The complex scattering coefficients of the target.
[0027] Range compression and range-dimensional FFT are performed on the echo signal to obtain the range-frequency-azimuth-time domain signal: , in, f r For distance frequency, B For signal bandwidth, v For the target radial velocity, a This represents the target's radial acceleration. At this point, the signal remains non-uniformly sampled in the slow time dimension.
[0028] Step S2: Non-uniform signal reconstruction and phase compensation based on two-dimensional parameter search.
[0029] To handle Doppler blur and phase changes caused by acceleration, a system containing Doppler blur numbers is constructed. M amb and acceleration a The search space for the two-dimensional parameters is defined. The search range for the Doppler ambiguity number is determined by the target's maximum possible velocity, and the search range for the acceleration is determined by the target's maximum possible maneuverability.
[0030] The parameter search employs a double-nested loop: the outer loop iterates through the Doppler ambiguity numbers. M amb Inner traversal acceleration a For each set of parameters ( M amb , a Perform the following processing: Step S21: Phase pre-compensation.
[0031] Before performing interpolation reconstruction, the Doppler blur number of the current search is used as a basis. M amb Phase compensation is performed on the range-frequency-azimuth-time domain signal to pre-compensate for the phase term caused by Doppler blur: , in, F 0 represents the fundamental frequency, typically taken as a reference value from the pulse repetition frequency. This compensation allows subsequent interpolation processing to be performed on the deblurred signal, improving interpolation accuracy.
[0032] Step S22: Improved local quadratic interpolation reconstruction.
[0033] An improved local quadratic interpolation method is used to transform the compensated signal from a non-uniform slow-time signal. t m Mapping to a uniform time grid η m superior.
[0034] The principle of the improved local quadratic interpolation method is as follows: In each local sub-interval Inside, the signal is approximated by a quadratic polynomial: , Coefficients are obtained through a local three-point template. and the corresponding phase-compensated signal value , ( To determine the interpolation coefficients, specifically: , , , in, , .
[0035] This method directly determines the interpolation coefficients using information from three adjacent nodes, so that the calculation of each reconstruction point depends only on the sampling points in the local neighborhood, avoiding the problems of global coupling and recursive error accumulation in traditional spline interpolation.
[0036] Using the above method, the compensated signal is transformed from a non-uniform slow-time signal. t m Interpolation to uniform time grid η m The interpolated signal is obtained.
[0037] Step S23: Reverse phase compensation.
[0038] After interpolation and reconstruction, inverse phase compensation is performed to restore the phase characteristics of the signal. , In this case, both phase compensation and inverse phase compensation use the same Doppler ambiguity number. M amb This ensures phase consistency before and after interpolation reconstruction, and avoids introducing additional errors due to parameter mismatch.
[0039] Thus, a uniformly sampled signal was obtained. .
[0040] Step S3: Motion compensation and coherence accumulation.
[0041] Step S31: Keystone transform corrects linear distance migration.
[0042] The Keystone Transform (KT) is used to correct linear distance migration. The expression for KT is: , in, τ m This represents the transformed slow-time variable. Substituting this expression into the uniformly sampled signal eliminates the linear distance migration term, yielding a signal with linear distance migration corrected. .
[0043] Step S32: Matched filter construction and phase compensation.
[0044] Based on the current Doppler blur number search M amb and acceleration a A range-frequency domain-azimuth time domain matched filter is constructed. This matched filter is used to simultaneously compensate for Doppler blur and the second phase caused by acceleration, and its expression is: , in, λ is the wavelength.
[0045] The signal processed by KT Multiplying by the matched filter yields the ambiguity number. and acceleration The compensated signal: .
[0046] Step S33: Coherent accumulation.
[0047] Perform inverse range Fourier transform (IFFT) and azimuth FFT on the phase-compensated signal to achieve coherent energy accumulation: , in, t For distance and time, f a The Doppler frequency is used. The energy peak of this focused image is extracted as the current parameter combination. M amb , a The result of energy accumulation.
[0048] Step S4: Parameter estimation and final focusing.
[0049] After iterating through all parameter combinations, a two-dimensional energy accumulation matrix is constructed from all energy accumulation results. The global maximum value in this matrix is then searched, and the Doppler blur number and acceleration corresponding to this maximum value are the optimal estimates. and the optimal estimate of acceleration .
[0050] Using the optimal estimate Replacement parameters Repeat steps S2 and S3 for reconstruction, motion compensation, and coherence accumulation to output the final target-focused image.
[0051] To verify the effectiveness of this invention, simulation experiments were conducted to verify its computational efficiency, detection performance under the PRI random agile mode, applicability under the PRI periodic non-uniform agile mode, and robustness compared to the Sinc interpolation method. The simulation platform was configured with an AMD Ryzen 5-5500U processor (2.10GHz) and 16GB of memory.
[0052] Experiment 1: Comparison Experiment of Computational Efficiency Simulation.
[0053] Experimental setup: Set a motion target with the following parameters: initial distance. ,speed acceleration ; The radar system parameters are as follows: carrier frequency signal bandwidth sampling frequency pulse number Pulse duration PRI: in Random and agile changes within a range.
[0054] In the number of pulses respectively M When the values are 1500, 2000, 2500, and 3000, the reconstruction time of the improved local quadratic interpolation method of this invention is compared with that of the Sinc interpolation method.
[0055] Experimental results show that when M is 1500, 2000, 2500, and 3000, the reconstruction times required by the method of this invention are 4.41s, 6.86s, 9.49s, and 11.83s, respectively. Under the same conditions, the reconstruction times required by the Sinc interpolation method are 50.16s, 86.20s, 135.07s, and 192.13s, respectively. The results indicate that, for the same number of pulses, the computational efficiency of the improved local quadratic interpolation method of this invention is more than ten times higher than that of Sinc interpolation, and the speed advantage becomes increasingly significant as the number of pulses increases.
[0056] Furthermore, the signal reconstruction effects of standard piecewise quadratic spline interpolation and the improved local quadratic interpolation of this invention are compared, such as... Figure 2As shown in (a), the standard piecewise quadratic spline interpolation reconstruction result shows severe spectral leakage due to the accumulation of recursive errors. This manifests as numerous transverse stray fringes and energy diffusion, with the target trajectory being submerged by noise and interference, making it impossible to clearly distinguish. Figure 2 (b) The improved local quadratic interpolation reconstruction results of the present invention effectively avoid recursion error and spectral leakage, the target energy is highly concentrated, presenting a clear and continuous oblique motion trajectory, which is clearly distinguished from the noise base, and the reconstruction accuracy and robustness are significantly better.
[0057] Experiment 2: Simulation of PRI random agile target detection.
[0058] Experimental setup: Set up two maneuvering targets with different intensities, with the following parameters: Target 1: Initial distance ,speed acceleration Signal-to-noise ratio ; Objective 2: Initial distance ,speed acceleration Signal-to-noise ratio ; The radar system parameters are the same as in Experiment 1.
[0059] like Figure 3 As shown in (a), the target energy is dispersed due to the non-uniform sampling and maneuvering of the PRI, exhibiting only a weak oblique trajectory, making it difficult to detect directly; as Figure 3 As shown in (b), the focusing result of the traditional non-uniform discrete Fourier transform method (NUDFT) is that the target trajectory is curved due to the influence of acceleration, and the energy is still not effectively concentrated, thus limiting the detection performance; Figure 3 (c) and (d) show the focusing results of the method of the present invention. The energy of both targets is focused into a sharp single peak, and clear focusing can be achieved even under the condition of low signal-to-noise ratio of target 2, with a significant improvement in signal-to-noise ratio. Figure 3 (e) and (f) show the parameter estimation results. The proposed method can accurately locate the global peak in the two-dimensional parameter space and accurately estimate the Doppler ambiguity number and acceleration of the target, which are consistent with the preset motion parameters. Figure 3 (g) and (h) show the results of the traditional two-dimensional frequency domain matched filtering method. The energy of targets 1 and 2 is submerged in the noise floor, making effective detection and focusing impossible. The results show that the method proposed in this invention can effectively eliminate the adverse effects of non-uniform sampling, achieve accurate focusing of two maneuvering targets, and accurately estimate the motion parameters of the targets.
[0060] Experiment 3: Simulation of PRI periodic non-uniform agile target detection.
[0061] Experimental setup: Set up a moving target with the following parameters: initial distance ,speed acceleration Signal-to-noise ratio ; PRI is set to periodic non-uniform agile mode: PRI is fixed at 3ms, and the transmission time sequence of the first pulse train is as follows: A total of 200 pulse trains were transmitted. The radar system parameters were the same as in Experiment 1.
[0062] like Figure 4 As shown in (a), the target energy is dispersed due to the non-uniform sampling of the PRI period and the maneuvering migration, and only shows a weak oblique trajectory, which is difficult to detect directly; Figure 4 (b) is the focusing result of the method of the present invention. The target energy is focused into a sharp single peak, achieving clear range-Doppler domain energy focusing and effectively eliminating the adverse effects of non-uniform sampling. Figure 4 (c) shows the parameter estimation results. The method of the present invention accurately locates the global peak in the two-dimensional parameter space of Doppler ambiguity number and acceleration, and accurately estimates the motion parameters of the target, which are consistent with the preset values. Figure 4 (d) shows the result of the traditional two-dimensional frequency domain matched filtering method. The target energy is almost completely submerged in the noise floor, with only weak stray peaks remaining, making effective detection and focusing impossible. The results indicate that the method proposed in this invention is also applicable to the PRI periodic non-uniform agile sampling mode, effectively overcoming the influence of non-uniform sampling and the migration of maneuvering targets, achieving accurate target focusing and accurate estimation of motion parameters, and outperforming traditional methods.
[0063] Experiment 4: Robustness Comparison Simulation.
[0064] Experimental setup: Set a moving target with the following parameters: initial distance ,speed acceleration Signal-to-noise ratio .
[0065] PRI is set to periodic non-uniform agile mode: the repetition period of the pulse train is fixed at 3ms, and the transmission time sequence of the first pulse train is as follows: The total number of pulse trains is 200. The radar system parameters are: , , , .
[0066] like Figure 5 (a) and (b) are the results of reconstruction based on Sinc interpolation. Although the target forms the main focusing peak, there are false peaks in the spectrum, which can easily cause false target interference and lack robustness. Figure 5 (c) and (d) show the results of the improved local quadratic interpolation reconstruction based on this invention. The target energy is focused into a single sharp main peak, the spectrum is clean and free of spurious peaks, achieving precise focusing and effectively avoiding reconstruction errors caused by non-uniform sampling. The results show that the method of this invention is free from spurious peak interference in the non-uniform agile PRI scenario, has stronger robustness and a wider range of applications, and is superior to the traditional Sinc interpolation reconstruction method.
[0067] This invention constructs a two-dimensional parameter search space for Doppler blur number and acceleration, and performs phase compensation, improved local quadratic interpolation reconstruction, and inverse compensation on each set of parameters to efficiently map non-uniform signals to a uniform time grid. The improved local quadratic interpolation directly determines the interpolation coefficients using a local three-point template, eliminating the need for global matrix inversion and avoiding spectral leakage caused by recursive error accumulation in standard piecewise quadratic splines. Based on this, Keystone transform and matched filtering are combined to complete motion compensation and coherence accumulation. The optimal estimate is obtained through parameter search, and the focused image of the target is output. Compared to existing Sinc interpolation methods, this invention reduces computational complexity from O(NM) to O(NM). 2 The efficiency is reduced to O(NM), which is more than ten times higher. It can achieve precise focusing and no spurious peaks in different PRI agility modes, and accurately estimate the Doppler ambiguity number and acceleration of maneuvering targets.
[0068] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.
Claims
1. A PRI agile radar system maneuvering target accumulation detection method, characterized in that, include: The echo signal of the PRI agile radar is acquired, and the echo signal is subjected to range compression and range-dimensional Fourier transform to obtain the range frequency domain-azimuth time domain signal. Construct a two-dimensional parameter search space containing Doppler blur number and acceleration; For each set of parameters in the two-dimensional parameter search space, an improved local quadratic interpolation method is used to reconstruct the non-uniform slow-time sampled range-frequency domain-azimuth time domain signal into a uniformly sampled signal. Motion compensation and coherent accumulation are then performed on the uniformly sampled signal based on the current set of parameters to obtain the energy accumulation result under the current parameter combination. The improved local quadratic interpolation method constructs a quadratic polynomial in each sub-interval and determines the interpolation coefficients using a three-point template of the local neighborhood. Determine the global maximum value among the energy accumulation results corresponding to all parameter combinations, and take the Doppler ambiguity number and acceleration corresponding to the global maximum value as the optimal estimate; The optimal estimate is used to reconstruct, perform motion compensation and coherent accumulation on the range-frequency domain-azimuth-time domain signal, and output a focused image of the target.
2. The method for accumulating and detecting maneuvering targets in a PRI agile radar system according to claim 1, characterized in that, The improved local quadratic interpolation method is used to construct a quadratic polynomial in a subinterval composed of adjacent sampling points constructing a quadratic polynomial wherein, is a resampling time point; Using a local three-point template and their corresponding signal values To determine the interpolation coefficients so that the computation at each reconstruction point depends only on the three local input sampling points; where, the coefficients , , , , .
3. The method for accumulating and detecting maneuvering targets in a PRI agile radar system according to claim 1, characterized in that, Before performing interpolation reconstruction using the improved local quadratic interpolation method, phase compensation is performed on the range frequency domain-azimuth time domain signal to pre-compensate the phase corresponding to the Doppler ambiguity number currently being searched; After interpolation and reconstruction to a uniform time grid, inverse phase compensation is performed to restore the phase characteristics of the signal.
4. The method for accumulating and detecting maneuvering targets in a PRI agile radar system according to claim 3, characterized in that, The phase compensation and inverse phase compensation use the same Doppler ambiguity number to match the phase pre-compensation before interpolation reconstruction with the phase recovery after interpolation reconstruction.
5. The method for accumulating and detecting maneuvering targets in a PRI agile radar system according to claim 1, characterized in that, Motion compensation for the uniformly sampled signal specifically includes: Construct a range-frequency domain-azimuth-time domain matched filter that matches the Doppler ambiguity number and acceleration of the current search, the range-frequency domain-azimuth-time domain matched filter being used to simultaneously compensate for the second phase caused by Doppler ambiguity and acceleration; The uniformly sampled signal is multiplied by the matched filter to obtain the phase-compensated signal; The phase-compensated signal is coherently accumulated to obtain the energy accumulation result.
6. The method for accumulating and detecting maneuvering targets in a PRI agile radar system according to claim 5, characterized in that, The expression for the range-frequency domain-azimuth-time domain matched filter is: ， in, The Doppler blur number for the current search. a The acceleration of the current search. For distance frequency, For carrier frequency, The reference pulse repetition frequency, λ For wavelength, The reconstruction time is on a uniform slow time grid.
7. The method for accumulating and detecting maneuvering targets in a PRI agile radar system according to claim 1, characterized in that, The coherent accumulation includes: The phase-compensated signal is subjected to range-dimensional inverse Fourier transform and azimuth-dimensional Fourier transform to obtain the target focused image under the current parameter combination, and the energy peak value is extracted from the target focused image as the energy accumulation result.
8. The method for accumulating and detecting maneuvering targets in a PRI agile radar system according to claim 1, characterized in that, Before performing motion compensation and coherent accumulation on the uniformly sampled signal, the method further includes: The uniformly sampled signal is subjected to Keystone transform to correct for linear distance migration.
9. The method for accumulating and detecting maneuvering targets in a PRI agile radar system according to claim 1, characterized in that, Determining the global maximum value from the energy accumulation results corresponding to all parameter combinations specifically includes: The energy accumulation results corresponding to each set of parameters are used to construct a two-dimensional energy accumulation matrix. The maximum value in the two-dimensional energy accumulation matrix is searched, and the Doppler ambiguity number and acceleration corresponding to the maximum value are the optimal estimates.
10. The method for accumulating and detecting maneuvering targets in a PRI agile radar system according to claim 1, characterized in that, The computational complexity of the improved local quadratic interpolation method is O(NM), where N is the number of distance gate units and M is the number of pulses.