A method and apparatus for three-dimensional constant false alarm rate (CFAR) multi-target detection and estimation for low-precision quantized pulse phased array radar.

Abstract

This invention discloses a three-dimensional constant false alarm rate (CFAR) multi-target detection and estimation method for low-precision quantized pulse phased array radar. It achieves high-accuracy target parameter estimation by introducing Newton's method to correct the azimuth frequency, range, Doppler frequency, and amplitude values ​​in the received signal, and by fixing a portion of the detected results and iteratively correcting the remaining results. By incorporating the results of detected targets and recalculating the detection statistics, the influence of detected targets on the detection of the current target is eliminated. Furthermore, by introducing a Rao detector to achieve joint detection and estimation of target azimuth, range, and velocity, the system maintains its CFAR characteristics in multi-target detection scenarios. This invention also provides a three-dimensional CFAR multi-target detection and estimation device. The method provided by this invention achieves high-accuracy joint detection and estimation of target azimuth, range, and velocity, while maintaining CFAR characteristics in multi-target detection scenarios.

Description

Technical Field

[0001] This invention belongs to the field of radar signal processing technology, and in particular relates to a method and apparatus for three-dimensional constant false alarm rate (CFAR) multi-target detection and estimation for low-precision quantized pulse phased array radar. Background Technology

[0002] Pulse phased array radars are widely used in target detection and estimation due to their high range resolution, long-range detection capabilities, and highly flexible beamforming capabilities. However, these radar systems are limited by their required wide bandwidth and large-scale antenna deployment, resulting in extremely high power consumption and cost. A promising solution is to introduce low-resolution quantization technology. However, the signal after this nonlinear transformation will generate rich harmonic components. In scenarios where strong and weak targets coexist, weak targets are easily masked by strong targets, leading to a decrease in dynamic range (DR) and severely affecting detection performance. Meanwhile, traditional matched filter (MF) methods typically perform pulse compression (PC) in the "fast time" dimension and discrete Fourier transform (DFT) in the "slow time" dimension between multiple pulses. This method is subject to the "discrete grid" effect, and there is a risk of energy leakage except for targets located exactly at the range and DFT grid points.

[0003] Patent document CN111896927A discloses a communication-assisted radar target detection method based on 1-bit quantization sampling. The method includes: performing 1-bit quantization sampling on the received radar detection signal and base station communication signal, and comparing the results with a constructed detection operator and a detection threshold to determine whether a target exists. This method leverages the coverage advantage of communication signals and the low power consumption of 1-bit quantization, demonstrating some innovation in signal coordination and detection frameworks. However, this invention focuses only on determining the existence of a target and does not address the estimation of specific target parameters used for subsequent tracking, localization, and decision-making, thus limiting its applicability in practical applications. Furthermore, this invention lacks a systematic analysis of key issues such as limited accuracy and dynamic range compression caused by 1-bit quantization.

[0004] Patent document CN109828253B discloses a multi-station radar quantization fusion target detection method, comprising: firstly, setting up a multi-station radar system covering multiple radar stations and a signal fusion center. Each radar station receives echo signals and calculates detection statistics. Subsequently, each detection statistic undergoes a mapping transformation to generate a mapped statistic, and a quantization threshold for each statistic is calculated using a set false alarm probability. Next, each transformed statistic is quantized to obtain a quantized label value. These label values ​​are transmitted to the signal fusion center, where a final fusion decision is made regarding the presence or absence of a target. This method, through mapping transformation and the setting of quantization thresholds, improves the efficiency of quantization fusion target detection to some extent. However, this invention lacks correction and feedback mechanisms, and may have a high risk of missed detections or false alarms when facing complex targets or uncertain environments. Summary of the Invention

[0005] The purpose of this invention is to provide a three-dimensional constant false alarm rate (CFAR) multi-target detection and estimation method for low-precision quantized pulse phased array radar. It addresses the problems of limited dynamic range and harmonic generation caused by the high nonlinearity of low-precision quantization, as well as the problem of mutual masking or interference between targets in multi-target detection in complex scenarios. It achieves high-accuracy joint detection and estimation of target azimuth, range and velocity, and maintains constant CFAR characteristics in multi-target detection scenarios.

[0006] To achieve the first objective of this invention, the following technical solution is provided: a method and apparatus for three-dimensional constant false alarm rate (CFAR) multi-target detection and estimation for low-precision quantized pulse phased array radar, comprising the following steps:

[0007] Step 1: Obtain the quantized echo signal, filter the targets in the echo signal through a pre-constructed binary hypothesis testing model, and divide the signal into discrete grid points in the range and velocity dimensions. Use the Rao detector to filter out the existing effective targets as the first effective targets.

[0008] Step 2: Estimate the signal parameters of the first effective target;

[0009] The signal parameters of the first effective target are corrected using the Newton orthogonal matching algorithm, and the corrected first result is added to the effective target set.

[0010] Step 3: Input the corrected first result from Step 2 into the binary hypothesis testing model to eliminate the influence of the first result on the remaining undetected targets. Recalculate the detection statistic for the remaining undetected targets to screen out effective targets as the second effective targets, and estimate the signal parameters of the second effective targets.

[0011] The signal parameters of the second effective target are corrected using the Newton orthogonal matching algorithm, and the corrected second result is added to the effective target set.

[0012] Simultaneously, the second result is fixed, and the first result in the set of valid targets is iteratively corrected.

[0013] Step 4: Repeat steps 2-3 until the termination condition is met to obtain the final valid target set.

[0014] This invention achieves high-accuracy estimation of target parameters by introducing Newton's method to correct the azimuth frequency, range, Doppler frequency, and amplitude values ​​in the received signal, and by fixing some of the detected results and cyclically correcting the remaining detected results.

[0015] By taking into account the results of detected targets and recalculating the detection statistics, the influence of detected targets on the detection of current targets is eliminated, the problem of strong targets masking weak targets and making it difficult to detect weak targets is solved, and the dynamic range is effectively broadened.

[0016] By introducing a Rao detector to achieve joint detection and estimation of target azimuth, distance and velocity, the system can maintain constant false alarm characteristics in multi-target detection scenarios, thus solving the harmonic false alarm problem.

[0017] Specifically, the pulse phased array radar echo signal is a shore-based radar, an airborne radar, or a shipborne radar.

[0018] Specifically, the Rao detector and Newton's method require parameter initialization before use. These parameters include the false alarm probability, the maximum number of iterations, the oversampling factor, and the number of corrections.

[0019] Specifically, the expression for the low-precision quantization is as follows:

[0020] ;

[0021] ;

[0022]

[0023] ;

[0024] ;

[0025] ;

[0026] In this case, it is assumed that there are K targets in the echo signal. Indicates the amplitude value. Indicates distance, Indicates the azimuth frequency. Represents the Doppler frequency, defined ,make , indicating the complex amplitude A two-dimensional real vector formed by concatenating the real and imaginary parts of , and . It is additive white Gaussian noise, satisfying ,in For noise variance, , , , , and Let represent the indices of the fast time domain, slow time domain, and spatial domain, respectively. Let G represent the total number of received samples, G = NML. The sampling interval is... Indicates distance manifold vectors, The bandwidth is The baseband waveform, Let be the chirp rate of the linear frequency modulated pulse. The pulse width of a linear frequency modulated pulse. Indicates the carrier frequency. Indicates the speed of electromagnetic wave propagation. Indicates the frequency of the azimuth angle Array manifold vector; Indicates the frequency of Doppler array manifold vectors, It is a uniform quantizer. The tensor... Vectorization yields .

[0027] Specifically, the expression for the uniform quantizer is as follows:

[0028] ;

[0029] in Indicates uniform quantizer A bit quantizer, its output number is , It is the maximum full-scale range.

[0030] Specifically, the binary hypothesis testing problem corresponding to the binary hypothesis testing model is as follows:

[0031]

[0032] Among them, for the first to be detected Each target, and the detected targets will be placed... In the middle, non-zero threshold When detecting the first detection unit .

[0033] Specifically, the echo signal needs to undergo linear processing after filtering. The linear processing process is as follows:

[0034] The signal is pulse-compressed using matched filtering to obtain pulse-compressed data.

[0035] Doppler oversampling FFT is performed on the pulse-compressed data.

[0036] Specifically, the Rao detector calculates detection statistics by calculating all detection units and compares them with the threshold of the current valid target to determine whether there is a target in the echo signal.

[0037] Specifically, the detection process of the Rao detector is as follows:

[0038] In Rao detection, the pulse-compressed data expression is as follows:

[0039] ;

[0040] in, Representing a finite discrete set of distances, n This indicates a fast-time index. Represents the speed of light. Indicates the fast sampling frequency. Represents the number of samples in the distance dimension. This indicates the number of samples corresponding to the pulse width. This represents the oversampling factor relative to the traditional pulse compression grid.

[0041] The FFT process expression is as follows:

[0042]

[0043] in, Discrete sets representing velocities k This indicates a slow-time dimension index. This represents the oversampling factor relative to the traditional discrete Fourier transform lattice. M This represents the number of velocity samples.

[0044] To ensure the independence between detection units, the detection process is designed to... For all detection units corresponding to the discrete sets of distance and velocity dimensions, the detection statistic is calculated, which takes the following form:

[0045] ;

[0046] in , .

[0047] ;

[0048] ;

[0049] ;

[0050] The set false alarm probability

[0051] ;

[0052] By deducing the detection threshold for the current effective target;

[0053] ;

[0054] Take the highest value and the threshold from the detection statistics. If the value is higher than the threshold, then there is a valid target at the detection unit; otherwise, it is considered not to exist.

[0055] Specifically, the estimation process of the Rao detector is as follows:

[0056] For the azimuth dimension, the main lobe width of the array is generally small; therefore, in the detection and initial estimation stages, it can be assumed that... , where is a known quantity. The initial value is the preset azimuth frequency. It is then corrected in subsequent iterations to improve the estimation accuracy.

[0057] Assuming the target is detected and determined to exist, the coarse estimates of the actual target distance, Doppler frequency, and amplitude corresponding to its detection unit are obtained by the following formula:

[0058] ;

[0059] ;

[0060] in:

[0061] ;

[0062] in, This indicates that the variable value corresponding to the maximum value is returned. This represents a rough estimate of the distance. This represents a rough estimate of the Doppler frequency. , This represents a rough estimate of the target magnitude.

[0063] Specifically, the signal parameters include the target's azimuth frequency, range, and Doppler frequency, and the correction process is as follows:

[0064] The amplitude value of a fixed target is corrected by adjusting the azimuth frequency, range, and Doppler frequency, as expressed by the formula:

[0065] ;

[0066] in, This represents a combination of coarse estimates of azimuth frequency, range, and Doppler frequency before correction. This is the result after correction. express exist Regarding gradient, express exist Regarding The Hessian matrix.

[0067] ;

[0068] According to the revised Update range

[0069] ;

[0070] in, This represents the corrected amplitude value. The corrected target azimuth frequency, range, Doppler frequency, and amplitude value are then added to the valid target set.

[0071] Specifically, the termination conditions include:

[0072] a. The number of repetitions has reached the maximum number of iterations preset by the Rao detector;

[0073] b. The Rao detector determines that there is no target, that is, the detection statistic corresponding to the maximum value point does not exceed the corresponding threshold.

[0074] Specifically, the process of constructing the reconstructed dataset is as follows:

[0075] The new round of testing will be based on the condition that the first result exists, that is, at this time... Substituting this into a binary hypothesis testing model yields new null and alternative hypothesis tests, thereby eliminating the influence of the detected targets.

[0076] The detection statistic is recalculated for the new detection object obtained in this way, and the highest value point is taken as the second detection unit. It is then input into the detector and compared with the threshold for detection.

[0077] The azimuth frequency, range, Doppler frequency, and amplitude value corresponding to the second detection unit that is detected as a valid target are corrected using Newton's method, and the corrected second result is added to the set of valid targets. At the same time, the second result is fixed, and the first result is cyclically corrected using Newton's method. After correction, it is stored in the set of valid targets again.

[0078] Whenever a new valid target is detected, it is first corrected using Newton's algorithm and added to the target set. Then, the target is fixed, and all existing targets are corrected and updated in turn.

[0079] To achieve the second objective of this invention, the following technical solution is provided: a three-dimensional constant false alarm rate (CFAR) multi-target detection and estimation device, used to perform the steps of the above-described three-dimensional CFAR multi-target detection and estimation method for low-precision quantized pulse phased array radar.

[0080] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0081] (1) To address the problem of limited signal detection and estimation capabilities caused by the severe compression of available information in low-precision quantized signals, the Newton method was introduced to correct each parameter, and the remaining detection results were cyclically corrected by fixing some of the detected results to achieve high-accuracy estimation of the target parameters. By taking the results of the detected targets into consideration, new null and alternative hypothesis tests were formed, and the detection statistics were recalculated, eliminating the influence of the detected targets on the current target detection, solving the problem of strong targets masking weak targets, which makes it difficult to detect weak targets, and effectively broadening the dynamic range.

[0082] (2) Compared with the traditional pulse compression method based on matched filtering, the method based on grid points is modified by Newton's method, which overcomes the problems of energy leakage and grid mismatch, improves the estimation accuracy of azimuth, distance and velocity, and ensures that the system can still maintain constant false alarm characteristics in multi-target detection scenarios.

[0083] (3) For the problem of mutual influence between targets in space, the cyclic correction algorithm is used to correct all targets in turn, which reduces the mutual influence between targets and further improves the accuracy of the estimation results. Attached Figure Description

[0084] Figure 1 This is a flowchart of the three-dimensional constant false alarm rate (CFAR) multi-target detection and estimation method for low-precision quantized pulse phased array radar provided in this embodiment.

[0085] Figure 2 This is a schematic diagram of the three-dimensional constant false alarm rate (CFAR) multi-target detection and estimation device provided in this embodiment;

[0086] Figure 3This is a schematic diagram illustrating the detection results of multiple targets at similar distances, provided in this embodiment.

[0087] Figure 4 This is a schematic diagram illustrating the detection results of multiple targets with similar velocities provided in this embodiment;

[0088] Figure 5 This is a schematic diagram of the detection results when strong and weak targets coexist, as provided in this embodiment. Detailed Implementation

[0089] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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 embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0090] like Figure 1 As shown in this embodiment, a three-dimensional constant false alarm rate (CFAR) multi-target detection and estimation method for low-precision quantized pulse phased array radar is provided, which includes:

[0091] S110 acquires the pulse phased array radar echo signal after a low-precision quantization process and performs linear processing to obtain target data.

[0092] First, initialize the Rao detector and Newton's method by inputting the false alarm probability, maximum number of iterations, oversampling factor, and correction number.

[0093] For a pulse phased array radar employing a few-bit analog-to-digital converter (ADC), after down-conversion, the baseband received signal within one coherent processing interval (CPI) can be expressed as:

[0094] ;

[0095] In this case, it is assumed that there are K targets in the echo signal. Indicates the amplitude value. It is additive white Gaussian noise, satisfying ,in This represents the noise variance. , , , , and These represent the indices for the fast time domain, slow time domain, and spatial domain, respectively.

[0096] ;

[0097] The bandwidth is The baseband waveform. Nyquist sampling is used, with a sampling interval of [missing information]. .set up Indicates the Pulse Repetition Interval (PRI). . and These represent the chirp rate and pulse width of a linear frequency modulation (LFM) pulse, respectively. It is an indicator function that takes the value 1 when the condition is true and 0 otherwise. Indicates the carrier frequency. The wavelength is defined as the speed of electromagnetic wave propagation. .

[0098] Antenna spacing is In practice, it is generally used . , , and They represent the first The complex amplitude, radial distance, radial velocity, and azimuth of each target.

[0099] This signal model can be further represented in tensor form, which is more suitable for algorithm design:

[0100] ;

[0101] The first in The element is .

[0102] in

[0103]

[0104] ;

[0105] ;

[0106] ;

[0107] Indicates distance, Indicates the azimuth frequency. Represents the Doppler frequency. Definition ,make , indicating the complex amplitude A two-dimensional real vector formed by concatenating the real and imaginary parts of , and . The sampling interval is... Indicates distance manifold vectors, Indicates the frequency of the azimuth angle Array manifold vector; Indicates the frequency of Doppler Array manifold vectors.

[0108] It is a uniform quantizer. .

[0109] It converts a continuous numerical input signal into a finite number of discrete amplitude values ​​through a preset mapping relationship. The bit depth is... A uniform quantizer can be represented as:

[0110] ;

[0111] The output of the bit quantizer is . This is the maximum full-scale range; signals outside this range will be mapped to a specific quantization value.

[0112] A greedy strategy is employed to sequentially detect and estimate the parameters of new targets. It is assumed that the parameters have already been estimated... There are several objectives, and their parameter sets are as follows: It is necessary to determine the first Does the target exist? The above problem can be transformed into the following binary hypothesis testing (BHT) problem:

[0113] ;

[0114] Among them, for the first to be detected Each target, and the detected targets will be placed... In the middle, non-zero threshold When detecting the first detection unit .

[0115] Matched filtering is used to compress the signal into a pulse, resulting in compressed pulse data, which can be expressed by the formula:

[0116] ;

[0117] in, Let n represent a finite discrete set of distances, where n denotes the fast-time index. Represents the speed of light. Indicates the fast sampling frequency. Represents the number of samples in the distance dimension. This indicates the number of samples corresponding to the pulse width. This represents the oversampling factor relative to the traditional pulse compression grid.

[0118] The Doppler oversampling FFT of the pulse-compressed data is expressed by the following formula:

[0119] ;

[0120] in, The discrete set represents speed, and k represents the index of the slow time dimension. The value represents the oversampling factor relative to the traditional discrete Fourier transform grid, and M represents the number of velocity samples.

[0121] In Rao detection, to ensure that the detection units are independent of each other, Set during the estimation process .

[0122] For the azimuth dimension, the main lobe width of the array is generally small; therefore, in the detection and initial estimation stages, it can be assumed that... , where is a known quantity. The initial value is the preset azimuth frequency. It is then corrected in subsequent iterations to improve the estimation accuracy.

[0123] The method for detecting the existence of a valid target using a Rao detector is as follows:

[0124] First, the detection statistics are calculated for all detection units corresponding to the discrete sets of distance and velocity dimensions in S110. The form of these statistics is as follows:

[0125] ;

[0126] in , .

[0127] ;

[0128] ;

[0129] ;

[0130] The set false alarm probability

[0131] ;

[0132] Deducing the detection threshold of the current effective target

[0133] ;

[0134] In this embodiment, the highest value in the detection statistic and the threshold are taken. If the value is higher than the threshold, then there is a valid target at the detection unit; otherwise, it is considered not to exist.

[0135] The coarse estimates of the actual target distance, Doppler frequency, and amplitude corresponding to the detection unit where a target is determined to exist are obtained by the following formula:

[0136] ;

[0137] ;

[0138] in:

[0139] ;

[0140] This indicates that the variable value corresponding to the maximum value is returned. This represents a rough estimate of the distance. This represents a rough estimate of the Doppler frequency. , This represents a rough estimate of the target magnitude.

[0141] S120, In this embodiment, if the detection result of the first detection unit is a valid target, then the Newton method is used to correct the azimuth frequency, range, Doppler frequency, and amplitude value corresponding to the valid target, including:

[0142] The amplitude value of a fixed target is corrected by adjusting the azimuth frequency, range, and Doppler frequency, as expressed by the formula:

[0143] ;

[0144] in, This represents a combination of coarse estimates of azimuth frequency, range, and Doppler frequency before correction. This is the result after correction. express exist Regarding gradient, express exist Regarding The Hessian matrix.

[0145] ;

[0146] According to the revised Update range

[0147] ;

[0148] in, This represents the corrected amplitude value. The corrected target azimuth frequency, range, Doppler frequency, and amplitude value are then added to the valid target set.

[0149] S130 establishes the new round of testing on the condition that the first result exists, i.e., at this point... Substituting these values ​​into a binary hypothesis testing model yields new null and alternative hypothesis tests, thus eliminating the influence of already detected targets. The detection statistic is then recalculated for the new target data, and its highest value is selected as the second detection unit, input into the detector for comparison with a threshold.

[0150] The azimuth frequency, range, Doppler frequency, and amplitude value corresponding to the second detection unit detected as a valid target are corrected using Newton's method, and the corrected second result is added to the valid target set. Simultaneously, the second result is fixed, and the first result is iteratively corrected using Newton's method, and then stored in the valid target set again after correction. That is, whenever a new valid target is detected, it is first corrected using Newton's method and added to the target set, then the target is fixed, and all existing targets are sequentially corrected and updated.

[0151] In this embodiment, it is assumed that the estimated front The parameter set of each effective target is as follows: ,So Substituting this into the binary hypothesis testing model yields new null and alternative hypothesis tests, which can then be used to test the first hypothesis. Each target is detected and estimated. After correction, it is added to the set of valid targets.

[0152] Fixed number The results of each objective are analyzed using Newton's method. Each target is iteratively corrected, and after correction, it is stored again in the set of valid targets.

[0153] S140, repeat S120 and S130 until the Rao detector's termination condition is met, that is, all target information has been detected or the maximum number of iterations of the Rao detector has been reached, and the azimuth, range, velocity and amplitude information of all targets in the signal are obtained.

[0154] Then, by fixing the azimuth, distance, and velocity data, the amplitude of the target is uniformly corrected, and the final set of valid targets is output.

[0155] like Figure 2 As shown, this embodiment provides a three-dimensional constant false alarm rate (CFAR) multi-target detection and estimation device, which is used to perform the steps of the three-dimensional CFAR multi-target detection and estimation method for low-precision quantized pulse phased array radar provided in the above embodiment. It includes a quantization data acquisition unit 510, a Newton correction unit 520, a cyclic correction unit 530, and a target output unit 540.

[0156] The quantization data acquisition unit 510 is used to acquire the echo signal of the pulse phased array radar after low-precision quantization processing and perform linear processing to obtain target data.

[0157] The Newton correction unit 520 is used to calculate the detection statistics of the two-dimensional data in the range and velocity dimensions of the target data, and compares the highest value point as the first detection unit with the threshold. The Newton method is used to correct the azimuth frequency, range, Doppler frequency, and amplitude values ​​corresponding to the validly detected targets, and the corrected first result is added to the set of valid targets.

[0158] The cyclic correction unit 530 is used to establish a new round of detection based on the existing detection results, updating... The results are then substituted into a binary hypothesis testing model to obtain new null and alternative hypothesis tests, thereby eliminating interference from already detected targets. The detection statistic is recalculated for the new target data, and its highest value is taken as the second detection unit, which is then input into the detector and compared with a threshold for detection. Newton's method is used to correct the azimuth frequency, range, Doppler frequency, and amplitude values ​​corresponding to the second detection unit that is detected as a valid target, and the corrected second result is added to the valid target set. Simultaneously, the second result is fixed, and Newton's method is used to iteratively correct the first result, which is then stored in the valid target set again after correction. In other words, whenever a new valid target is detected, it is first corrected using Newton's method and added to the target set; then, the target is fixed, and all existing targets are sequentially corrected and updated.

[0159] The target output unit 540 is used to repeat the Newton correction unit 520 and the cyclic correction unit 530 until the Rao detector's judgment termination condition is met, and output the final set of valid targets, obtaining the azimuth frequency, range and Doppler frequency information of all targets in the signal.

[0160] To better illustrate the technical effects of the method provided in this embodiment, the following specific example process is provided.

[0161] The radar parameters were set as follows in the experiment: carrier frequency =3 GHz, pulse width T p =20us, pulse repetition period T=200us, bandwidth B=4MHz, sampling frequency =4MHz, pulse number 16, array unit number 32.

[0162] Experiment Example 1: Real target distances were set to 10000 m, 10020 m, and 10400 m; speeds to 80 m / s, 75 m / s, and 44 m / s; azimuth angles to 0.5°, 0.4°, and 0.2°; and target intensities to 25 dB, 24 dB, and 23 dB. In a multi-target proximity scenario, with a minimum real target proximity of 20 m, the target detection results are shown in the appendix. Figure 3 .

[0163] like Figure 3 (2) is the detection result of the method proposed in this invention (Generalized Newtonized Orthogonal Matching Pursuit for Pulsed Phased Array Radar, GNOMP-PPAR): distance error is less than 1m, velocity error is less than 1m / s, and azimuth error is less than 0.1°.

[0164] like Figure 3 (1) shows the detection result of the traditional pulse compression method based on matched filtering: Since the distance resolution of the matched filtering method is 37.5m, it cannot accurately detect two targets that are close to each other. The present invention has a significant improvement in both distance resolution and accuracy.

[0165] Experiment Example 2: Real target distances were set to 10200m, 10800m, and 12000m; velocities to 80m / s, 79m / s, and 40m / s; azimuth angles to 0.5°, 0.4°, and 0.2°; and target intensities to 25dB, 24dB, and 23dB. In this multi-target proximity scenario, the minimum difference in radial velocity between real targets was 1m / s. Target detection results are shown in the appendix. Figure 4 .

[0166] like Figure 4 (2) is the detection result of the GNOMP-PPAR method proposed in this invention: the detection result has a distance error of less than 1m, a velocity error of less than 1m / s, and an azimuth error of less than 0.1°.

[0167] like Figure 4 (1) shows the detection result of the traditional pulse compression method based on matched filtering: Since the velocity resolution of the matched filtering method is 15.6 m / s, it cannot accurately detect two targets with similar velocities. The present invention has a significant improvement in velocity resolution.

[0168] Experiment Example 3: Real-world target distances were set to 10200m, 10800m, and 12000m; speeds to 80m / s, 60m / s, and 40m / s; azimuth angles to 0.5°, 0.4°, and 0.2°; and target intensities to 25dB, 44dB, and 23dB. The difference in intensities between the strongest and weakest targets was 21dB. Target detection results in the scenario where strong and weak targets coexist are shown in the appendix. Figure 5 .

[0169] like Figure 5 (2) is the detection result of the GNOMP-PPAR method proposed in this invention: the detection result has a distance error of less than 1m, a velocity error of less than 1m / s, and an azimuth error of less than 0.1°.

[0170] like Figure 5 (1) shows the detection results of the traditional pulse compression method based on matched filtering. It can be seen that weak targets are missed when strong and weak targets coexist. This invention effectively reduces the influence of strong targets on weak targets and improves the detection probability of weak targets.

[0171] Furthermore, the terms "upper," "lower," "inner," "outer," "front," and "rear" are used for descriptive purposes only and should not be construed as indicating or implying relative importance. Unless otherwise specifically stated, the relative steps, numerical expressions, and values ​​of the components and steps set forth in these embodiments do not limit the scope of the invention.

[0172] Of course, the above description is only a specific embodiment of the present invention and is not intended to limit the scope of the present invention. All equivalent changes or modifications made to the structure, features and principles described in the claims of the present invention should be included in the scope of the claims of the present invention.

[0173] Finally, it should be noted that the above-described embodiments are merely specific implementations of the present invention, used to illustrate the technical solutions of the present invention, and not to limit it. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments within the technical scope disclosed in the present invention, or make equivalent substitutions for some of the technical features; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and 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 three-dimensional constant false alarm rate (CFAR) multi-target detection and estimation method for low-precision quantized pulse phased array radar, characterized in that, Includes the following steps: Step 1: Obtain the quantized echo signal, filter targets in the echo signal using a pre-constructed binary hypothesis testing model, and divide the signal into discrete grid points in the range and velocity dimensions. Use a Rao detector to filter out valid targets as the first valid targets. The expression for the low-precision quantization is as follows: ; ; ; ; ; Assume there are K targets in the echo signal. Indicates the amplitude value. Indicate distance, Indicates the azimuth frequency. Represents the Doppler frequency, defined ,make , indicating the complex amplitude A two-dimensional real vector formed by concatenating the real and imaginary parts of , and ; It is additive white Gaussian noise, satisfying ,in For noise variance, , , , , and These represent the indices for the fast time domain, slow time domain, and spatial domain, respectively. The sampling interval is... Indicates distance manifold vectors, The bandwidth is The baseband waveform, Let be the chirp rate of the linear frequency modulated pulse. The pulse width of a linear frequency modulated pulse. Indicates the carrier frequency. Indicates the speed of electromagnetic wave propagation. Indicates the frequency of the azimuth angle Array manifold vector; Indicates the frequency of Doppler array manifold vectors, It is a uniform quantizer; it converts tensors Vectorization yields Step 2: Estimate the signal parameters of the first effective target; The signal parameters of the first effective target are corrected using the Newton-Orthogonal Matching Algorithm, and the corrected first result is added to the effective target set. Step 3: Input the corrected first result from Step 2 into the binary hypothesis testing model to eliminate the influence of the first result on the remaining undetected targets. Recalculate the detection statistic for the remaining undetected targets to screen out effective targets as the second effective targets, and estimate the signal parameters of the second effective targets. The signal parameters of the second effective target are corrected using the Newton orthogonal matching algorithm, and the corrected second result is added to the effective target set. Simultaneously, the second result is fixed, and the first result in the set of valid targets is iteratively corrected. Step 4: Repeat steps 2-3 until the termination condition is met to obtain the final valid target set.

2. The method for three-dimensional constant false alarm rate (CFAR) multi-target detection and estimation for low-precision quantized pulse phased array radar according to claim 1, characterized in that, The expression for the uniform quantizer is as follows: ;in, Indicates uniform quantizer A bit quantizer whose output number is , It is the maximum full-scale range.

3. The method for three-dimensional constant false alarm rate (CFAR) multi-target detection and estimation for low-precision quantized pulse phased array radar according to claim 1, characterized in that, The echo signal needs to undergo linear processing after filtering. The linear processing procedure is as follows: The signal is pulse-compressed using matched filtering to obtain pulse-compressed data. Doppler oversampling FFT is performed on the pulse-compressed data.

4. The method for three-dimensional constant false alarm rate (CFAR) multi-target detection and estimation for low-precision quantized pulse phased array radar according to claim 1, characterized in that, The Rao detector calculates detection statistics by calculating all detection units and compares them with the threshold of the current valid target to determine whether there is a target in the echo signal.

5. The method for three-dimensional constant false alarm rate (CFAR) multi-target detection and estimation for low-precision quantized pulse phased array radar according to claim 4, characterized in that, The expression for the threshold is as follows: ;in, This represents the set false alarm probability, and G represents the total number of received samples.

6. The method for three-dimensional constant false alarm rate (CFAR) multi-target detection and estimation for low-precision quantized pulse phased array radar according to claim 1, characterized in that, The signal parameters include the target's azimuth frequency, range, and Doppler frequency, and the correction process is as follows: Construct the corresponding gradient and Hansen matrix based on the target's azimuth frequency, range, and Doppler frequency; The coarsely estimated amplitude, azimuth frequency, distance, and Doppler results are updated based on the gradient and Hansen matrix. The corrected azimuth frequency, range, Doppler frequency, and amplitude values ​​are added to the effective target set.

7. The method for three-dimensional constant false alarm rate (CFAR) multi-target detection and estimation for low-precision quantized pulse phased array radar according to claim 1, characterized in that, The termination conditions include: a. The number of repetitions has reached the maximum number of iterations preset by the Rao detector; b. The Rao detector determines that there is no target, that is, the detection statistic corresponding to the maximum value point does not exceed the corresponding threshold.

8. A three-dimensional constant false alarm rate (CFAR) multi-target detection and estimation device, characterized in that, The steps are for performing the three-dimensional constant false alarm rate (CFAR) multi-target detection and estimation method for low-precision quantized pulse phased array radar as described in any one of claims 1 to 7.

Images