Ultrahigh-resolution millimeter wave radar output side lobe suppression method

Abstract

The invention belongs to the technical field of signal processing, and discloses an ultrahigh-resolution millimeter-wave radar output side lobe suppression method, which comprises the following two steps of: 1, performing two-dimensional mathematical modeling on a process of transmitting, target and receiving of an ultrahigh-resolution millimeter-wave radar detection target, and processing a received millimeter-wave radar echo signal; therefore, the distance-angle two-dimensional spectrum data of the ultrahigh-resolution millimeter-wave radar about the target is obtained. And secondly, corrected two-dimensional CLEAN algorithm processing is carried out on the distance-angle two-dimensional spectrum data, a candidate target parameter matrix is constructed through local neighborhood search of the first K maximum peak values, then the real target position is determined, and the sidelobe level of the distance-angle two-dimensional spectrum is reduced. According to the method, the target detection rate of the weak target in the two-dimensional scene can be greatly improved on the basis of meeting the accuracy requirement. Meanwhile, under the complex problems of high sidelobe level, false peaks, peak position offset and the like, accurate target detection performance can still be realized.

Description

Technical Field

[0001] The invention belongs to the technical field of signal processing, and in particular relates to an ultra-high resolution millimeter wave radar output sidelobe suppression method. Background Art

[0002] At present, the side lobes in the range or angle direction of radar data can greatly limit the dynamic range of radar system detection, reduce image contrast, and cause the range resolution or angle resolution of the radar to fail to reach the theoretical resolution. At this time, the echo of the strong target will interfere with the echo signal in the surrounding range gate or angle gate through the side lobe, thereby affecting or even drowning other weak real echo signals, reducing the accuracy of radar measurement. Even though high-resolution millimeter-wave radar is widely used due to its own advantages, such problems cannot be easily ignored. Therefore, in order to improve the measurement accuracy of ultra-high-resolution millimeter-wave radar on targets, it is necessary to suppress the side lobes in the radar data in both the range and angle directions.

[0003] The CLEAN deconvolution algorithm solves the problem of how to quickly and effectively restore the real target and suppress the sidelobes in the polluted scene. It is the earliest deconvolution technology used in wireless astronomy and is also used in sidelobe suppression. Its basic idea is to extract the strongest signal one by one and eliminate the influence of the sidelobes on other signals. The CLEAN algorithm can reconstruct the ideal point spread function at the peak position every time the maximum value is found, and subtract it from the current image, iteratively and cyclically execute the process of "finding the maximum value-reconstructing the ideal point spread function-subtraction", and finally obtain the optimized clean result image. The mainstream CLEAN algorithms are as follows: coherent CLEAN algorithm, sequence-CLEAN algorithm, Lean-CLEAN algorithm, active-CLEAN algorithm, adaptive CLEAN algorithm, etc. CLEAN algorithms are complex and diverse, and different algorithms can complement each other, which will improve the recovery effect of the real target in the scene to varying degrees.

[0004] However, the existing CLEAN algorithm has the following problems:

[0005] 1. There is no detailed algorithm to explain the scenario of ultra-high resolution millimeter-wave radar, and there is no detailed introduction to analyze the sidelobe suppression of radar data collected using this system radar in the range-azimuth two-dimensional spectrum.

[0006] 2. In actual scenarios, weak targets are often obscured by the side lobes of strong targets, resulting in errors in target detection. And targets usually exist continuously, which is contrary to the ideal assumption that "point targets are independent and isolated". Although continuous targets can be modeled as a series of closely spaced different point targets, these closely spaced point targets will be affected by factors such as sidelobe-mainlobe, sidelobe-sidelobe coherent interference and noise in the two-dimensional spectrum, resulting in problems such as false peaks or peak position shifts in the two-dimensional spectrum. These will cause certain errors in the correct estimation of target parameters, seriously affecting the subsequent target detection effect. Therefore, for the problem of weak targets being obscured by strong targets and continuous target detection, improving the accuracy and stability of sidelobe suppression technology is a very meaningful topic.

[0007] 3. The sidelobe suppression technology for the range-azimuth two-dimensional spectrum is not mature enough, and there are few related achievements. Most of the existing sidelobe suppression technologies based on the CLEAN algorithm are introduced with one-dimensional range spectrum. Although they are applicable to two-dimensional scenes, they do not provide a detailed derivation process of the two-dimensional data model. Therefore, there is still a lot of room for development and research value in this regard. The current sidelobe suppression technology for the CLEAN algorithm mainly uses the coherent CLEAN algorithm, the Sequence-CLEAN algorithm, and the Lean-CLEAN algorithm to achieve target detection and recognition. The sidelobe suppression technology using the two-dimensional CLEAN algorithm in two-dimensional scenes has high practicality, but also faces severe challenges. Summary of the invention

[0008] The present invention proposes an ultra-high resolution millimeter-wave radar output sidelobe suppression method based on a modified two-dimensional CLEAN algorithm based on local neighborhood search. The invention mainly aims at the problems of strong target blocking weak target, excessive sidelobe, false peak, peak position offset and the like in the range-azimuth two-dimensional spectrum of targets in two-dimensional scenes. The invention proposes a sidelobe suppression method suitable for two-dimensional scenes and capable of meeting high accuracy and stability.

[0009] To achieve the above object, the present invention provides the following technical solutions:

[0010] A method for suppressing output sidelobe of ultra-high resolution millimeter-wave radar comprises the following steps:

[0011] Step 1: 2D data modeling of ultra-high resolution millimeter-wave radar echo signals

[0012] Step 1: Establish radar transmission signal expression;

[0013] Step 2: Establish the target echo delay expression;

[0014] Step 3: Establish radar echo signal expression;

[0015] Step 4: Establish radar beat signal expression under single cycle;

[0016] Step 5: Establish a simplified radar beat signal expression;

[0017] Step 6: Establish radar beat signal expression under multi-cycle;

[0018] Step 7: Establish a discrete expression of the radar beat signal under multiple cycles;

[0019] Step 8: Establish the radar antenna array receiving signal expression;

[0020] Step 9: Establish the two-dimensional data expression of ultra-high resolution millimeter-wave radar echo signal about time-antenna;

[0021] Step 10: Establish the two-dimensional data expression of the distance-antenna of the radar echo signal;

[0022] Step 11: Establish the distance-angle two-dimensional data expression of the radar echo signal;

[0023] Step 12: The distance-angle two-dimensional spectrum data obtained in Step 11 is used as the ultra-high resolution millimeter-wave radar data to be processed, and is input into the following improved two-dimensional CLEAN algorithm for processing to achieve the purpose of sidelobe suppression of the two-dimensional data;

[0024] Step 2: Modified 2D CLEAN algorithm based on local search

[0025] Step 1: Input distance-angle two-dimensional spectrum data and determine the data to be processed;

[0026] Step 2: Determine the top K highest peaks and their corresponding parameter information;

[0027] Step 3: Locally search the neighborhood of the Kth peak to determine the candidate target area;

[0028] Step 4: Determine K groups of peak clusters; repeat the operations in Step 2-3 until all the data of K peaks and their neighboring peaks are found, thereby constructing K groups of peak clusters;

[0029] Step 5: Determine the candidate target parameter matrix; take the K groups of peak clusters found in Step 4 as nodes, sequentially combine all peaks in each peak cluster to construct a dendrogram (the depth of the dendrogram is K), and determine the target parameter information of each branch in the dendrogram, so as to construct a complete candidate target parameter matrix;

[0030] Step 6: Construct the reference two-dimensional spectrum data of the Kth peak;

[0031] Step 7: Perform a subtraction operation; subtract the reference distance-angle two-dimensional data in Step 6 from the current distance-angle two-dimensional data. This step can eliminate the influence of the side lobes of the target corresponding to the peak in the current distance-angle two-dimensional spectrum:

[0032] Step 8: Repeat Step 6-7 to obtain U sets of residual distance-angle two-dimensional spectrum data; each residual data is obtained by subtracting the reference data of K candidate targets from the current data at one time;

[0033] Step9: Calculate the information entropy of each group of data in Step8;

[0034] Step 10: Determine the optimal estimated target parameter sequence; compare U groups of information entropies, find the corresponding target parameter sequence that reduces the overall information entropy, and select a set of parameter information (corresponding to K values) corresponding to the minimum information entropy as the estimated target for this iteration;

[0035] Step 11: Repeat Step 2-10 until all targets in all two-dimensional data are detected;

[0036] Step 12: Present the final distance-angle two-dimensional spectrum.

[0037] Preferably, the specific sub-steps of each step in step 1 are:

[0038] Step 1: Under a single linear frequency modulation transmission pulse, assuming that the radar transmitter transmits a linear frequency modulation signal, its mathematical expression is:

[0039]

[0040] Where f0 represents the initial frequency of the Chirp signal; K represents the frequency modulation slope of the Chirp signal; φ0 represents the initial phase of the Chirp signal; t represents the time series, t∈[0,T P ], T P Indicates pulse width; the subscript 't' is the abbreviation of 'transmit';

[0041] Step 2: Assuming that the radial distance between a moving target and the radar at the initial moment is R0 (m), and the target moves away from the radar at a constant radial speed v (m / s), the mathematical expression of the echo delay is:

[0042]

[0043] Where c represents the speed of light;

[0044] Step 3: According to the working principle of radar, the mathematical expression of the echo signal of the target received by the radar is:

[0045]

[0046] in, It represents the additional phase caused by the target reflection; the subscript 'r' is the abbreviation of 'receive';

[0047] Step 4: The echo signal represented by equation (1-3) and the transmission signal represented by equation (1-1) are sent to the mixer for mixing and passed through a low-pass filter. The output result is the beat signal of the echo signal, which is mathematically expressed as:

[0048]

[0049] Among them, (g) * Indicates the conjugation operation of data; the subscript 'I' is the abbreviation of 'Intermediate';

[0050] Step 5: Substitute equation (1-2) into equation (1-4) and ignore τ 2 Item, perform relevant mathematical processing, and obtain its simplified mathematical expression:

[0051]

[0052] Step 6: Consider multi-cycle Chirp signal analysis, R0+nvT p Substituting into equation (1-5) and replacing R0, the mathematical expression of the radar echo beat signal under the nth Chirp pulse is:

[0053]

[0054] Where n represents the Chirp pulse index number, n∈[0,N], and N represents the total number of pulses;

[0055] Step 7: Apply equation (1-6) with sampling period T s Sampling is performed to obtain the mathematical expression of the sampled beat signal:

[0056]

[0057] Where, m∈[0,M-1], M is the number of sampling points in each pulse period; n∈[0,N-1];

[0058] Step 8: Considering that multiple antenna array elements receive radar echo signals, the signal received by the pth antenna is mathematically expressed as:

[0059]

[0060] Where d is the distance between antenna elements, which is usually set to half the wavelength to avoid the appearance of grating lobes; θ0 is the angle of the target; λ is the wavelength; m∈[0,M-1]; n∈[0,N-1]; p is the antenna index number, p∈[0,P-1], and P is the total number of antennas;

[0061] Step 9: Temporarily ignore the relevant information about speed in equation (1-8), then the mathematical expression of the two-dimensional signal about time-antenna in the target echo signal received by the radar is:

[0062]

[0063] Among them, m∈[0,M-1], p∈[0,P-1];

[0064] Step 10: Perform Fourier transform operation on the time-antenna two-dimensional signal in equation (1-9) in the time direction to obtain the mathematical expression of the distance-antenna data of the radar echo signal:

[0065]

[0066] Wherein, FFT{g} denotes a fast Fourier transform operation; the subscript r denotes the abbreviation of distance ‘range’;

[0067] Step 11: Perform beamforming transformation operation on the distance-antenna two-dimensional signal in equation (1-10) with respect to the antenna direction to obtain the mathematical expression of the distance-angle data of the radar echo signal:

[0068]

[0069] Wherein, BF{g} represents the beamforming transformation operation; the subscript a is the abbreviation of azimuth.

[0070] Preferably, the specific sub-steps of each step in step 2 are:

[0071] Step 1: The distance-angle two-dimensional spectrum data F of the ultra-high resolution millimeter-wave radar echo signal obtained by formula (1-11) r,a (l,θ) is used as the input data of the improved two-dimensional CLEAN algorithm. The two-dimensional radar spectrum data contains all relevant information about the target of interest;

[0072] Step 2: Search for ultra-high resolution millimeter-wave radar 2D data | F r,a (l,θ)|, find the first K values ​​with the largest amplitude, and record their amplitude estimates and its corresponding distance estimate and the angle estimate

[0073]

[0074] Wherein, k=1,2,…,K represents the kth peak value; |g| represents the modulo operation of the data; argmax{g} represents the corresponding parameter value operation when the data takes the maximum value;

[0075] Step 3: Take the kth peak determined in Step 2 as the search center, search and determine the I neighboring peaks in its surrounding area, and record their corresponding amplitude estimates Distance estimate and the angle estimate

[0076]

[0077] Where i=1,2,…,I represents the i-th neighborhood peak; neighbor{g} represents its neighborhood search operation when the data takes the largest k-th peak;

[0078] Step 5: Take the K groups of peak clusters found in Step 4 as nodes, and sequentially combine all the peaks in each peak cluster to construct a dendrogram with a depth of K. At the same time, determine the target parameter information of each branch in the dendrogram, so as to construct a complete candidate target parameter matrix with a dimension of U×K (U=I K ), in which each row corresponds to the K peaks subtracted at one time by the two-dimensional CLEAN algorithm;

[0079] Step 6: Select the candidate target parameter information in the uth row and kth column in the candidate target parameter matrix (u = 1, 2, ..., U), and use the target amplitude estimation value, distance estimation value and angle estimation value to construct a reference distance-angle two-dimensional spectrum in the form of formula (1-11):

[0080]

[0081] in, represents the target amplitude estimate corresponding to the u-th row and k-th column in the candidate target parameter matrix; represents the target distance estimate corresponding to the uth row and kth column in the candidate target parameter matrix; is the target angle estimate corresponding to the uth row and kth column in the candidate target parameter matrix;

[0082] Step 7: Subtract the reference distance-angle two-dimensional data in Step 6 from the current distance-angle two-dimensional data. This step can eliminate the influence of the side lobe of the target corresponding to the peak in the current distance-angle two-dimensional spectrum:

[0083]

[0084] Where γ represents the loop gain factor;

[0085] Step 9: This information entropy uses the power spectrum information entropy, which can reflect the uncertainty of signal energy under power spectrum division:

[0086]

[0087] Among them, pro v is the proportion of the vth power spectrum component in the entire power spectrum.

[0088] Compared with the prior art, the present invention has the following beneficial effects:

[0089] The ultra-high resolution millimeter wave radar sidelobe suppression based on the modified two-dimensional CLEAN algorithm based on local neighborhood search proposed in the present invention can greatly improve the target detection rate of weak targets in two-dimensional scenes on the basis of meeting the accuracy requirements. At the same time, it can still achieve accurate target detection performance under complex problems such as high sidelobe level, false peaks and peak position offset. BRIEF DESCRIPTION OF THE DRAWINGS

[0090] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, constitute a part of this application, and do not constitute a limitation of the embodiments of the present invention. In the drawings:

[0091] Figure 1 The present invention is a flow chart of two-dimensional data modeling of ultra-high resolution millimeter wave radar echo signals.

[0092] Figure 2 This is a processing flow chart of the modified two-dimensional CLEAN algorithm based on local search of the present invention.

[0093] Figure 3 The present invention is a tree diagram construction structure in the modified two-dimensional CLEAN algorithm based on local search.

[0094] Figure 4 This is a comparison chart of the two-dimensional spectra before and after the two-dimensional CLEAN algorithm is processed in the target covering problem experiment of the present invention.

[0095] Figure 5 This is a comparison diagram of the two-dimensional spectra before and after processing by the two-dimensional CLEAN algorithm in the sidelobe suppression experiment of the continuous target of the present invention.

[0096] Figure 6 This is a comparison diagram of the sidelobe suppression experiment of the present invention before and after the two-dimensional CLEAN algorithm is processed. DETAILED DESCRIPTION

[0097] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation mode, structure, characteristics and effects of the present invention are described in detail below in combination with the accompanying drawings and preferred embodiments.

[0098] A method for suppressing the output sidelobe of an ultra-high resolution millimeter wave radar is provided. The first method is to perform a two-dimensional mathematical modeling on the process of "transmitting-targeting-receiving" of the ultra-high resolution millimeter wave radar detecting a target, and process the received millimeter wave radar echo signal, including performing Fourier transform in the time direction to obtain data on the distance dimension, and performing beamforming transform in the antenna direction to obtain data on the angle dimension, thereby obtaining the ultra-high resolution millimeter wave radar's distance-angle two-dimensional spectrum data on the target. The second method is to perform a two-dimensional CLEAN algorithm process to correct the distance-angle two-dimensional spectrum data, and construct a candidate target parameter matrix by searching the local neighborhood of the first K maximum peaks, and then determine the true target position, thereby reducing the sidelobe level of the distance-angle two-dimensional spectrum. The reduction in the sidelobe level can reveal weak targets in a two-dimensional scene, especially when the sidelobe of a strong amplitude target covers up a weak amplitude target. The specific process of the two parts of the algorithm is described in detail below.

[0099] 1. Two-dimensional data modeling of ultra-high resolution millimeter-wave radar echo signals, see Figure 1 ,

[0100] Step 1: Establish the radar transmission signal expression. Consider the signal of a single linear frequency modulation transmission pulse. Under a single Chirp pulse, assuming that the radar transmitter transmits a linear frequency modulation (LFM) signal, its mathematical expression is:

[0101]

[0102] Where f0 represents the initial frequency of the Chirp signal; K represents the frequency modulation slope of the Chirp signal; φ0 represents the initial phase of the Chirp signal; t represents the time series, t∈[0,T P ], T P Indicates pulse width; the subscript 't' is the abbreviation of 'transmit'.

[0103] Step 2: Establish the target echo delay expression. Assuming that the radial distance between a moving target and the radar at the initial moment is R0 (m), and the target moves away from the radar at a constant radial speed v (m / s), the echo delay can be mathematically expressed as:

[0104]

[0105] Here, c represents the speed of light.

[0106] Step 3: Establish the radar echo signal expression. The path attenuation of the echo signal is not considered for now. According to the working principle of radar (the echo signal is the delay of the transmitted signal), the echo signal of the target received by the radar can be mathematically expressed as:

[0107]

[0108] in, Represents the additional phase caused by target reflection; the subscript 'r' is the abbreviation of 'receive'.

[0109] Step 4: Establish the radar beat signal expression under single cycle. The echo signal represented by equation (1-3) and the transmission signal represented by equation (1-1) are sent to the mixer for mixing and passed through a low-pass filter. The output result is the beat signal of the echo signal, which can be mathematically expressed as:

[0110]

[0111] Among them, (g) * Indicates the conjugation operation of data; the subscript 'I' is the abbreviation of 'Intermediate'.

[0112] Step 5: Establish a simplified radar beat signal expression. Substitute equation (1-2) into equation (1-4) and ignore τ 2 Item (due to c -2 The term is very small), and through relevant mathematical processing, its simplified mathematical expression can be obtained:

[0113]

[0114] Step 6: Establish radar beat signal expression under multi-cycle. Consider multi-cycle Chirp signal analysis, R0+nvT p Substituting into equation (1-5) and replacing R0, the mathematical expression of the radar echo beat signal under the nth Chirp pulse is:

[0115]

[0116] Where n represents the Chirp pulse index number, n∈[0,N], and N represents the total number of pulses.

[0117] Step 7: Establish a discrete expression of the radar beat signal under multiple cycles. s After sampling, the mathematical expression of the sampled beat signal can be obtained:

[0118]

[0119] Among them, m∈[0,M-1], M is the number of sampling points in each pulse period; n∈[0,N-1].

[0120] Step 8: Establish the expression of the received signal of the radar antenna array. Considering that multiple antenna array elements receive radar echo signals, the signal received by the pth antenna can be mathematically expressed as:

[0121]

[0122] Where d is the distance between antenna elements, which is usually set to half a wavelength to avoid the appearance of grating lobes; θ0 is the angle of the target; λ is the wavelength; m∈[0,M-1]; n∈[0,N-1]; p is the antenna index number, p∈[0,P-1], and P is the total number of antennas.

[0123] Step 9: Establish a two-dimensional data expression of the ultra-high resolution millimeter-wave radar echo signal about time-antenna. Temporarily ignoring the relevant information about speed in equation (1-8), the two-dimensional signal about time-antenna in the target echo signal received by the radar can be mathematically expressed as:

[0124]

[0125] Among them, m∈[0,M-1], p∈[0,P-1].

[0126] Step 10: Establish a two-dimensional data expression of the distance-antenna of the radar echo signal. Perform a Fourier transform operation on the time-antenna two-dimensional signal in equation (1-9) in the time direction to obtain the mathematical expression of the distance-antenna data of the radar echo signal:

[0127]

[0128] Where FFT{g} denotes a fast Fourier transform operation and the subscript r is an abbreviation for range.

[0129] Step 11: Establish the distance-angle two-dimensional spectrum expression of the radar echo signal. Perform beamforming transformation operation on the distance-antenna two-dimensional signal in equation (1-10) with respect to the antenna direction, and the mathematical expression of the distance-angle data of the radar echo signal can be obtained:

[0130]

[0131] Wherein, BF{g} represents the beamforming transformation operation; the subscript a is the abbreviation of azimuth.

[0132] Step 12: The distance-angle two-dimensional spectrum data obtained by formula (1-11) is used as the ultra-high resolution millimeter-wave radar data to be processed, and is input into the following improved two-dimensional CLEAN algorithm for processing to achieve the purpose of sidelobe suppression of the two-dimensional data.

[0133] 2. Modified 2D CLEAN algorithm based on local search, see Figure 2 ,

[0134] Step 1: Determine the data to be processed. The distance-angle two-dimensional spectrum data F of the ultra-high resolution millimeter-wave radar echo signal obtained by formula (1-11) r,a (l,θ) is used as the input data of the improved two-dimensional CLEAN algorithm. The two-dimensional radar spectrum data contains all relevant information about the target of interest, including the amplitude, distance, angle, etc. of the target.

[0135] Step 2: Determine the top K highest peaks and their corresponding parameter information. Searching for ultra-high resolution millimeter-wave radar two-dimensional data | F r,a (l,θ)|, find the first K values ​​with the largest amplitude, and record their amplitude estimates and its corresponding distance estimate and the angle estimate

[0136]

[0137] Among them, k=1,2,…,K represents the kth peak; |g| represents the modulo operation of the data; argmax{g} represents the corresponding parameter value operation when the data takes the maximum value.

[0138] Step 3: Locally search the peak neighborhood to determine the candidate target area. Take the kth peak determined in Step 2 as the search center, search and determine the I neighboring peaks in its surrounding area, and record their corresponding amplitude estimates. Distance estimate and the angle estimate

[0139]

[0140] Among them, i=1,2,…,I represents the i-th neighborhood peak; neighbor{g} represents its neighborhood search operation when the data takes the largest k-th peak.

[0141] Step 4: Determine K groups of peak clusters. Repeat the operations in Step 2-3 until all the data of K peaks and their neighboring peaks are found, thus constructing K groups of peak clusters.

[0142] Step 5: Determine the candidate target parameter matrix. Take the K groups of peak clusters found in Step 4 as nodes, combine all the peaks in each peak cluster in turn to build a dendrogram (the depth of the dendrogram is K), and determine the target parameter information of each branch in the dendrogram, so as to build a complete candidate target parameter matrix with a dimension of U×K (U=I K ), in the candidate matrix, each row corresponds to the K peaks subtracted at one time by the two-dimensional CLEAN algorithm. The dendrogram is constructed as follows Figure 3 shown.

[0143] Step 6: Construct reference two-dimensional spectrum data. Select the candidate target parameter information in the uth row and the kth column in the candidate target parameter matrix (u=1,2,…,U), and use the target amplitude estimation value, distance estimation value and angle estimation value to construct the reference distance-angle two-dimensional spectrum in the form of formula (1-11):

[0144]

[0145] in, represents the target amplitude estimate corresponding to the u-th row and k-th column in the candidate target parameter matrix; represents the target distance estimate corresponding to the uth row and kth column in the candidate target parameter matrix; is the target angle estimate corresponding to the u-th row and k-th column in the candidate target parameter matrix.

[0146] Step 7: Perform a subtraction operation. Subtract the reference distance-angle two-dimensional data in Step 6 from the current distance-angle two-dimensional data. This step can eliminate the influence of the side lobes of the target corresponding to the peak in the current distance-angle two-dimensional spectrum:

[0147]

[0148] Where γ represents the loop gain factor.

[0149] Step 8: Repeat Step 6-7 to obtain U groups of residual distance-angle two-dimensional data. Each residual data is obtained by subtracting the reference data of K candidate targets from the current data at one time;

[0150] Step 9: Calculate the information entropy of each group of data in Step 8. This information entropy uses the power spectrum information entropy, which can reflect the uncertainty of signal energy under power spectrum division:

[0151]

[0152] Among them, pro v is the proportion of the vth power spectrum component in the entire power spectrum.

[0153] Step 10: Determine the optimal estimated target parameter sequence. Compare U groups of information entropies, find the target parameter sequence that reduces the overall information entropy, and select a set of parameter information (corresponding to K values) corresponding to the minimum information entropy as the estimated target for this iteration.

[0154] Step 11: Repeat Step 2-10 until all targets in all two-dimensional data are detected.

[0155] Step 12: Present the final distance-angle two-dimensional spectrum.

[0156] Experimental data:

[0157] 1. Sidelobe suppression experiment of simulated data

[0158] In this experiment, the simulated data is mainly used as the research object.

[0159] 1) Experiment on target occlusion problem

[0160] In the data used for simulation, four targets with large amplitude differences are used as research objects (the amplitudes of the two large targets are much larger than the amplitudes of the two small targets, so as to achieve the effect of strong targets covering weak targets). After the two-dimensional data of the millimeter-wave radar echo signal is processed by modeling program, the two-dimensional distance-angle data of the target is obtained, which is used as the input data of the two-dimensional CLEAN algorithm. The comparison results before and after the two-dimensional CLEAN algorithm are shown in Figure 2. Figure 4 shown.

[0161] From the experimental results, before the two-dimensional CLEAN algorithm was processed, the two strong targets with larger amplitudes in the scene were observable. However, in the two-dimensional spectrum of distance-angle, the side lobes of the two targets were too high (about -3.5dB), resulting in the target with smaller amplitude being completely submerged by the side lobes of the target with larger amplitude, causing the weak amplitude target to completely disappear in the two-dimensional spectrum, seriously affecting the detection and recognition performance of the target. After the two-dimensional CLEAN algorithm was processed, the side lobe level of the target with larger amplitude was greatly reduced (approximately from -3.5dB to -59.29dB), thereby exposing the target with smaller amplitude, making all targets in the scene observable, greatly improving the target detection performance.

[0162] 2) Sidelobe suppression experiment of continuous target

[0163] In the data used for simulation, four continuous point target models are used as the research object (the point targets are separated by their own resolutions in the distance and angle directions). After the two-dimensional data of the millimeter-wave radar echo signal is processed by modeling program, the distance-angle two-dimensional data of the target is obtained and used as the input data of the two-dimensional CLEAN algorithm. The comparison results before and after the two-dimensional CLEAN algorithm are shown in Figure 2. Figure 5 shown.

[0164] From the experimental results, before the 2D CLEAN algorithm was used, the positions of the various points of the continuous target in the scene were offset to a certain extent, and the sidelobe level in the 2D spectrum was high, about -10.22dB, which seriously affected the detection and recognition performance of the target. After the 2D CLEAN algorithm was used, the various points of the continuous target were fully restored. More importantly, the sidelobe level in the 2D spectrum was reduced from -10.22dB before processing to -37.2dB after processing, which greatly improved the target detection performance.

[0165] 2. Sidelobe suppression experiment of measured data

[0166] In the experimental part of the measured data, the school bus was selected as the research object. The actual radar data was collected in the test environment such as Figure 3 As shown, the millimeter-wave radar board is placed about 5 meters in front of the campus bus, the millimeter-wave radar board and the various interfaces of the laptop are connected, the radar collection application is started, the parameters are adjusted, and radar data is collected for the current scene.

[0167] like Figure 6 As shown in the figure, before the two-dimensional CLEAN algorithm is processed, although the shape of the bus can be observed in the figure, the overall sidelobe level is relatively high, about -33dB. If there are small targets with weak amplitudes in the surrounding area, the sidelobe level will obviously cover up the small targets, seriously hindering the detection and recognition of all targets in the scene. After the two-dimensional CLEAN algorithm is processed, it can be clearly observed that the shape of the bus in the figure is clearly highlighted. More importantly, the sidelobe level of the surrounding area has been greatly reduced to about -85dB. At this sidelobe level, small targets with weak amplitudes will be clearly exposed, which can greatly improve the detection performance of the radar. It can be clearly observed from the before-and-after comparison effect diagram of the two that the two-dimensional CLEAN algorithm is feasible under the measured millimeter-wave radar data, and the effect of the two-dimensional CLEAN algorithm in suppressing the sidelobe level is significant, which greatly proves the effectiveness of the two-dimensional CLEAN algorithm.

[0168] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technical personnel in this field can make some changes or modify the technical contents disclosed above into equivalent embodiments without departing from the scope of the technical solution of the present invention. However, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. A method for suppressing output sidelobe of ultra-high resolution millimeter-wave radar, characterized in that: The following steps are involved: Step 1: 2D data modeling of ultra-high resolution millimeter-wave radar echo signals Step 1: Establish radar transmission signal expression; Step 2: Establish the target echo delay expression; Step 3: Establish radar echo signal expression; Step 4: Establish radar beat signal expression under single cycle; Step 5: Establish a simplified radar beat signal expression; Step 6: Establish radar beat signal expression under multi-cycle; Step 7: Establish a discrete expression of the radar beat signal under multiple cycles; Step 8: Establish the radar antenna array receiving signal expression; Step 9: Establish the two-dimensional data expression of ultra-high resolution millimeter-wave radar echo signal about time-antenna; Step 10: Establish the two-dimensional data expression of the distance-antenna of the radar echo signal; Step 11: Establish the distance-angle two-dimensional data expression of the radar echo signal; Step 12: The distance-angle two-dimensional spectrum data obtained in Step 11 is used as the ultra-high resolution millimeter-wave radar data to be processed, and is input into the following improved two-dimensional CLEAN algorithm for processing to achieve the purpose of sidelobe suppression of the two-dimensional data; Step 2: Modified 2D CLEAN algorithm based on local search Step 1: Input distance-angle two-dimensional spectrum data and determine the data to be processed; Step 2: Determine the top K highest peaks and their corresponding parameter information; Step 3: Locally search the neighborhood of the Kth peak to determine the candidate target area; Step 4: Determine K groups of peak clusters; repeat the operations in Step 2-3 until all the data of K peaks and their neighboring peaks are found, thereby constructing K groups of peak clusters; Step 5: Determine the candidate target parameter matrix; take the K groups of peak clusters found in Step 4 as nodes, sequentially combine all peaks in each peak cluster to construct a dendrogram (the depth of the dendrogram is K), and determine the target parameter information of each branch in the dendrogram, so as to construct a complete candidate target parameter matrix; Step 6: Construct the reference two-dimensional spectrum data of the Kth peak; Step 7: Perform a subtraction operation; subtract the reference distance-angle two-dimensional data in Step 6 from the current distance-angle two-dimensional data. This step can eliminate the influence of the side lobes of the target corresponding to the peak in the current distance-angle two-dimensional spectrum: Step 8: Repeat Step 6-7 to obtain U sets of residual distance-angle two-dimensional spectrum data; each residual data is obtained by subtracting the reference data of K candidate targets from the current data at one time; Step9: Calculate the information entropy of each group of data in Step8; Step 10: Determine the optimal estimated target parameter sequence; compare U groups of information entropies, find the corresponding target parameter sequence that reduces the overall information entropy, and select a set of parameter information (corresponding to K values) corresponding to the minimum information entropy as the estimated target for this iteration; Step 11: Repeat Step 2-10 until all targets in all two-dimensional data are detected; Step 12: Present the final distance-angle two-dimensional spectrum.

2. The method for suppressing output sidelobe of an ultra-high resolution millimeter-wave radar according to claim 1, characterized in that: The specific sub-steps of each step in step 1 are: Step 1: Under a single linear frequency modulation transmission pulse, assuming that the radar transmitter transmits a linear frequency modulation signal, its mathematical expression is: Where f0 represents the initial frequency of the Chirp signal; K represents the frequency modulation slope of the Chirp signal; φ0 represents the initial phase of the Chirp signal; t represents the time series, t∈[0,T P ], T P Indicates pulse width; the subscript 't' is the abbreviation of 'transmit'; Step 2: Assuming that the radial distance between a moving target and the radar at the initial moment is R0 (m), and the target moves away from the radar at a constant radial speed v (m / s), the mathematical expression of the echo delay is: Where c represents the speed of light; Step 3: According to the working principle of radar, the mathematical expression of the echo signal of the target received by the radar is: in, It represents the additional phase caused by the target reflection; the subscript 'r' is the abbreviation of 'receive'; Step 4: The echo signal represented by equation (1-3) and the transmission signal represented by equation (1-1) are sent to the mixer for mixing and passed through a low-pass filter. The output result is the beat signal of the echo signal, which is mathematically expressed as: Among them, (g) * Indicates the conjugation operation of data; the subscript 'I' is the abbreviation of 'Intermediate'; Step 5: Substitute equation (1-2) into equation (1-4) and ignore τ 2 Item, perform relevant mathematical processing, and obtain its simplified mathematical expression: Step 6: Consider multi-cycle Chirp signal analysis, R0+nvT p Substituting into equation (1-5) and replacing R0, the mathematical expression of the radar echo beat signal under the nth Chirp pulse is: Where n represents the Chirp pulse index number, n∈[0,N], and N represents the total number of pulses; Step 7: Apply equation (1-6) with sampling period T s Sampling is performed to obtain the mathematical expression of the sampled beat signal: Where, m∈[0,M-1], M is the number of sampling points in each pulse period; n∈[0,N-1]; Step 8: Considering that multiple antenna array elements receive radar echo signals, the signal received by the pth antenna is mathematically expressed as: Where d is the distance between antenna elements, which is usually set to half the wavelength to avoid the appearance of grating lobes; θ0 is the angle of the target; λ is the wavelength; m∈[0,M-1]; n∈[0,N-1]; p is the antenna index number, p∈[0,P-1], and P is the total number of antennas; Step 9: Temporarily ignore the relevant information about speed in equation (1-8), then the mathematical expression of the two-dimensional signal about time-antenna in the target echo signal received by the radar is: Among them, m∈[0,M-1], p∈[0,P-1]; Step 10: Perform Fourier transform operation on the time-antenna two-dimensional signal in equation (1-9) in the time direction to obtain the mathematical expression of the distance-antenna data of the radar echo signal: Wherein, FFT{g} denotes a fast Fourier transform operation; the subscript r denotes the abbreviation of distance ‘range’; Step 11: Perform beamforming transformation operation on the distance-antenna two-dimensional signal in equation (1-10) with respect to the antenna direction to obtain the mathematical expression of the distance-angle data of the radar echo signal: Wherein, BF{g} represents the beamforming transformation operation; the subscript a is the abbreviation of azimuth.

3. The method for suppressing output sidelobe of an ultra-high resolution millimeter-wave radar according to claim 2, characterized in that: The specific sub-steps of each step in step 2 are: Step 1: The distance-angle two-dimensional spectrum data F of the ultra-high resolution millimeter-wave radar echo signal obtained by formula (1-11) r,a (l,θ) is used as the input data of the improved two-dimensional CLEAN algorithm. The two-dimensional radar spectrum data contains all relevant information about the target of interest; Step 2: Search for ultra-high resolution millimeter-wave radar 2D data | F r,a (l,θ)|, find the first K values ​​with the largest amplitude, and record their amplitude estimates and its corresponding distance estimate and the angle estimate Wherein, k=1,2,…,K represents the kth peak value; |g| represents the modulo operation of the data; argmax{g} represents the corresponding parameter value operation when the data takes the maximum value; Step 3: Take the kth peak determined in Step 2 as the search center, search and determine the I neighboring peaks in its surrounding area, and record their corresponding amplitude estimates Distance estimate and the angle estimate Where i=1,2,…,I represents the i-th neighborhood peak; neighbor{g} represents the neighborhood search operation when the data takes the largest k-th peak; Step 5: Take the K groups of peak clusters found in Step 4 as nodes, and sequentially combine all the peaks in each peak cluster to construct a dendrogram with a depth of K. At the same time, determine the target parameter information of each branch in the dendrogram, so as to construct a complete candidate target parameter matrix with a dimension of U×K (U=I K ), in which each row corresponds to the K peaks subtracted at one time by the two-dimensional CLEAN algorithm; Step 6: Select the candidate target parameter information in the uth row and kth column in the candidate target parameter matrix (u = 1, 2, ..., U), and use the target amplitude estimation value, distance estimation value and angle estimation value to construct a reference distance-angle two-dimensional spectrum in the form of formula (1-11): in, represents the target amplitude estimate corresponding to the u-th row and k-th column in the candidate target parameter matrix; represents the target distance estimate corresponding to the uth row and kth column in the candidate target parameter matrix; is the target angle estimate corresponding to the uth row and kth column in the candidate target parameter matrix; Step 7: Subtract the reference distance-angle two-dimensional data in Step 6 from the current distance-angle two-dimensional data. This step can eliminate the influence of the side lobe of the target corresponding to the peak in the current distance-angle two-dimensional spectrum: Where γ represents the loop gain factor; Step 9: This information entropy uses the power spectrum information entropy, which can reflect the uncertainty of signal energy under power spectrum division: Among them, pro v is the proportion of the vth power spectrum component in the entire power spectrum.

Images