Processing method and system for improving target positioning robustness based on smoothing filtering

Abstract

The invention discloses a processing method and system for improving target positioning robustness based on smooth filtering, and the method comprises the steps: generating and obtaining at least two pieces of first data corresponding to a target object to be positioned by an FMCW radar through the method, and generating corresponding second data according to the first data; based on the second data, generating and acquiring eighth data corresponding to the FMCW radar, and generating ninth data corresponding to the target object according to the eighth data; according to the second data and the eighth data, through two-dimensional Fourier transform processing, sparse MUSIC algorithm processing and smooth filtering processing in sequence, corresponding third data with high robustness are generated; the invention further discloses a system corresponding to the method so as to realize super-resolution positioning of the target objects and improve the positioning precision of the multiple target objects to the maximum extent. Moreover, the target positioning effect in a strong noise environment can be improved, the time complexity and the space complexity are reduced, and the robustness of target object positioning is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of multi-target FMCW radar positioning processing, and particularly relates to a method and system for improving the robustness of target positioning based on smoothing filtering. Background Art

[0002] Currently, the increasing application scenarios have continuously increased the demand for radar positioning. There are technical defect problems such as poor accuracy in positioning target objects, high time complexity and space complexity, and low robustness. Especially in a strong noise environment, the accuracy of multiple target objects will be worse, the time complexity and space complexity will be higher, and the robustness will be lower.

[0003] That is to say, during the process of positioning a target object by radar, it is impossible to achieve noise-free simulation radar positioning, and it is impossible to achieve noise reduction and precise positioning of the positions of multiple objects; moreover, the time complexity and space complexity of positioning the target object are high, and the robustness is low.

[0004] Therefore, aiming at the above technical problems and defects that it is impossible to achieve noise-free simulation radar positioning, it is impossible to achieve noise reduction and precise positioning of the positions of multiple objects, and the time complexity and space complexity of positioning the target object are high and the robustness is low, it is urgently necessary to design and develop a method and system for improving the robustness of target positioning based on smoothing filtering. Summary of the Invention

[0005] To overcome the deficiencies and difficulties of the above-mentioned prior art, the purpose of the present invention is to provide a method and system for improving the robustness of target positioning based on smoothing filtering, so as to achieve super-resolution positioning to the target object, and greatly improve the positioning accuracy of multiple target objects, reduce the time complexity and space complexity, and improve the robustness of positioning the target object.

[0006] The first object of the present invention is to provide a method for improving the robustness of target positioning based on smoothing filtering; the second object of the present invention is to provide a system for improving the robustness of target positioning based on smoothing filtering.

[0007] The first object of the present invention is achieved as follows: The method includes:

[0008] Generating and obtaining at least two pieces of first data corresponding to the target object to be positioned by the FMCW radar, and generating corresponding second data according to the first data; wherein, the first data is the reflected signal data of the target object; the second data includes the target object distance data and the target object azimuth data;

[0009] Generate and obtain the eighth data corresponding to the FMCW radar based on the second data, and generate the ninth data corresponding to the target object according to the eighth data; wherein, the eighth data is the positioning deviation data of the antenna; the ninth data is the offset data corresponding to the target object and the eighth data;

[0010] According to the second data and in combination with the eighth data, perform two-dimensional Fourier transform processing, sparse MUSIC algorithm processing, and smoothing filter processing in sequence to generate the corresponding third data with high robustness; wherein, the third data is the target object position positioning data.

[0011] The second object of the present invention is achieved as follows: The system is applied to the method for improving the robustness of target positioning based on smoothing filtering. The system includes: a first data generation unit for generating and obtaining at least two first data corresponding to the target object to be positioned by the FMCW radar, and generating the corresponding second data according to the first data; wherein, the first data is the reflection signal data of the target object; the second data includes the target object distance data and the target object azimuth data;

[0012] A second data generation unit for generating and obtaining the eighth data corresponding to the FMCW radar based on the second data, and generating the ninth data corresponding to the target object according to the eighth data; wherein, the eighth data is the positioning deviation data of the antenna; the ninth data is the offset data corresponding to the target object and the eighth data;

[0013] A data processing and generation unit for generating the corresponding third data with high robustness through two-dimensional Fourier transform processing, sparse MUSIC algorithm processing, and smoothing filter processing in sequence according to the second data and in combination with the eighth data; wherein, the third data is the target object position positioning data.

[0014] The present invention generates and obtains at least two first data corresponding to the target object to be located by an FMCW radar, and generates corresponding second data according to the first data; wherein, the first data is the reflected signal data of the target object; the second data includes the target object distance data and the target object azimuth data; based on the second data, eighth data corresponding to the FMCW radar is generated and obtained, and ninth data corresponding to the target object is generated according to the eighth data; wherein, the eighth data is the positioning deviation data of the antenna; the ninth data is the offset data of the target object corresponding to the eighth data; according to the second data, and in combination with the eighth data, two-dimensional Fourier transform processing, sparse MUSIC algorithm processing, and smoothing filtering processing are sequentially performed to generate corresponding third data with high robustness; wherein, the third data is the target object position positioning data, and a system corresponding to the method is used to achieve super-resolution positioning of the target object, and greatly improve the positioning accuracy of multiple target objects; and, the target positioning effect in a strong noise environment can also be improved, and the time complexity and space complexity can be reduced, and the robustness of the target object positioning can be improved. Description of the Drawings

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0016] Figure 1 Schematic diagram of multiple IF single-tone signals transmitted for multiple object detections;

[0017] Figure 2 Schematic diagram of the original signal, the signal after fft transformation, and the spectrum after fft transformation of the 21st antenna;

[0018] Figure 3 Two-dimensional projection (left) and three-dimensional diagram (right) of the distance-angle distribution diagram of the signal intensity;

[0019] Figure 4 Schematic diagram of the position of the target point in the coordinate system;

[0020] Figure 5 Schematic diagram of one of the target positioning signal intensity (dB) distributions restored by the MUSIC algorithm ( Figure Ⅰ Three-dimensional distribution diagram of the signal intensity. The image on the xyz plane is the projection of the three-dimensional diagram, and the color represents the signal intensity; the black frame represents the area where the target of key concern is located; Figure Ⅱ Enlarged view of the area of key concern);

[0021] Figure 6 It is a schematic diagram of one of the positions of the target point in the coordinate system based on the MUSIC algorithm;

[0022] Figure 7 It is one of the result diagrams after compressive sensing sparsification in both the distance and angle directions (the result of angle-direction sparsification (left), the exact position of the target point (upper right), and the result of distance-direction sparsification (lower right));

[0023] Figure 8 It is one of the time-frequency diagrams (left) and spectrograms (right) of the signals received by the 1st (upper), 43rd (middle), and 85th (lower) antennas after wavelet transform (Gaussian 2 wavelet basis);

[0024] Figure 9 It is a schematic diagram of the distance between the left and right antennas of the radar and the target;

[0025] Figure 10 It is for |d right -d left | distribution schematic diagram;

[0026] Figure 11 It is the second diagram of the distribution of the target positioning signal intensity (dB) restored by the MUSIC algorithm; ((Ⅰ) The diagram is a three-dimensional distribution diagram of the signal intensity. The image on the xyz plane is the projection of this three-dimensional diagram, and the color represents the signal intensity; the black frame represents the area where the target of key concern is located. (Ⅱ) The right diagram is an enlarged view of the area of key concern);

[0027] Figure 12 It is a schematic diagram of the second position of the target point in the coordinate system based on the MUSIC algorithm;

[0028] Figure 13 It is the second diagram of the result after compressive sensing sparsification in both the distance and angle directions; (the result of angle-direction sparsification (left), the exact position of the target point (upper right), the result of distance-direction sparsification (lower right));

[0029] Figure 14 It is the second time-frequency diagram (left) and spectrogram (right) of the signals received by the 1st (upper), 43rd (middle), and 85th (lower) antennas after wavelet transform (Gaussian 2 wavelet basis);

[0030] Figure 15 It is a schematic diagram of the motion trajectory of an object within one frame (32 Chirp cycles);

[0031] Figure 16Schematic diagram of the relative motion trajectory of an object calculated for a provided frame of data (polar coordinate diagram of the target point (upper left), rectangular coordinate diagram (upper right, the black frame represents the enlarged area), enlarged area of the rectangular coordinate diagram (lower left), trajectory diagram of the target point (lower right, the straight line is the trajectory, and the arrow represents the trajectory direction));

[0032] Figure 17 Schematic diagram of the results within the 1st, 5th, 9th, 13th, 17th, 21st, 25th, and 29th chirp periods in one frame for wave analysis (wavelet basis: Gaussian 2);

[0033] Figure 18 Schematic diagram of the distance between the left and right antennas of the radar and the target;

[0034] Figure 19 The third diagram of the target localization signal intensity (dB) restored by the MUSIC algorithm; (The left diagram is the three-dimensional distribution diagram of the signal intensity, the image on the xyz plane is the projection of this three-dimensional diagram, and the color represents the signal intensity; the black frame represents the area where the target of key concern is located; the right diagram is the enlarged diagram of the area of key concern);

[0035] Figure 20 Schematic diagram of the results after smoothing filtering (filter window length: 7 in the angular direction and 3 in the distance direction);

[0036] Figure 21 Schematic diagram of the system architecture of a method for improving the robustness of target localization based on smoothing filtering according to the present invention

[0037] Figure 22 Schematic diagram of the process flow of a method for improving the robustness of target localization based on smoothing filtering according to the present invention Detailed implementation manners

[0038] For a better understanding of the purpose, technical solution, and advantages of the present invention, the present invention will be further described below in conjunction with the accompanying drawings and specific implementation manners. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification

[0039] The present invention can also be implemented or applied through other different specific examples, and various details in this specification can also be modified and changed based on different viewpoints and applications without departing from the spirit of the present invention

[0040] The present invention will be further described in detail below in conjunction with the accompanying drawings. As Figures 1 - 20 、 Figure 22 shown, the present invention provides a method for improving the robustness of target localization based on smoothing filtering, and the method includes the following steps:

[0041] S1. Generate and obtain at least two pieces of first data corresponding to the target object to be located by the FMCW radar, and generate corresponding second data according to the first data; wherein, the first data is the reflected signal data of the target object; the second data includes the target object distance data and the target object azimuth data;

[0042] S2. Based on the second data, generate and obtain the eighth data corresponding to the FMCW radar, and generate the ninth data corresponding to the target object according to the eighth data; wherein, the eighth data is the positioning deviation data of the antenna; the ninth data is the offset data corresponding to the target object and the eighth data;

[0043] S3. According to the second data, and in combination with the eighth data, sequentially perform two-dimensional Fourier transform processing, sparse MUSIC algorithm processing, and smoothing filter processing to generate corresponding third data with high robustness; wherein, the third data is the target object position positioning data.

[0044] The generating and obtaining at least two pieces of first data corresponding to the target object to be located by the FMCW radar, and generating corresponding second data according to the first data, further includes:

[0045] S11. Generate the fourth data corresponding to the target object to be located by the FMCW radar according to the first data; wherein, the fourth data is the double echo distance data;

[0046] S12. Based on the fourth data, determine and generate the fifth data corresponding to the target object; wherein, the fifth data is the coordinate data of the target object.

[0047] The generating and obtaining the eighth data corresponding to the FMCW radar based on the second data, and generating the ninth data corresponding to the target object according to the eighth data, further includes:

[0048] S21. Generate and obtain at least one frame of tenth data corresponding to the moving target object; wherein, the tenth data is the intermediate frequency signal data;

[0049] S22. Generate the eleventh data corresponding to the moving target object and with online low complexity according to the tenth data; wherein, the eleventh data is the target object motion trajectory data.

[0050] The generating the corresponding third data with high robustness by sequentially performing two-dimensional Fourier transform processing, sparse MUSIC algorithm processing, and smoothing filter processing according to the second data and in combination with the eighth data, further includes:

[0051] S31. Construct a distance-angle grid corresponding to the target object to be located by the FMCW radar;

[0052] S32. Generate conjugate signal data corresponding to the intermediate-frequency signal data of the target object based on the distance-angle grid;

[0053] S33. Generate sixth data corresponding to the target object to be located by the FMCW radar according to the intermediate-frequency signal data of the target object and the conjugate signal data; wherein, the sixth data is the target object position estimation data.

[0054] The step of generating corresponding and highly robust third data by performing two-dimensional Fourier transform processing, sparse MUSIC algorithm processing, and smoothing filtering processing in sequence according to the second data and in combination with the eighth data further includes:

[0055] S34. Generate a first matrix corresponding to the array signal, and perform eigenvalue decomposition on the first matrix; wherein, the first matrix is the autocorrelation matrix corresponding to the array received signal data;

[0056] S35. Generate a first subspace corresponding to the first matrix based on the first matrix; wherein, the first subspace is the noise subspace;

[0057] S36. Construct a corresponding spatial spectrum function according to the first subspace, and generate corresponding seventh data through spectrum peak search; wherein, the seventh data is the steering vector data for each angle in the spatial domain.

[0058] The step of generating corresponding and highly robust third data by performing two-dimensional Fourier transform processing, sparse MUSIC algorithm processing, and smoothing filtering processing in sequence according to the second data and in combination with the eighth data further includes:

[0059] S37. Respectively perform compressive sensing sparsification processing on the angle direction and distance direction corresponding to the target object to be located;

[0060] S38. Generate corresponding twelfth data according to the eighth data and in combination with the spectrum waveform corresponding to the radar antenna; wherein the twelfth data is the antenna damage state data, including single-antenna damage situation data and multi-antenna damage situation data.

[0061] Specifically, in the embodiments of the present invention, for the problem of super-resolution positioning in mobile scenarios, a method and system for improving the robustness of target positioning based on smoothing filtering are provided; by using the linearly frequency-modulated signal wave and the echo signal transmitted by a frequency-modulated continuous-wave (FMCW) radar, an improved mathematical model and corresponding algorithms are established to improve the resolution to meet the requirements of super-resolution positioning, and the following problems are solved:

[0062] (1) Noise-free simulation of radar positioning; wherein, given the intermediate-frequency signal received by the radar array, the target position is solved. The main solution idea is as follows: based on the one-to-one correspondence between the frequency of the intermediate-frequency signal received by FMCW and the distance, the frequency of the intermediate-frequency signal is calculated by Fourier transform to obtain the distance between the target and the antenna, and the MUSIC algorithm is used for verification. Finally, the compressive sensing algorithm is used to improve the resolution of the result.

[0063] (2) Design a super-resolution algorithm to achieve noise reduction and accurately locate the positions of multiple objects; wherein, given that the received intermediate-frequency signal is superimposed with Gaussian noise, a super-resolution algorithm is designed to accurately locate multiple objects. The main idea is as follows: by using the orthogonality of the signal subspace and the noise subspace through the MUSIC algorithm, a spatial spectrum function is constructed, and the position of the signal is estimated through spectrum peak search. Similarly, the resolution of the result is improved by compressive sensing to achieve target positioning.

[0064] (3) Based on one frame of data, calculate the relative motion trajectory of the object and visualize it in a two-dimensional graph; wherein, for a moving object, one frame of intermediate-frequency signal is used for super-resolution positioning to calculate the relative motion trajectory of the object, and at the same time, the algorithm is required to be online and of low complexity. The main idea is as follows: the row-by-row and column-by-column operations of the matrix that occupy a large number of for loops in the algorithm are improved to a form of direct operation on the entire matrix, reducing the time complexity.

[0065] (4) Design an improved algorithm to enhance the robustness of the positioning algorithm for the positioning error of the antenna array itself; wherein, the algorithm is improved for the problem of the positioning error of the antenna itself in order to improve the robustness of the positioning. The solution idea is as follows: the positioning deviation of the antenna is equivalent to the offset of the target object, so as to distinguish the damaged antenna according to the difference in the spectrum waveform and correct it, and the antenna damage situation is divided into single-antenna damage situation and multi-antenna damage situation for discussion.

[0066] In the solution of the present invention, the radius of the radar antenna is much smaller than the distance of the object. Since the time of one chirp period (about 50 microseconds) is extremely short, it can be considered that the object remains stationary within the period. However, within one frame time, the relative position of the object has an obvious movement. The reflection of the target object is isotropic, and the reflectivity of different objects is consistent. In each scenario, it is assumed that there are K objects within the detection range of the radar (a sector area with a radius of 10 meters centered at the origin and an opening angle of 100° upward). In a fast FMCW radar, the dependence of the frequency of the IF signal on the object velocity can be ignored. Equivalent virtual antenna array: After modulating the waveform of the electromagnetic wave, the NTX transmitting antennas transmit sequentially, and the NRX receiving antennas will also sequentially receive the returned signals. Since this period is extremely short, it can be equivalent to NTX×NRX antennas transmitting and receiving simultaneously. The number of targets is less than the number of antennas to ensure that the columns of the array manifold matrix in the MUSIC algorithm are linearly independent. The antenna signals and noise are independent of each other. Among them, in the solution and formulas, the meanings of the symbols involved are shown in the following table:

[0067] Table 1 Symbol Explanation of This Solution

[0068]

[0069]

[0070] Note: For symbols not listed and important symbols, the appearance location shall prevail.

[0071] According to the number of target objects, the positioning of noise-free simulation data is mainly divided into two types: single target and multiple targets. Among them, the single-target positioning can directly calculate the accurate solution of the target position according to the formula, while the multi-target positioning inevitably introduces errors due to the need to separate the received mixed and superimposed signals. Model 1 uses wavelet transform to perform time-frequency domain analysis on the signal.

[0072] Fourier Transform:

[0073] The Fourier transform is an important time-frequency analysis method that expands the signal in frequency and is very important for extracting the frequency information of the signal. Let f(t) be a one-dimensional signal, and its continuous Fourier transform is:

[0074] F(ω) = ∫f(t)e ―iωt dt (3.1)

[0075] In Equation (3.1), F(ω) is the Fourier integral of f(t). The function F(ω) can also be written in complex form:

[0076] F(ω) = R(ω) + iX(ω) = A(ω)e iΦ(ω) (3.2)

[0077] In Equation (3.2), A(ω) is the amplitude spectrum of f(t).

[0078] Advantages of the MUSIC algorithm: The basic idea of the MUSIC algorithm is to perform eigenvalue decomposition on the covariance matrix of the output data of an arbitrary array, so as to obtain the signal subspace corresponding to signal classification and the noise subspace orthogonal to the signal components. Then, the orthogonality of these two subspaces is used to construct a spatial spectrum function, and through spectrum peak search, the parameters of the signal are estimated. Its main advantages are as follows: Simultaneous lateral direction finding of multiple signals; High-precision direction finding; High-resolution lateral direction finding for signals within the antenna beam; Applicable to the case of small data samples; Real-time processing can be achieved after adopting high-speed processing technology.

[0079] Wavelet analysis theory: The important feature that the continuous wavelet transform is different from other transforms is that there is no fixed kernel function [1]. In wavelet transform, the wavelet must fully satisfy the reconstruction condition to perform the inverse wavelet transform. Let ψ(t) ∈ L 2 (R), when When the following conditions are satisfied [2], then:

[0080]

[0081] In practical applications, discretizing the continuous wavelet can overcome information redundancy and avoid information loss. The discrete wavelet transform mainly discretizes the scale factor a and the dilation factor b in the continuous wavelet transform, that is:

[0082]

[0083] Single target without noise (k = 1): At time t, the intermediate frequency signal received by the nth antenna of the FMCW radar is (here and afterwards, uniform sampling is performed with respect to time, and t is represented by T s t to represent t):

[0084]

[0085] That is:

[0086] s n,k (t) = sqrt(a^2 + b^2)e j(arctan(b / a)) (3.6)

[0087] Where: arctan(b / a) is the phase, γ = 78.986 × 10 12 hz / s, T s = 1.25 × 10 ―7 s, t = 0:1:255 arithmetic sequence (256 sampling points within 1 chirp period), c = 3 × 10 8m / s, f0=78.8×10 9 hz. According to the question, the coordinates of the nth antenna are:

[0088]

[0089] Distance to the kth object (X k =r k sinθ k ,Y k =r k cosθ k ) is:

[0090]

[0091] Among them, R n,k is the calculated two-way echo distance, aperture L = 0.0815, n = 0, 1, ..., 85 is the arithmetic progression corresponding to each antenna, N a = 86. The coordinates of the kth object (X k and Y k ) at the center of the circle Radius R n,k Just find the intersection point on the circle.

[0092] Multiple targets without noise (k>1): At time t within a chirp cycle, the intermediate frequency signal received by the nth antenna is a mixture of the noisy intermediate frequency signals of K target objects:

[0093]

[0094] Right now:

[0095]

[0096] Three different RX chirp pulses received from different objects (top) and the corresponding multiple IF single tone signals with constant frequency (bottom) Figure 1 shown.

[0097] According to the dependence of the frequency of the IF signal on the object speed, which can be ignored in fast FMCW radar, it can be known that for an object at a distance d from the radar, the IF signal will be a sine wave Asin(2πf o t+φ0).

[0098] In the scheme of the present invention, according to Solve the number of peaks and corresponding frequencies in the spectrum after wavelet transformation to determine the target number K and the corresponding frequency f0; thus obtain R n,k ; Determine the coordinates of the kth object (X k and Yk ) at the center of the circle radius R n,k on the circle. Draw all the circles, and at the same time, according to the fact that all the targets are located within a sector area with the origin as the center, a radius of less than 10 meters, and an opening angle of 100° upward, exclude the infeasible intersections.

[0099] Limitations of Fourier transform: First, perform Fourier analysis on the intermediate frequency (IF) signal sequences (with a length of 256) received by 86 antennas respectively. It is found that the spectra of all antennas reach their peaks at 3.69 MHz and have the same amplitude. Judging from the spectra of the intermediate frequency signals of each antenna, the number of spectral peaks being 1 indicates that there is only a single target. According to the relationship, the radial distance of the target corresponding to the intermediate frequency of 3.69 MHz is 7.003 m. Therefore, based on the above processing process, the result of the solution of the present invention is that a single target is approximately 7.003 meters away from the antenna. As Figure 2 shown.

[0100] However, due to the insufficient frequency resolution of the FFT algorithm, it is impossible to reflect the frequency differences of the intermediate frequency signals received by different antennas. Therefore, it is difficult to further determine the target azimuth, and there must be a certain error in the target distance deduced from the intermediate frequency. For this reason, a two-dimensional Fourier transform in the azimuth and distance domains is proposed. The peak value on the obtained two-dimensional ambiguity plane is the estimated target position. The specific implementation process is as follows:

[0101] (1) Construct a distance-angle grid. The Y-axis is the distance, the search range is 0 - 10 m, and the search step is 0.1 m. The X-axis is the angle, the search range is -50° to 50°, and the search step is 0.1°.

[0102] (2) At each point on the grid, generate the corresponding conjugate signal of the actual intermediate frequency signal according to the distance d of this point from the nth antenna n

[0103] (3) Multiply the conjugate signal by the actual intermediate frequency signal given in the problem. When d n and R n,k are equal, the phases cancel each other out, and the signal intensity (modulus) reaches the maximum.

[0104] Finally, the obtained result is as Figure 3 shown. The distance result is basically the same as that obtained directly using the Fourier transform before. Comprehensive analysis of the Fourier transform results shows that the target is a single object, 7 m away from the antenna, at an angle of 5°. The coordinate position is as Figure 4 shown in the right coordinate system. Obviously, the calculation result of the Fourier algorithm is extremely inaccurate, and the calculation of the angle is too rough, as Figure 3 ​​The signal average intensity obtained from the three-dimensional map shows a gentle peak at 7m instead of a sharp peak. This gentle peak is prone to misjudgment (such as overlooking the superimposed signal or misjudging the signal position). In the solution of the present invention, the correct result is two targets, but the Fourier calculation result is misjudged as one target. Therefore, it is necessary to select other better algorithms.

[0105] Solution result of MUSIC algorithm: Based on the MUSIC algorithm, the positioning solution of noise-free simulation data is realized. It can be seen that the signal intensity of the rear target is about 3 - 4dB stronger than that of the front target. The stripes around the peak should be the sidelobes of the azimuth spectrum. The result is as Figure 5 shown.

[0106] The result shows that the number of targets is 2. The coordinates of target 1 are (0, 7) and the coordinates of target 2 are (0.1, 8.2) in the (θ, r) coordinate system, as Figure 6 shown.

[0107] Compressive sensing sparsification: Through the sparsification process of compressive sensing, the result of the MUSIC algorithm can be further refined, as Figure 7 shown.

[0108] Wavelet algorithm test: Except for the noise in the low-frequency region, four significant peak points are found after wavelet transform, but there are only two strong targets. Combining with the MUSIC result, it is judged that there is a possibility in the solution of the present invention, that is, two large targets (as shown in the MUSIC result) and two small targets. Due to the resolution reason, only the accurate positions of the two large targets can be found, as Figure 8 shown.

[0109] In the solution of the present invention, two algorithms (Fourier transform, MUSIC algorithm) are used to solve the solution of the present invention and are used as a comparison with each other.

[0110] The result of Fourier transform is incorrect, but it can be used as a comparison. The solution result is a single target at 7.003m and an angle of 5°. This is the optimal result that can be obtained by the current radar resolution. According to the definition of the radar frequency resolution Δf, it satisfies According to it can be obtained That is, the radar range resolution

[0111] According to the radar range resolution Δd, for a target at a distance d and an angle θ, it can be calculated as Figure 9 shown; Explanation: The red dots are the leftmost and rightmost antennas, the red squares are the targets, the x-axis is the antenna array, the y-axis is the direction perpendicular to the array, L is the radar aperture, the shaded area symmetric about the y-axis is the detectable range of the radar given in the question, and θ is positive on the right and negative on the left.

[0112] |d right -d left |As Figure 10 shown. When the radar detection target is within the green dashed box |d right -d left |<Δd. When |d right -d left |<Δd, the influence of the radar aperture on target detection cannot be theoretically distinguished. Therefore, if the Fourier transform is used to calculate the distance according to the frequency, the radar can only be regarded as a point, which is extremely inaccurate.

[0113] Note: The area within the green dashed box is |d right -d left |<Δd area, and the area within the red dashed box is the radar detection range. The solution of the present invention selects the MUSIC algorithm to solve the problem.

[0114] The solution results of the present invention are shown in the following table:

[0115] Table 2 Coordinates of the target in the rectangular coordinate system and the polar coordinate system

[0116]

[0117] When using the Fourier transform, due to the limitation of resolution, only target 2 was found, and target 1 could not be distinguished. Therefore, this problem cannot be solved only by the analysis in the frequency domain. Therefore, the solution of the present invention adopts the MUSIC algorithm for solution and improves the resolution based on compressive sensing. Taking the midline of the fan-shaped area with the antenna opening upward and the central angle of 100° as 0°, target 1 (r: 8.20 m, θ: 0.00°) and target 2 (r: 7.00 m, θ: -0.01°) are clearly shown in the results. However, when using wavelet transform for testing, at least 4 obvious peaks were found. Therefore, there may be more targets in the solution of the present invention, but due to the resolution limitation of the measuring instrument, they cannot be accurately located.

[0118] Multi-target FMCW radar array noise reduction and positioning model based on sparse MUSIC algorithm, MUSIC technical principle: According to the signal and the array source noise are independent of each other. The MUSIC algorithm uses the orthogonality between the signal direction vector and the noise subspace to construct a spatial scanning spectrum, performs global search for spectral peaks, and thus realizes signal parameter estimation and noise reduction.

[0119] Calculation process of the MUSIC algorithm: Step1: Solve the autocorrelation matrix R of the received signal X of the array; Step2: Perform eigenvalue decomposition on R, and the eigenvectors corresponding to the smaller N-P eigenvalues are denoted as the noise subspace Q n; Step 3: Conduct spectral peak search to obtain the steering vectors at each angle in the spatial domain.

[0120] Based on the MUSIC algorithm, perform positioning solution for the noise-free simulation data, and the results are as Figure 11 shown. It can be seen that the intensity of the rear target signal is about 3 - 4 dB stronger than that of the front target. Figure 11 (Top) The fringes around the peak should be the sidelobes of the azimuth spectrum.

[0121] The results show that the number of targets is 2, and the coordinates of Target 1 (r: 8.20 m, θ: -0.01°) and Target 2 (r: 7.00 m, θ: -0.01°) in the (θ, r) coordinate system are as Figure 12 shown.

[0122] Compressive sensing sparsification: Through the sparsification process of compressive sensing, the results of the MUSIC algorithm can be further refined; as Figure 13 shown.

[0123] Wavelet test: Except for the noise in the low-frequency region, four significant peak points are found after wavelet transform, but there are only two strong targets. There are two relatively weak peaks near 2 MHz, suspected to be two weak targets. Combining the MUSIC results and the waveform of the first chirp period in Model 3, it is judged that there is a possibility in the solution of the present invention, that is, there are two large targets (as shown in the MUSIC results) and two small targets, but due to resolution reasons, only the exact positions of the two large targets can be found; as Figure 14 shown.

[0124] The solution results of the present invention are shown in the following table:

[0125] Table 3 Coordinates of the Targets in the Rectangular Coordinate System and the Polar Coordinate System

[0126]

[0127] The solution of the present invention uses the MUSIC algorithm for noise reduction and solution, and improves the resolution based on compressive sensing. The results clearly show Target 1 (r: 8.20 m, θ: -0.01°) and Target 2 (r: 7.00 m, θ: -0.01°). However, when using wavelet transform for testing, at least 4 obvious peaks are found. Therefore, there may be more targets in the solution of the present invention, but due to the resolution limitation of the measuring instrument, they cannot be accurately positioned.

[0128] Online low-time-complexity FMCW radar array positioning model based on improved MUSIC algorithm; Definition of online low-complexity algorithm: Online: In computer science, an online algorithm is one that can process inputs one by one in a serialized manner, meaning that it does not need to know all the inputs at the beginning. In contrast, for an offline algorithm, all the input data of the problem needs to be known at the start, and the result has to be output immediately after solving a problem. For example, selection sort needs to know all the elements to be sorted before sorting, while insertion sort does not.

[0129] Low complexity: The evaluation of an algorithm mainly considers time complexity and space complexity. Assuming that the time required for each statement to execute once is unit time, the time complexity of an algorithm is the sum of the frequencies of all statements in the algorithm, denoted as T(n), where n is called the scale of the problem. Similar to time complexity, space complexity refers to the measure of the storage space required by an algorithm when executing on a computer. It is denoted as: S(n) = O(f(n)). For the solution of the present invention, since the online algorithm has no storage requirements, the impact of time complexity is mainly considered.

[0130] Algorithm design: To reduce the time complexity of the MUSIC algorithm [7], first start from matrix operations and improve the row-by-row and column-by-column operations of the matrix that occupy a large number of for loops in the algorithm to a form of direct operation on the entire matrix.

[0131] The initial algorithm has more nested for loops and is less efficient when calculating conjugates. To make full use of the matrix operation capabilities of Matlab, we perform operations such as transposing and dot multiplying the coefficient matrices in the program to remove one layer of for loop and improve the running efficiency through overall matrix operations. The running time before improvement was 372 seconds, and the running time after improvement was 370 seconds.

[0132] Solution result of the present invention; As Figure 15 shown; Explanation: In the initial few Chirp cycles, only two energy peaks can be obtained using the MUSIC algorithm, that is, it is judged that the number of targets is 2, the target angles are all near 0 degrees, and the distances are approximately 6m and 9m respectively. Starting from the fourth Chirp cycle, due to the dispersion of target movement, the algorithm can further distinguish the target positions and finally obtain 4 energy spectral peaks.

[0133] Compressed sensing improves resolution:

[0134] Table 4 Results of MUSIC target point coordinates (r, θ) after compressed sensing improves resolution, where r (m) represents distance and θ (°) represents angle

[0135] chirp period Target 1 Target 2 Target 3 Target 4 1 (6.0,-0.8) (9.2,-0.8) (6.0,0.3) (9.2,0.3) 2 (6.0,-0.9) (9.1,-1.3) (6.1,0.9) (9.2,0.5) 3 (6.0,-0.8) (9.1,-1.4) (6.1,0.9) (9.2,0.5) 4 (6.0,-0.8) (9.1,-1.5) (6.1,1.0) (9.2,0.6) 5 (5.9,-2.0) (9.1,-1.6) (6.1,1.1) (9.2,0.8) 6 (5.9,-2.1) (9.1,-1.7) (6.1,1.3) (9.3,1.6) 7 (5.9,-2.3) (9.1,-1.9) (6.1,1.4) (9.3,1.8) 8 (5.9,-2.5) (9.1,-2.1) (6.1,1.6) (9.3,2.0) 9 (5.9,-2.7) (9.1,-2.3) (6.1,1.8) (9.3,2.2) 10 (5.9,-2.9) (9.0,-2.6) (6.1,2.1) (9.3,2.4) 11 (5.9,-3.2) (9.0,-2.8) (6.2,2.3) (9.3,2.7) 12 (5.8,-3.4) (9.0,-3.1) (6.2,2.6) (9.4,2.9) 13 (5.8,-3.6) (9.0,-3.3) (6.2,2.8) (9.4,3.1) 14 (5.8,-3.9) (9.0,-3.5) (6.2,3.0) (9.4,3.4) 15 (5.8,-4.1) (9.0,-3.8) (6.2,3.3) (9.4,3.6) 16 (5.8,-4.3) (9.0,-4.0) (6.2,3.5) (9.4,3.8) 17 (5.8,-4.5) (8.9,-4.2) (6.3,3.7) (9.4,4.0) 18 (5.7,-4.8) (8.9,-4.4) (6.3,3.9) (9.5,4.3) 19 (5.7,-5.0) (8.9,-4.7) (6.3,4.2) (9.5,4.5) 20 (5.7,-5.2) (8.9,-4.9) (6.3,4.4) (9.5,4.7) 21 (5.7,-5.5) (8.9,-5.1) (6.3,4.6) (9.5,5.0) 22 (5.7,-5.7) (8.9,-5.4) (6.3,4.9) (9.5,5.2) 23 (5.7,-5.9) (8.8,-5.6) (6.3,5.1) (9.5,5.4) 24 (5.6,-6.2) (8.8,-5.8) (6.4,5.3) (9.5,5.7) 25 (5.6,-6.4) (8.8,-6.0) (6.4,5.6) (9.6,5.9) 26 (5.6,-6.6) (8.8,-6.3) (6.4,5.8) (9.6,6.1) 27 (5.6,-6.8) (8.8,-6.5) (6.4,6.0) (9.6,6.3) 28 (5.6,-7.1) (8.8,-6.7) (6.4,6.2) (9.6,6.6) 29 (5.6,-7.3) (8.8,-7.0) (6.4,6.5) (9.6,6.8) 30 (5.6,-7.5) (8.7,-7.2) (6.5,6.7) (9.6,7.0) 31 (5.5,-7.8) (8.7,-7.4) (6.5,6.9) (9.7,7.3) 32 (5.5,-8.0) (8.7,-7.6) (6.5,7.2) (9.7,7.5)

[0136] Wavelet test, asFigure 17 As shown, it can be seen from the figure that wavelet analysis always believes that there are four peaks, that is, there are four targets, which is consistent with the conclusion in the solution of the present invention. The only difference is that the resolution at the beginning of the MUSIC algorithm is not sufficient to fully resolve the four objects, and it can be resolved after the fourth cycle.

[0137] The solution of the present invention uses the MUSIC algorithm to solve noise reduction, improves the resolution based on compressive sensing, and finally uses wavelet transform for verification. When inputting, a per-cycle sequence is input to meet the requirements of the online algorithm. The result shows that the number of targets is 4. In the first approximately 3 chirp cycles, the MUSIC algorithm cannot fully resolve the 4 targets, and it can be clearly seen only after the 4th chirp cycle. However, the high-resolution result after compressive sensing can identify all 4 targets. The object trajectories are as shown in the above figure, all of which are linear motions. Among them, target 3 (azimuth 29.68°, origin coordinates (6.0, 0.3), destination coordinates (6.5, 7.2)) and target 4 ((azimuth 19.80°), origin coordinates (9.2, 0.3), destination coordinates (9.7, 7.5)) move forward to the right, and target 1 (azimuth 201.80°, origin coordinates (6.0, -0.8), destination coordinates (5.5, -8.0)) and target 2 (azimuth 197.74°, origin coordinates (9.2, -0.8), destination coordinates (8.7, -7.6)) move backward to the left, and the coordinates are distance-angle coordinates.

[0138] For the positioning robustness improvement algorithm based on smoothing filtering, when there is an error in the positioning of a certain damaged antenna array for itself, the intermediate frequency signal it receives can be equivalent to the result measured when the target deviates a certain distance before the antenna is damaged, that is, all targets only deviate when measured by the damaged antenna, and the deviation amount is consistent with the deviation of the radar's own positioning at that moment, and there is no deviation when measured by the normal antenna. And this deviation can be used to determine and correct the damaged antenna. This model analyzes and solves the damage of the radar array antenna according to the number of damaged antennas, including single-antenna damage and multi-antenna damage situations.

[0139] Single-antenna fault: In the case of antenna damage, assume that the number of damaged antennas is 1, and its own positioning offset amounts are Δx and Δy. At this time, for this damaged antenna (such as Figure 18 the blue dot), the received intermediate frequency signal is equivalent to the target point (such as Figure 18 the green dot) moving -Δx and -Δy when it is intact, so the damaged antenna can be determined by observing the consistency of the antenna spectrogram.

[0140] Multi - antenna fault: According to the analysis of single - antenna damage, the self - positioning offset can be replaced by the offset of the target point. For the MUSIC algorithm, the offset of the target point will generate a secondary peak on the edge of the original peak or cause the original peak to split into sidelobes. After sparsification, this peak becomes more obvious, as Figure 19 shown. Obviously, the influence of the antenna's self - positioning error in the solution of the present invention is to generate sidelobes in the signal intensity. Here, the smoothing filtering method is used to remove the sidelobes, and the removal result is as Figure 20 shown.

[0141] Equivalent the positioning deviation of the antenna to the offset of the target object, so as to distinguish the damaged antenna according to the difference in the spectral waveform and correct it. The antenna damage situation is divided into single - antenna damage situation and multi - antenna damage situation for discussion, realizing the optimization of the robustness of antenna positioning, and based on this, using smoothing filtering to achieve the optimization of the antenna result.

[0142] Through the solution of the present invention, a solution model based on the sparsified MUSIC algorithm is constructed to accurately solve the problem target when solving Problem 1. At the same time, the characteristics of Fourier transform and wavelet transform are used to give the analysis of the target characteristics, verifying the accuracy of the algorithm.

[0143] Use the signal subspace corresponding to signal classification and the noise subspace orthogonal to the signal components in the MUSIC algorithm to distinguish the noise, so as to realize the noise reduction and positioning of the target in Problem 2.

[0144] The solution of the present invention replaces the for loop with matrix operations, reducing the time complexity of the MUSIC algorithm from T(n3) to T(n2) at the technical level, and constructing the algorithm input in a sequential manner to achieve online low complexity, and finally accurately identifying all 4 target trajectories.

[0145] The solution of the present invention determines the damaged antenna according to the difference in the spectral waveform, and then uses the smoothing filtering algorithm to correct the sidelobes of the signal, realizing accurate correction and reducing the resource waste caused by traversal.

[0146] The solution of the present invention uses the sparsified MUSIC algorithm, Fourier transform, and wavelet transform to study the positioning problem, avoiding the problem that when the system resolution is insufficient, only relying on the frequency - domain analysis can only judge the distance of the object but not the azimuth of the object. For example, in the solution of the present invention, using the sparsified MUSIC algorithm can effectively improve the resolution of the FMCW radar array positioning system, while using Fourier transform cannot achieve this resolution. Using the MUSIC algorithm can realize the positioning of the distance and azimuth of the target in the presence of noise, and using compressive sensing sparsification for the result can improve its resolution.

[0147] The solution of the present invention uses wavelet transform as an auxiliary means to judge the number of targets, enriching the basis for positioning judgment. The main influence of the self-positioning deviation of the antenna on the signal is the generation of side lobes, and the solution is to perform smoothing filtering on the signal. The low radar resolution of the solution of the present invention limits the use of the algorithm. For example, too low a sampling rate results in indistinguishable differences in the Fourier transform results of different antennas. Assuming that the instrument resolution increases, there can be more abundant solutions to this problem. For example, the solution based on the Fourier algorithm can utilize the strong robustness of the Fourier transform.

[0148] To achieve the above object, the present invention also provides a processing system for improving the robustness of target positioning based on smoothing filtering, as Figure 21 shown. The system is applied to the method for improving the robustness of target positioning based on smoothing filtering. The system includes: a first data generation unit, configured to generate and obtain at least two first data corresponding to the target object to be positioned by the FMCW radar, and generate corresponding second data according to the first data; wherein, the first data is the reflected signal data of the target object; the second data includes the distance data of the target object and the azimuth data of the target object; a second data generation unit, configured to generate and obtain eighth data corresponding to the FMCW radar based on the second data, and generate ninth data corresponding to the target object according to the eighth data; wherein, the eighth data is the positioning deviation data of the antenna; the ninth data is the offset data corresponding to the target object and the eighth data; a data processing and generation unit, configured to generate corresponding and highly robust third data through two-dimensional Fourier transform processing, sparse MUSIC algorithm processing, and smoothing filtering processing in sequence according to the second data and in combination with the eighth data; wherein, the third data is the position positioning data of the target object.

[0149] The first data generation unit further includes: a first generation module, configured to generate fourth data corresponding to the target object to be positioned by the FMCW radar according to the first data; wherein, the fourth data is the double echo distance data; a first determination module, configured to determine and generate fifth data corresponding to the target object based on the fourth data; wherein, the fifth data is the coordinate data of the target object. And / or, the second data generation unit further includes: a seventh generation module, configured to generate and obtain at least one frame of tenth data corresponding to the moving target object; wherein, the tenth data is the intermediate frequency signal data; an eighth generation module, configured to generate eleventh data corresponding to the moving target object and being online low complexity according to the tenth data; wherein, the eleventh data is the motion trajectory data of the target object.

[0150] The data processing and generation unit further includes: a first construction module for constructing a distance-angle grid corresponding to the target object to be located by the FMCW radar; a second generation module for generating conjugate signal data corresponding to the intermediate-frequency signal data of the target object based on the distance-angle grid; a third generation module for generating sixth data corresponding to the target object to be located by the FMCW radar according to the intermediate-frequency signal data of the target object and the conjugate signal data; wherein the sixth data is the target object position estimation data.

[0151] The data processing and generation unit further includes: a fourth generation module for generating a first matrix corresponding to the array signal and performing eigenvalue decomposition processing on the first matrix; wherein the first matrix is the autocorrelation matrix corresponding to the array received signal data; a fifth generation module for generating a first subspace corresponding to the first matrix based on the first matrix; wherein the first subspace is the noise subspace; a sixth generation module for constructing a corresponding spatial spectrum function according to the first subspace and generating corresponding seventh data through spectrum peak search; wherein the seventh data is the steering vector data of each angle in the spatial domain.

[0152] The data processing and generation unit further includes: a first processing module for respectively performing compressive sensing sparsification processing on the angle direction and the distance direction corresponding to the positioning target object; a ninth generation module for generating corresponding twelfth data according to the eighth data and in combination with the spectrum waveform corresponding to the radar antenna; wherein the twelfth data is the antenna damage state data, including single-antenna damage situation data and multi-antenna damage situation data.

[0153] In the embodiment of the system solution of the present invention, the method steps involved in the improvement of target positioning robustness processing based on smoothing filtering have been described in detail above. That is to say, the functional modules in the system are used to implement the steps or sub-steps in the above method embodiment, which will not be elaborated here.

[0154] The present invention generates and obtains at least two first data corresponding to a target object to be located by an FMCW radar, and generates corresponding second data according to the first data; wherein, the first data is reflection signal data of the target object; the second data includes target object distance data and target object azimuth data; based on the second data, eighth data corresponding to the FMCW radar is generated and obtained, and ninth data corresponding to the target object is generated according to the eighth data; wherein, the eighth data is antenna positioning deviation data; the ninth data is offset data of the target object corresponding to the eighth data; according to the second data, and in combination with the eighth data, two-dimensional Fourier transform processing, sparse MUSIC algorithm processing, and smoothing filtering processing are sequentially performed to generate corresponding third data with high robustness; wherein, the third data is target object position positioning data, and a system corresponding to the method is provided to achieve super-resolution positioning of the target object and greatly improve the positioning accuracy of multiple target objects; moreover, the target positioning effect in a strong noise environment can be improved, and the time complexity and space complexity can be reduced, and the robustness of target object positioning can be improved.

[0155] That is to say, mobile scenario super-resolution positioning has broad application prospects in people's work and life. The solution of the present invention is based on algorithms such as sparse MUSIC algorithm, wavelet transform, Fourier transform, and smoothing filtering, and conducts research on the problem of mobile scenario super-resolution positioning, realizing the precise positioning of FMCW (Frequency Modulated Continuous Wave) array radar targets with low time complexity online. A solution model based on the sparse MUSIC algorithm is constructed. The results show that the number of targets in the solution of the present invention is 2, the coordinates of target 1 are (r: 7.00 m, θ: -0.01°), and the coordinates of target 2 are (r: 8.20 m, θ: 0.00°); due to resolution limitations, the Fourier transform can only detect target 1, and the wavelet transform verification shows 4 significant peak points (2 strong and 2 weak), so it is judged that there may be 4 targets (2 large and 2 small) in the solution of the present invention, and the small targets cannot be accurately positioned due to resolution limitations. A noise reduction solution model based on the sparse MUSIC algorithm is constructed. The results show that there are 2 targets in the solution of the present invention, namely target 1 (r: 8.20 m, θ: -0.01°) and target 2 (r: 7.00 m, θ: -0.01°); through wavelet transform, 4 obvious wave peaks are found, indicating that there may be more targets. By replacing the for loop with matrix operations, the time complexity of the traversal link of the MUSIC algorithm is reduced from T(n3) to T(n2). At the same time, the algorithm input is constructed in a sequential manner, and then an improved MUSIC algorithm with low complexity online is realized. The results show that the resolution optimization results are obtained by processing the results of the MUSIC algorithm based on the compressive sensing sparsification method, and finally the trajectories of all 4 targets can be accurately identified. By equating the positioning deviation of the antenna to the offset of the target object, the damaged antenna is positioned according to the difference in the spectral waveform, and the robustness optimization of the antenna positioning is realized by using the smoothing filtering algorithm.

[0156] The above-described embodiments merely represent several implementation manners of the present invention. The description is relatively specific and detailed, but it should not be construed as a limitation on the scope of the patent of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several deformations and improvements can still be made, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the appended claims.

Claims

1. A method for improving the robustness of target positioning based on smoothing filtering, characterized in that: The method comprises: Generate and obtain at least two first data corresponding to the target object to be located by the FMCW radar, and generate corresponding second data according to the first data; wherein the first data is the reflected signal data of the target object; and the second data includes the target object distance data and the target object azimuth data; Based on the second data, eighth data corresponding to the FMCW radar is generated and acquired, and ninth data corresponding to the target object is generated according to the eighth data; wherein the eighth data is positioning deviation data of the antenna; and the ninth data is offset data of the target object corresponding to the eighth data; According to the second data and in combination with the eighth data, corresponding and highly robust third data are generated through two-dimensional Fourier transform processing, sparse MUSIC algorithm processing, and smoothing filtering processing in sequence; wherein the third data is target object position positioning data.

2. The method for improving the robustness of target positioning based on smoothing filtering according to claim 1, characterized in that: The generating and acquiring at least two first data corresponding to the target object to be positioned by the FMCW radar, and generating corresponding second data according to the first data, further includes: Generate fourth data corresponding to the target object to be located by the FMCW radar according to the first data; wherein the fourth data is double echo distance data; Based on the fourth data, it is determined to generate fifth data corresponding to the target object; wherein the fifth data is the coordinate data of the target object.

3. The method for improving the robustness of target positioning based on smoothing filtering according to claim 1, characterized in that: The method of generating and acquiring eighth data corresponding to the FMCW radar based on the second data, and generating ninth data corresponding to the target object according to the eighth data, further includes: Generate and acquire tenth data corresponding to the moving target object and of at least one frame; wherein the tenth data is intermediate frequency signal data; According to the tenth data, an eleventh data corresponding to the moving target object and having online low complexity is generated; wherein the eleventh data is the target object motion trajectory data.

4. The method for improving the robustness of target positioning based on smoothing filtering according to claim 1, characterized in that: The method further comprises: generating corresponding third data with high robustness by sequentially performing two-dimensional Fourier transform processing, sparse MUSIC algorithm processing, and smoothing filter processing according to the second data and in combination with the eighth data, and further comprising: Construct a range angle grid corresponding to the target object to be located by the FMCW radar; Based on the distance angle grid, generating conjugate signal data corresponding to the intermediate frequency signal data of the target object; According to the intermediate frequency signal data of the target object and the conjugate signal data, sixth data corresponding to the target object to be positioned by the FMCW radar is generated; wherein the sixth data is the estimated position data of the target object.

5. A method for improving target positioning robustness based on smoothing filtering according to claim 1 or 3, characterized in that: The method further comprises: generating corresponding third data with high robustness by sequentially performing two-dimensional Fourier transform processing, sparse MUSIC algorithm processing, and smoothing filter processing according to the second data and in combination with the eighth data, and further comprising: Generate a first matrix corresponding to the array signal, and perform eigenvalue decomposition processing on the first matrix; wherein the first matrix is ​​an autocorrelation matrix corresponding to the array received signal data; Based on the first matrix, generating a first subspace corresponding to the first matrix; wherein the first subspace is a noise subspace; According to the first subspace, a corresponding spatial spectrum function is constructed, and corresponding seventh data is generated through spectrum peak search; wherein the seventh data is the steering vector data of each angle in the spatial domain.

6. A method for improving the robustness of target positioning based on smoothing filtering according to claim 5, characterized in that: The method further comprises: generating corresponding third data with high robustness by sequentially performing two-dimensional Fourier transform processing, sparse MUSIC algorithm processing, and smoothing filter processing according to the second data and in combination with the eighth data, and further comprising: Compressed sensing and sparse processing are respectively performed on the angle direction and distance direction corresponding to the positioning target object; According to the eighth data and in combination with the frequency spectrum waveform corresponding to the radar antenna, the corresponding twelfth data is generated; wherein the twelfth data is antenna damage status data, including single antenna damage status data and multi-antenna damage status data.

7. A processing system for improving target positioning robustness based on smoothing filtering, characterized in that: The system is applied to the method for improving the robustness of target positioning based on smoothing filtering as claimed in claim 1, and the system comprises: A first data generating unit is used to generate and obtain at least two first data corresponding to the target object to be positioned by the FMCW radar, and generate corresponding second data according to the first data; wherein the first data is the reflection signal data of the target object; and the second data includes the target object distance data and the target object azimuth data; a second data generating unit, configured to generate and acquire eighth data corresponding to the FMCW radar based on the second data, and generate ninth data corresponding to the target object according to the eighth data; wherein the eighth data is positioning deviation data of the antenna; and the ninth data is offset data of the target object corresponding to the eighth data; A data processing and generating unit is used to generate corresponding and highly robust third data based on the second data and in combination with the eighth data, by sequentially performing two-dimensional Fourier transform processing, sparse MUSIC algorithm processing, and smoothing filtering processing; wherein the third data is target object position positioning data.

8. A system for improving target positioning robustness based on smoothing filtering according to claim 7, characterized in that: The first data generating unit further includes: A first generating module is used to generate fourth data corresponding to the target object to be positioned by the FMCW radar according to the first data; wherein the fourth data is double echo distance data; A first determination module, configured to determine and generate fifth data corresponding to the target object based on the fourth data; wherein the fifth data is coordinate data of the target object; And / or, the second data generating unit further includes: A seventh generating module is used to generate and obtain tenth data corresponding to the moving target object and of at least one frame; wherein the tenth data is intermediate frequency signal data; The eighth generation module is used to generate, based on the tenth data, eleventh data corresponding to the moving target object and having online low complexity; wherein the eleventh data is the target object motion trajectory data.

9. The system for improving the robustness of target positioning based on smoothing filtering according to claim 7, characterized in that: The data processing and generating unit further includes: A first construction module is used to construct a range angle grid corresponding to the target object to be positioned by the FMCW radar; A second generating module, used for generating conjugate signal data corresponding to the intermediate frequency signal data of the target object based on the distance angle grid; The third generating module is used to generate sixth data corresponding to the target object to be positioned by the FMCW radar according to the intermediate frequency signal data of the target object and the conjugate signal data; wherein the sixth data is the estimated position data of the target object.

10. The system for improving the robustness of target positioning based on smoothing filtering according to claim 7, characterized in that: The data processing and generating unit further includes: A fourth generating module is used to generate a first matrix corresponding to the array signal, and simultaneously perform eigenvalue decomposition processing on the first matrix; wherein the first matrix is ​​an autocorrelation matrix corresponding to the array received signal data; a fifth generating module, configured to generate a first subspace corresponding to the first matrix based on the first matrix; wherein the first subspace is a noise subspace; A sixth generating module, used to construct a corresponding spatial spectrum function according to the first subspace, and generate corresponding seventh data by spectrum peak search; wherein the seventh data is the steering vector data of each angle in the spatial domain; And / or, the data processing and generating unit further includes: The first processing module is used for compressing and sparsely processing the angle direction and the distance direction corresponding to the positioning target object; The ninth generating module is used to generate corresponding twelfth data according to the eighth data and in combination with the frequency spectrum waveform corresponding to the radar antenna; wherein the twelfth data is antenna damage status data, including single antenna damage status data and multi-antenna damage status data.

Images