Echo analysis method and device for radar detection, detection equipment and computer readable storage medium

Abstract

The invention relates to the technical field of radar signal processing, and discloses an echo analysis method for radar detection, and the method comprises the steps: carrying out the detection of a target scene through a coprime linear frequency modulation signal of a pulse repetition period, and obtaining a demodulated time domain echo signal; constructing a distance dimension sparse observation model according to the demodulated time domain echo signal; performing iterative solution on the distance dimension sparse observation model through a sparse Bayesian learning reconstruction algorithm to obtain a reconstructed target unambiguous distance dimension scattering feature vector, and constructing a long pulse signal model after distance dimension folding; and according to the target unambiguous distance dimension scattering feature vector and a long pulse signal model, obtaining a time domain echo signal after distance dimension unfolding. The invention further discloses an echo analysis device for radar detection, detection equipment and a computer readable storage medium.

Description

Technical Field

[0001] The present application relates to the field of radar signal processing technology, for example, to an echo analysis method and apparatus for radar detection, detection equipment, and computer-readable storage medium. Background Art

[0002] Radar systems such as imaging radar, ranging radar, and weather radar generally use linear frequency modulation (LFM) signals as a conventional detection mechanism. Accurately estimating the distance, position, and echo intensity of external scene targets is key to radar systems achieving target detection and high-precision imaging. However, the radar's maximum unambiguous range is limited by the radar system's pulse repetition time (PRT). In practical applications, shorter PRTs may be used to meet the requirements of velocity estimation and high range resolution. When the target range is too large or the distance is far, the target may be located outside the radar's maximum unambiguous range. Echoes from distant targets and unambiguous echoes from nearby targets will alias, causing the ambiguous echo to overlap with the original echo. This not only limits the radar's detection range, but the false target echoes caused by range aliasing will also interfere with echoes from scene targets within the detection range, leading to errors in the radar's estimation of target position, intensity, velocity, and other information, seriously affecting the data quality and imaging accuracy of radar detection products.

[0003] Traditional radars primarily remove range ambiguity by discarding the radar reflectivity factor data collected above, or by employing a more complex random phase (RANDP) method, which randomly adds phase offsets to different pulse trains before transmission and performs correlation modulation on echoes from different range intervals. Because this method relies primarily on modulation and correlation extraction of different echoes, the white noise introduced by phase modulation reduces the signal-to-noise ratio of the signal in the corresponding range interval, resulting in power loss of scene targets and reduced echo signal data quality. Summary of the Invention

[0004] In order to provide a basic understanding of some aspects of the disclosed embodiments, a brief summary is given below. The summary is not an extensive review, nor is it intended to identify key / critical elements or delineate the scope of protection of these embodiments, but rather serves as a prelude to the detailed description that follows.

[0005] The embodiments of the present disclosure provide an echo analysis method and apparatus, detection equipment, and computer-readable storage medium for radar detection, so as to utilize sparse recovery theory to complete the reconstruction of time domain echo signals, maintain the reconstruction accuracy of time domain echo signals, and optimize the data quality of time domain echo signals.

[0006] In some embodiments, the method includes: detecting the target scene through a linear frequency modulation signal with mutually prime pulse repetition periods to obtain a demodulated time domain echo signal; constructing a distance-dimensional sparse observation model based on the demodulated time domain echo signal; iteratively solving the distance-dimensional sparse observation model through a sparse Bayesian learning reconstruction algorithm to obtain a reconstructed target unambiguous distance-dimensional scattering feature vector, and constructing a long pulse signal model after distance-dimensional unfolding; obtaining a time domain echo signal after distance-dimensional unfolding based on the target unambiguous distance-dimensional scattering feature vector and the long pulse signal model.

[0007] In some embodiments, detecting a target scene using a linear frequency modulation signal with mutually prime pulse repetition periods to obtain a demodulated time domain echo signal includes: detecting the target scene using a linear frequency modulation signal with the same signal bandwidth and mutually prime pulse repetition periods to obtain an echo signal; mixing the echo signal with a reference signal to obtain a demodulated time domain echo signal; determining the least common multiple of the pulse repetition periods of all linear frequency modulation signals as the pulse repetition period T of the long pulse signal. m , and determine c·T m / 2 is the unambiguous distance.

[0008] In some embodiments, the distance-dimensional sparse observation model includes Ψ·σ=s(t); wherein Ψ, σ, and s(t) represent the sparse recovery basis matrix and the reconstructed target unambiguous distance-dimensional scattering feature vector and observation vector, respectively; the distance-dimensional sparse observation model is iteratively solved by a sparse Bayesian learning reconstruction algorithm to obtain the reconstructed target unambiguous distance-dimensional scattering feature vector, including: iteratively solving the distance-dimensional sparse observation model by a sparse Bayesian learning reconstruction algorithm; calculating the mean of the posterior probability based on the distance unfolding equation and the initialization hyperparameters and noise variance; updating the hyperparameters and noise variance, and determining the target mean corresponding to the target hyperparameters that meet the iteration termination conditions to obtain the reconstructed target unambiguous distance-dimensional scattering feature vector based on the target mean.

[0009] In some embodiments, the hyperparameters γ and noise variance σ are updated 2 ,include: in, Γ≡diag(γ),Σ x =σ 2 I+AΓA H ; A represents the M×N observation basis matrix in the distance unfolding equation, Σ n,n denotes the diagonal elements of Σ.

[0010] In some embodiments, constructing a long pulse signal model after distance dimension unfolding includes: obtaining the sampling frequency F of all linear frequency modulation signals sSum signal bandwidth B and spatial propagation distance R of each linear frequency modulation signal i , carrier frequency f0; determine the signal bandwidth B and the pulse repetition period T of the long pulse signal m The ratio is the frequency modulation slope μ of the long pulse signal; a long pulse signal model Φ is constructed; wherein the long pulse signal model Φ is

[0011]

[0012] In some embodiments, according to the target unambiguous range-dimensional scattering characteristic vector and the long pulse signal model, obtaining the time domain echo signal after range-dimensional unfolding includes: according to s(t M )=Φ·σ and the long pulse signal model Φ is solved to obtain the time domain echo signal s(t M ).

[0013] In some embodiments, a distance dimension sparse observation model is constructed based on the demodulated time domain echo signal, including: converting the demodulated time domain echo signal sequence [s(t1), s(t2), ..., s(t N )] are arranged vertically to form a one-dimensional column vector as the observation vector s(t); the beat signal data model is used to construct a sparse recovery basis matrix Ψ and a distance-dimensional sparse observation model; wherein the distance-dimensional sparse observation model includes: Ψ·σ=s(t); Ψ represents a sparse recovery basis matrix composed of a plurality of sub-matrices arranged vertically based on a linear frequency modulation signal with mutually prime pulse repetition periods, σ represents the unambiguous distance-dimensional scattering feature vector of the reconstructed target, and the plurality of sub-matrices are represented as [Ψ1,Ψ2,…,Ψ N ].

[0014] In some embodiments, the apparatus includes a processor and a memory storing program instructions, and the processor is configured to execute the aforementioned echo analysis method for radar detection when running the program instructions.

[0015] In some embodiments, the detection device includes: a detection device body; and an echo analysis device for radar detection as described above, installed on the detection device body.

[0016] In some embodiments, the storage medium stores program instructions, which, when executed, enable a computer to execute the aforementioned echo analysis method for radar detection.

[0017] The echo analysis method, apparatus, detection device, and computer-readable storage medium for radar detection provided by the embodiments of the present disclosure can achieve the following technical effects:

[0018] The detection device first detects the target scene through a linear frequency modulation signal with mutually prime pulse repetition periods to obtain a demodulated time domain echo signal and constructs a demodulated signal echo model. The detection device then constructs a distance dimension sparse observation model based on the demodulated time domain echo signal. After constructing the distance dimension sparse observation model, the detection device iteratively solves the distance sparse observation model through a sparse Bayesian learning reconstruction algorithm to obtain a reconstructed target unambiguous distance dimension scattering feature vector and constructs a distance dimension unfolded long pulse signal model, and obtains a distance dimension unfolded time domain echo signal based on the target unambiguous distance dimension scattering feature vector and the long pulse signal model, so as to realize echo distance unfolding through sparse recovery theory and ensure the reconstruction accuracy of the time domain echo signal. The embodiment of the present disclosure utilizes sparse recovery theory to complete the reconstruction of the time domain echo signal on the basis of realizing echo distance unfolding, maintain the reconstruction accuracy of the time domain echo signal, and optimize the data quality of the time domain echo signal.

[0019] The above general description and the following description are exemplary and explanatory only and are not intended to limit the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] One or more embodiments are exemplarily described by corresponding drawings. These exemplary descriptions and drawings do not limit the embodiments. Elements with the same reference numerals in the drawings are shown as similar elements. The drawings do not constitute a scale limitation. In addition,

[0021] Figure 1 is a schematic diagram of the change of the frequency of the linear frequency modulation signal over time provided by an embodiment of the present disclosure;

[0022] Figure 2 1 is a schematic diagram of a beat signal after a linear frequency modulation signal transmission pulse and an echo signal are demodulated according to an embodiment of the present disclosure;

[0023] Figure 3 is a schematic diagram of an echo analysis method for radar detection provided by an embodiment of the present disclosure;

[0024] Figure 4 is a schematic diagram of another echo analysis method for radar detection provided by an embodiment of the present disclosure;

[0025] Figure 5 is a schematic diagram of another echo analysis method for radar detection provided by an embodiment of the present disclosure;

[0026] Figure 6 It is a schematic diagram of the scattering point target scene setting in the simulation experiment;

[0027] Figure 7 This is a schematic diagram of the echo range-dimensional spectrum after folding the range dimension using the sparse recovery theory in the simulation experiment;

[0028] Figure 8 This is a schematic diagram of the target scene imaging after the range dimension is folded using the sparse recovery theory in the simulation experiment;

[0029] Figure 9 This is a schematic diagram of an echo analysis device for radar detection provided by an embodiment of the present disclosure. DETAILED DESCRIPTION

[0030] In the description and claims of the embodiments of the present disclosure, as well as in the accompanying drawings, the terms "first," "second," and the like are used to distinguish similar items and are not necessarily used to describe a particular order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate to describe the embodiments of the present disclosure herein. In addition, the terms "including," "having," and any variations thereof are intended to cover non-exclusive inclusions.

[0031] Unless otherwise stated, the term "plurality" means two or more.

[0032] In the embodiment of the present disclosure, the character " / " indicates that the preceding and following objects are in an "or" relationship. For example, A / B means: A or B.

[0033] The term "and / or" describes an association between objects, indicating that three relationships can exist. For example, A and / or B means: A or B, or A and B.

[0034] The term "correspondence" may refer to an association relationship or a binding relationship. The correspondence between A and B means that there is an association relationship or a binding relationship between A and B.

[0035] First, the theoretical derivation and simulation experimental verification process of the present invention are briefly described. Specifically, the echo analysis process of linear frequency modulation radar echo range unfolding is realized based on the linear frequency modulation radar signal processing theory, the principle of multi-PRT to achieve unambiguous distance extension, the Bayesian and statistical method theory and the sparse recovery theory.

[0036] For radar systems, their maximum detection range is proportional to the pulse width (hereinafter referred to as signal pulse width) T of the transmitted pulse. This means that the longer the signal pulse width, the longer the radar's maximum range. The radar's range resolution is ΔR = c / 2B. It can be seen that the radar's range resolution is inversely proportional to the signal bandwidth, meaning that the larger the signal bandwidth, the better the radar's range resolution. For general pulse signals, there is a conflict between the signal pulse width and the signal bandwidth, that is, B·T = 1. Linear Frequency Modulation (LFM), a core technology in modern radar, sonar, and communication systems, has revolutionized the field of signal processing with its large time-bandwidth product and time-frequency characteristics, becoming a core means of achieving high-resolution detection and high-precision measurement.

[0037] The mathematical model of the LFM signal can be expressed as:

[0038]

[0039] in, μ represents the frequency modulation slope of the long pulse signal, B represents the signal bandwidth of the linear frequency modulation signal, T represents the signal pulse width of the transmitted pulse, f0 represents the carrier frequency, A0 represents the initial signal amplitude of the transmitted pulse, and the time-frequency characteristics of the initial signal amplitude present the classic sawtooth wave characteristics, such as Figure 1 As shown, its instantaneous frequency is f(t)=f0+μt.

[0040] As modern radar applications expand, the demand for range resolution continues to increase, and the bandwidth of transmitted pulses is also increasing. Processing such wide-bandwidth signals, even though the analog-to-digital conversion performance of receivers has greatly improved with recent hardware developments, the large amount of collected data still poses significant challenges in storage capacity and real-time processing. Therefore, before signal processing, dechirping (de-chirping) is often used to reduce the signal bandwidth and lower the required sampling rate, thus alleviating the trade-off between high resolution and input signal bandwidth.

[0041] The demodulation frequency processing is achieved by mixing the local reference signal with the received signal, converting the time difference between the echo signal and the local oscillator signal at different distances into different frequency differences. Finally, a Fast Fourier Transform (FFT) is used to achieve pulse compression in the frequency domain, distinguishing targets at different distance units and achieving high resolution.

[0042] Transmit pulse and local reference signal s ref (t) As shown in formula (1), the time delay of the echo signal reflected by the target at distance R1 after the signal propagates through space compared to the reference signal is τ1 = 2R1 / c, where c is the speed of light. The mathematical model of the echo signal is:

[0043]

[0044] In formula (2), σ1 uniformly represents the distance transmission loss of target 1 in the echo signal and the multiplication coefficient caused by the scattering intensity determined by the target's own scattering characteristics, scattering cross-sectional area, incident angle, etc. on the initial signal amplitude A0. The beat signal data model obtained after demodulation is:

[0045]

[0046] The echo signal received by a single array element is the superposition of complex signals scattered from all target points in the scene. For digital signal processing, after determining the current pulse repetition period T and the sampling frequency F of the distance dimension, s After that, the division of distance dimension is also fixed, that is, the distance dimension sequence is R = [0, R1, R2, ..., R max ]. Therefore, the total echo beat signal can be modeled as the sum of the target beat signals in all distance dimensions within the scene.

[0047]

[0048] In formula (4), Among them, τ i =2R i / c, where I represents the total number of scattered signals from all target points in the scene. Furthermore, in formula (4), not all distance dimensions contain targets. That is, the vector σ formed by the amplitude and phase response scattering coefficients of scene targets in the distance dimension satisfies sparsity. In this case, the equation Φ·σ=s(t) is obtained, and the scattering characteristic vector σ of the target scene can be solved using sparse recovery methods.

[0049] Next, let's briefly discuss the range folding phenomenon: In pulse radar, when the target's distance from the radar exceeds the maximum distance corresponding to the pulse repetition period, the backscattered echo generated by the previous pulse encountering the target returns to the radar after the next pulse has already been emitted. This causes the radar to mistakenly interpret the target echo's time delay as the time interval between it and the start edge of the second pulse. The target echo will appear somewhere within the maximum detection range, and the measured target distance in this case is not the actual distance. This phenomenon is called range folding.

[0050] In the case of range folding, when there is only a single pulse repetition period sampling sample, the true distance of the folded echo has multiple possibilities. The above underdetermined equation has no unique solution, which will lead to equation solution errors. Therefore, it is necessary to solve the problem by using multiple sets of samples with mutually prime pulse repetition periods to eliminate the situation that does not conform to the actual scenario, such as Figure 2 shown.

[0051] Taking the sampling solution under two pulse repetition periods as an example, it is assumed that the pulse repetition periods of PRT1 and PRT2 are mutually prime in ratio a:b, and the maximum unambiguous distances are R a 、R b , and its sampling frequency is F s , then the unambiguous distance after expansion achieved by both is R m , and R m =aR b =bR a The unambiguous scene scattering feature vector σ=[σ R1 ,σ R2 ,σ R3 ,…,σ Rm ], which is an M×1-dimensional vector.

[0052] For a linear frequency modulation signal with a pulse repetition period of PRT1 (hereinafter referred to as signal 1), its transformation matrix Ψ a So that Φ a σ=s a (t) is:

[0053]

[0054] In formula (5), Ψ a It is an A×M dimensional matrix, where A is the pulse repetition period PRT1 and the sampling frequency is F s The number of distance dimension sampling points when A=T1·F s M indicates that the pulse repetition period is b·PRT1 and the sampling frequency is F s The number of distance dimension sampling points when M=T m ·F s . μ=B / T1 is the frequency modulation slope of signal 1.

[0055] For a linear frequency modulation signal with a pulse repetition period of PRT2 (hereinafter referred to as signal 2), its transformation matrix Ψ b So that Φ b σ=s b (t) is:

[0056]

[0057] In formula (6), Ψ b is a B×M dimensional matrix, where B is the pulse repetition period PRT2 and the sampling frequency is F s The number of distance dimension sampling points when B = T2·F s M indicates that the pulse repetition period is a·PRT2 and the sampling frequency is F s The number of distance dimension sampling points when M=T mFs. μ = B / T2 is the frequency modulation slope of signal 2.

[0058] From formula (5) and formula (6), the range unfolding equation of the linear frequency modulation radar demodulated frequency echo signal can be obtained as:

[0059]

[0060] After solving formula (7) using the SBL (Sparse Bayesian Learning) algorithm, the distance dimension unambiguous scattering feature vector σ is obtained. At this time, the long pulse signal model Φ corresponding to the distance dimension collapse is constructed as follows:

[0061]

[0062] s(t M )=Φ·σ Formula (9)

[0063] In formula (8), μ = B / T m is the frequency modulation slope of the long pulse signal, and Φ is an M×M dimensional matrix. Then, the time domain echo signal s(t M ).

[0064] However, the σ obtained by formula (9) is only the unambiguous scattering characteristic distribution after reconstructing the collected data of a single array element. For array signal processing, it is necessary to reconstruct the s collected by each array element. a (t), s b (t) The above distance unfolding equations are solved separately to finally obtain the echo data of all array elements.

[0065] SBL is a typical method for sparse signal reconstruction. It combines Bayesian thinking with sparsity, achieving performance comparable to reweighted l1-norm recovery without requiring a regularization parameter. After constructing the observation vector x and the observation basis matrix A, it uses adaptive adjustment to select the most relevant features from the high-dimensional data, gradually reducing the weights of unimportant, redundant features to zero, ultimately achieving deconstruction of the target data.

[0066] For the above equation When reconstructing the solution, you need to As the observation basis matrix A, the array element observation data is the observation vector, and σ is the unknown number to be solved.

[0067] Then the parameters are initialized so that all elements of the hyperparameter γ are 1 and the noise variance σ 2is 0.0001. It's important to note that for array signal processing, the sparse features of the entire scene captured by each array element are similar. Therefore, placing hyperparameter initialization outside the loop can improve iteration efficiency to a certain extent.

[0068] The mean μ and covariance matrix ∑ of the posterior probability are calculated as follows:

[0069]

[0070] In formula (10) and formula (11), Γ≡diag(γ), ∑ x =σ 2 I+AΓA H .

[0071] The EM (Expectation-Maximization) algorithm is used to adjust the hyperparameter γ and noise variance σ 2 Updates as follows:

[0072]

[0073] In formula (12), ∑ n,n are the diagonal elements of ∑, M and N are the number of rows and columns of matrix A. Repeat the calculation process of formulas (10) to (13) until the hyperparameter γ meets the iteration condition and converges to a fixed point γ * Or the number of iterations meets the preset conditions.

[0074] According to formula (10), the hyperparameter γ is calculated * The corresponding mean μ is the required unambiguous scattering feature vector σ in the range dimension. The effectiveness of this method in removing range folding can be verified by the measured data of linear frequency modulation radar correlation echoes.

[0075] Based on the echo analysis process of the linear frequency modulation radar echo range unfolding, combined with Figure 3 As shown, the embodiment of the present disclosure provides an echo analysis method for radar detection, comprising:

[0076] S01, the detection device detects the target scene through a linear frequency modulation signal with mutually prime pulse repetition periods, and obtains a demodulated time domain echo signal.

[0077] S02: The detection device constructs a distance-dimensional sparse observation model based on the demodulated time-domain echo signal.

[0078] S03, the detection device iteratively solves the distance dimension sparse observation model through the sparse Bayesian learning reconstruction algorithm, obtains the reconstructed target unambiguous distance dimension scattering feature vector, and constructs a long pulse signal model after the distance dimension is unfolded.

[0079] S04, the detection device obtains a time domain echo signal after range dimension unfolding based on the target's unambiguous range dimension scattering characteristic vector and the long pulse signal model.

[0080] Using the echo analysis method for radar detection provided by the embodiments of the present disclosure, the detection device first detects the target scene using a linear frequency modulation signal with mutually prime pulse repetition periods to obtain a demodulated time-domain echo signal and construct an echo model for the demodulated signal. The detection device then constructs a range-dimensional sparse observation model based on the demodulated time-domain echo signal. Because sparse recovery theory can utilize observation data from linear frequency modulation radars with staggered repetition frequencies to reconstruct the original echo signal power spectrum without distortion, it can separate the aliased single-pass and multi-pass target echoes, achieving echo range unfolding to deconstruct the target's original echo signal in the range dimension. Based on this, after constructing the distance-dimensional sparse observation model, the detection device iteratively solves the distance-dimensional sparse observation model through a sparse Bayesian learning reconstruction algorithm to obtain a reconstructed target unambiguous distance-dimensional scattering feature vector and constructs a long pulse signal model after generating distance-dimensional unfolding. Based on the target unambiguous distance-dimensional scattering feature vector and the long pulse signal model, a time-domain echo signal after distance-dimensional unfolding is obtained, so as to achieve echo distance unfolding through sparse recovery theory and ensure the reconstruction accuracy of the time-domain echo signal. The disclosed embodiment utilizes sparse recovery theory to complete the reconstruction of the time-domain echo signal on the basis of achieving echo distance unfolding, and maintain the reconstruction accuracy of the time-domain echo signal.

[0081] Furthermore, the random phase system used in conventional radars requires the generation of optimized random phase codes to control the transmitter phase shifter, further increasing the difficulty of system design and data analysis. However, the disclosed embodiments do not require the configuration of transmitter phase shifters, thus reducing the system design and data analysis difficulties to a certain extent.

[0082] Optionally, combined Figure 4 As shown, the detection device detects the target scene through a linear frequency modulation signal with mutually prime pulse repetition periods, and obtains a demodulated time domain echo signal, including:

[0083] S11, the detection device detects the target scene through a linear frequency modulation signal with the same signal bandwidth and mutually prime pulse repetition period, and obtains the echo signal s r (t).

[0084] In this step, the pulse repetition period of different linear frequency modulation signals is constructed as [PRT1, PRT 12 ,…,PRT N ]sequence.

[0085] S12, the detection device will echo the signal s r (t) and the reference signal Perform mixing processing to obtain the demodulated time domain echo signal.

[0086] In this step, the demodulated time domain echo signal is [s(t1),s(t2),…,s(t N )] sequence. Wherein, N represents the number of linear frequency modulation signals with the same signal bandwidth and mutually prime pulse repetition periods.

[0087] In the embodiment of the present disclosure, the detection device sends the echo signal s r (t) and the reference signal Perform mixing processing to obtain the demodulated time domain echo signal, including:

[0088] The detection device obtains the demodulated beat signal data model s corresponding to the scattering of any target point i (t) and the distance dimension sequence R. Where R=[0,R1,R2,…,R max ],s i (t) = f(R i )σ i and τ i =2R i / c, A0, T represent the initial signal amplitude of the transmitted pulse and the signal pulse width of the transmitted pulse, f0, μ represent the carrier frequency and the frequency modulation slope of the long pulse signal respectively, c represents the speed of light, σ i It represents the multiplicative coefficient of the scattering intensity of the i-th target point on the initial signal amplitude. The scattering intensity represents the scattering intensity value determined by the distance transmission loss of the i-th target point and the scattering characteristics of the target itself, the scattering cross-sectional area and the incident angle.

[0089] The detection equipment superimposes the demodulated beat signal data models corresponding to all target point scattering to obtain the demodulated time domain echo signal s(t). I represents the total number of scattered signals from all target points in the scene.

[0090] S13, the detection device determines that the lowest common multiple of the pulse repetition periods of all linear frequency modulation signals is the pulse repetition period T of the long pulse signal m , and determine c·T m / 2 is the unambiguous distance.

[0091] In this way, the detection device detects the target scene using linear frequency modulation signals with the same signal bandwidth and mutually prime pulse repetition periods, obtains the echo signal, and mixes the echo signal with the reference signal to obtain the demodulated time domain echo signal. Here, the detection device can also obtain the pulse repetition period of the long pulse signal corresponding to the unambiguous range extension based on the pulse repetition period of different linear frequency modulation signals as the least common multiple of all pulse repetition periods, and extend the unambiguous range of the long pulse signal to c·T m / 2.

[0092] Optionally, the range-dimensional sparse observation model includes Ψ·σ=s(t), where Ψ, σ, and s(t) represent the sparse recovery basis matrix and the reconstructed target unambiguous range-dimensional scattering feature vector and observation vector, respectively.

[0093] Combine Figure 5 As shown in the figure, the detection device iteratively solves the distance-dimensional sparse observation model through the sparse Bayesian learning reconstruction algorithm to obtain the reconstructed target unambiguous distance-dimensional scattering feature vector, including:

[0094] S21, the detection device iteratively solves the distance dimension sparse observation model through a sparse Bayesian learning reconstruction algorithm.

[0095] S22, the detection device uses the distance unfolding equation and initializes the hyperparameters γ and noise variance σ 2 , calculate the mean μ of the posterior probability.

[0096] In this step, taking the samples with two mutually prime pulse repetition periods as an example, the distance unfolding equation is: represents the observation basis matrix, represents the observation vector.

[0097] S23, the detection device updates the hyperparameter γ and noise variance σ 2 , and determine the target hyperparameter γ that meets the iteration termination condition * The corresponding target mean μ * According to the target mean μ * The unambiguous range-dimensional scattering feature vector σ of the reconstructed target is obtained.

[0098] In this step, the iteration termination conditions include: the hyperparameter γ converges or the number of iterations is greater than or equal to the upper limit of the number of iterations.

[0099] In this way, the detection device continuously iteratively solves the distance-dimensional sparse observation model using a sparse Bayesian learning reconstruction algorithm. It then calculates the mean of the posterior probability based on the distance unfolding equation and the initialization hyperparameters and noise variance, and then updates the hyperparameters and noise variance until the hyperparameters meet the iteration termination criteria. At this point, the detection device determines the target mean corresponding to the target hyperparameter that meets the iteration termination criteria and uses this mean to obtain the sparsely reconstructed target unambiguous distance-dimensional scattering feature vector.

[0100] Optionally, the detection device updates the hyperparameter γ and the noise variance σ 2 ,include:

[0101]

[0102] in, Γ≡diag(γ),∑ x =σ 2 I+AΓA H .

[0103] A represents the M×N observation basis matrix in the distance unfolding equation, ∑ n,n represents the ∑ diagonal elements.

[0104] Optionally, the detection device constructs a long pulse signal model after the range dimension is unfolded, including:

[0105] The detection device obtains the sampling frequency F of all linear frequency modulation signals s Sum signal bandwidth B and spatial propagation distance R of each linear frequency modulation signal i , carrier frequency f0.

[0106] The detection equipment determines the signal bandwidth B and the pulse repetition period T of the long pulse signal m The ratio is the frequency modulation slope μ of the long pulse signal.

[0107] The detection equipment constructs a long pulse signal model Φ. Among them, the long pulse signal model Φ is

[0108]

[0109] Thus, the pulse repetition period T of the long pulse signal corresponding to the unambiguous distance obtained in step 13 is m , construct and generate a long pulse signal model Φ, and multiply the long pulse signal model Φ with the range dimension unambiguous scattering vector σ obtained in step 23 to complete the time domain echo signal s(t M )reconstruction.

[0110] Optionally, the detection device obtains the time domain echo signal after the range dimension is unfolded according to the target unambiguous range dimension scattering characteristic vector and the long pulse signal model, including: the detection device obtains the time domain echo signal after the range dimension is unfolded according to s(tM )=Φ·σ and the long pulse signal model Φ is solved to obtain the time domain echo signal s(t M ).

[0111] In this way, by multiplying the long pulse signal model Φ with the range dimension unambiguous scattering vector σ obtained in step 23, the long pulse signal echo s (t M )reconstruction.

[0112] Optionally, the detection device constructs a distance-dimensional sparse observation model based on the demodulated time-domain echo signal, including:

[0113] The detection equipment will demodulate the time domain echo signal sequence [s(t1), s(t2), ..., s(t N )] are arranged vertically to form a one-dimensional column vector as the observation vector s(t).

[0114] The detection device uses the beat signal data model to construct a sparse recovery basis matrix Ψ and a distance dimension sparse observation model.

[0115] The distance-dimensional sparse observation model includes: Ψ·σ=s(t); Ψ represents a sparse recovery basis matrix composed of multiple sub-matrices arranged vertically based on a linear frequency modulation signal with mutually prime pulse repetition periods, σ represents the unambiguous distance-dimensional scattering feature vector of the reconstructed target, and multiple sub-matrices are expressed as [Ψ1,Ψ2,…,Ψ N ].

[0116] In the disclosed embodiment, the beat signal data model can refer to the above formula (3), that is, the beat signal data model is:

[0117]

[0118] Wherein, μ represents the frequency modulation slope of the long pulse signal, T represents the signal pulse width of the transmitted pulse, f0 represents the carrier frequency, A0 represents the initial signal amplitude of the transmitted pulse, and σ1 uniformly represents the range transmission loss of target 1 in the echo signal and the multiplication coefficient caused by the scattering intensity determined by the target's own scattering characteristics, scattering cross-sectional area, and incident angle on the initial signal amplitude A0.

[0119] In practical applications, the echo analysis method for radar detection specifically performs the following steps:

[0120] S101, respectively using the same bandwidth B and different pulse repetition period PRT linear frequency modulation signal to detect the target scene and collect the time domain echo signal, respectively the time domain echo signal s r (t) and the reference signal Mixing process, get the demodulated time domain echo signal [s(t1), s(t2), ..., s(tN )].

[0121] Here, according to different pulse repetition periods [PRT1, PRT2, ..., PRT N ], the pulse period PRT of the long pulse signal corresponding to the unambiguous distance expansion can be obtained M is the lowest common multiple of the pulse repetition period PRT, and its unambiguous distance extends to c·PRT M / 2.

[0122] S102, the demodulated time domain echo signal [s(t1), s(t2), ..., s(t N )] are arranged vertically to form a one-dimensional column vector to construct the observation vector s(t), and the beat signal data model is used to construct the sparse recovery basis matrix Ψ and the distance dimension sparse observation model Ψ·σ=s(t).

[0123] Where σ is the unambiguous range-dimensional scattering feature vector of the reconstructed target, and Ψ is a sub-matrix [Ψ1, Ψ2, ..., Ψ N ]The sparse recovery basis matrix is ​​arranged vertically.

[0124] S103, using the SBL reconstruction algorithm to continuously iteratively solve the distance-dimensional sparse observation model Ψ·σ=s(t) constructed in step 102 until the hyperparameters converge or the number of iterations is greater than or equal to the upper limit of the number of iterations, and obtain the reconstructed target unambiguous distance-dimensional scattering feature vector σ.

[0125] S104: The pulse period PRT of the long pulse signal corresponding to the unambiguous range extension obtained by calculation in step 101 M , construct a long pulse signal model Φ, and multiply the target unambiguous range-dimensional scattering feature vector σ obtained in step 103 by the long pulse signal model Φ to complete the time domain echo signal s(t M )reconstruction.

[0126] Based on the above steps, the present invention uses two linear frequency modulation signals with different pulse repetition periods to detect and simulate scene targets. The carrier frequency is 77.1 GHz, the pulse repetition period of signal 1 is 100, the pulse repetition period of signal 2 is 150, the signal bandwidth B is 800 MHz, and the sampling frequency is 2 MHz. The maximum unambiguous ranges for signal 1 are 37.5 m and 56.25 m, respectively, with 200 and 300 sampling points in the range dimension, respectively.

[0127] Set the target scattering scenario in the simulation as Figure 6As shown in the figure, it can be seen that the nearby "A" target is within the maximum unambiguous distance of signal 1 and signal 2, while the rectangular target is outside the maximum unambiguous distance of signal 1 and signal 2, resulting in range folding.

[0128] According to step 101, a one-dimensional array consisting of 100 elements acquires two echo signals for the above scenario and performs frequency demodulation processing. It can be seen that the maximum unambiguous range after range dimension expansion is increased to 112.5 meters, and the corresponding long pulse signal has a pulse repetition period of 300. Then, according to step 102, an observation basis matrix and a range-dimensional sparse observation model are constructed based on the demodulated frequency signal model.

[0129] According to the method of step 103, the SBL reconstruction algorithm is used to solve the distance-dimensional scattering characteristic vector of the signals collected by the 100 array elements respectively, and finally according to step 104, the time domain echo signal after the range unfolding detected by the long pulse signal is reconstructed.

[0130] Figure 7 and Figure 8 They are respectively the range dimension amplitude response of the echo signal after range unfolding by the present invention and the azimuth dimension imaging result of the DBF algorithm. Figure 7 and Figure 8 As shown in the range-dimensional spectrum and imaging results, the distant rectangular target echo and the nearby "A"-shaped echo are effectively separated, and the range folding is completely eliminated.

[0131] Combine Figure 9 As shown, an embodiment of the present disclosure provides an echo analysis device 70 for radar detection, including a processor 700 and a memory 701. Optionally, the device 70 may further include a communication interface 702 and a bus 703. The processor 700, the communication interface 702, and the memory 701 may communicate with each other via the bus 703. The communication interface 702 may be used for information transmission. The processor 700 may call the logic instructions in the memory 701 to execute the echo analysis method for radar detection of the above embodiment.

[0132] In addition, the logic instructions in the memory 701 can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product.

[0133] Memory 701, as a computer-readable storage medium, can be used to store software programs and computer-executable programs, such as program instructions / modules corresponding to the methods in the embodiments of the present disclosure. Processor 700 executes the program instructions / modules stored in memory 701 to perform functional applications and data processing, thereby implementing the echo analysis method for radar detection in the above-described embodiments.

[0134] The memory 701 may include a program storage area and a data storage area. The program storage area may store an operating system and at least one application required for a function; the data storage area may store data generated based on the use of the terminal device. Furthermore, the memory 701 may include high-speed random access memory and non-volatile memory.

[0135] The embodiment of the present disclosure provides a detection device, comprising: a detection device body, and the above-mentioned echo analysis device 70 for radar detection. The echo analysis device 70 for radar detection is installed on the detection device body. The installation relationship described here is not limited to placement inside the detection device body, but also includes installation connections with other components of the detection device, including but not limited to physical connections, electrical connections or signal transmission connections. It can be understood by those skilled in the art that the echo analysis device 70 for radar detection can be adapted to a feasible detection device body, thereby realizing other feasible embodiments.

[0136] An embodiment of the present disclosure provides a computer-readable storage medium storing computer-executable instructions, wherein the computer-executable instructions are configured to execute the above-mentioned echo analysis method for radar detection.

[0137] The technical solutions of the embodiments of the present disclosure may be embodied in the form of a software product, which is stored in a storage medium and includes one or more instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in the embodiments of the present disclosure. The aforementioned storage medium may be a non-transitory storage medium, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, among other media capable of storing program code.

[0138] The above description and the accompanying drawings fully illustrate the embodiments of the present disclosure so that those skilled in the art can practice them. Other embodiments may include structural, logical, electrical, process and other changes. The embodiments represent only possible variations. Unless explicitly required, individual components and functions are optional, and the order of operations may vary. Parts and features of some embodiments may be included in or replace parts and features of other embodiments. Moreover, the words used in this application are only used to describe the embodiments and are not used to limit the claims. As used in the description of the embodiments and claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to also include plural forms. Similarly, the term "and / or" as used in this application refers to any and all possible combinations of one or more associated listings. In addition, when used in this application, the term "comprise" and its variations "comprises" and / or comprising refer to the presence of stated features, wholes, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or groups of these. In the absence of further restrictions, an element defined by the sentence "comprising a..." does not exclude the presence of other identical elements in the process, method or device that includes the element. In this article, each embodiment may focus on the differences from other embodiments, and the same and similar parts between the various embodiments can be referenced to each other. For the methods, products, etc. disclosed in the embodiments, if they correspond to the method part disclosed in the embodiments, then the relevant parts can be found in the description of the method part.

[0139] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software may depend on the specific application and design constraints of the technical solution. The technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the embodiments of the present disclosure. The technicians will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

Claims

1. An echo analysis method for radar detection, characterized in that: include: The target scene is detected by using a linear frequency modulation signal with mutually prime pulse repetition periods, and the demodulated time domain echo signal is obtained; According to the demodulated time domain echo signal, a distance dimension sparse observation model is constructed; The sparse Bayesian learning reconstruction algorithm is used to iteratively solve the distance-dimensional sparse observation model to obtain the reconstructed target unambiguous distance-dimensional scattering feature vector and construct a long pulse signal model after range-dimensional unfolding. According to the target's unambiguous range-dimensional scattering feature vector and the long pulse signal model, the time-domain echo signal after range-dimensional unfolding is obtained.

2. The method according to claim 1, characterized in that The target scene is detected by using a linear frequency modulation signal with mutually prime pulse repetition periods, and the demodulated time domain echo signal is obtained, including: The target scene is detected by using linear frequency modulation signals with the same signal bandwidth and mutually prime pulse repetition periods to obtain echo signals; The echo signal is mixed with the reference signal to obtain a demodulated time domain echo signal; Determine the lowest common multiple of the pulse repetition periods of all linear frequency modulation signals as the pulse repetition period T of the long pulse signal m , and determine c·T m / 2 is the unambiguous distance.

3. The method according to claim 2, characterized in that The range-dimensional sparse observation model includes Ψ·σ=s(t); where Ψ, σ, and s(t) represent the sparse recovery basis matrix and the reconstructed target unambiguous range-dimensional scattering feature vector and observation vector, respectively; The distance-dimensional sparse observation model is iteratively solved by the sparse Bayesian learning reconstruction algorithm to obtain the reconstructed target unambiguous distance-dimensional scattering feature vector, including: The distance dimension sparse observation model is iteratively solved through the sparse Bayesian learning reconstruction algorithm; Calculate the mean μ of the posterior probability according to the distance unfolding equation and the initialization hyperparameters and noise variance; The hyperparameters and noise variance are updated, and the target mean corresponding to the target hyperparameters that meet the iteration termination conditions is determined to obtain the reconstructed target unambiguous distance-dimensional scattering feature vector according to the target mean.

4. The method according to claim 3, characterized in that Update hyperparameters γ and noise variance σ 2 ,include: Among them, Γ≡diag(γ),Σ x =s 2 I+AGA H ; A represents the M×N observation basis matrix in the distance unfolding equation, Σ n,n denotes the diagonal elements of Σ.

5. The method according to claim 2, characterized in that Construct a long pulse signal model after distance dimension unfolding, including: Get the sampling frequency F of all linear FM signals s Sum signal bandwidth B and spatial propagation distance R of each linear frequency modulation signal i , carrier frequency f0; Determine the signal bandwidth B and the pulse repetition period T of the long pulse signal m The ratio is the frequency modulation slope μ of the long pulse signal; Construct a long pulse signal model Φ; where the long pulse signal model Φ is 6. The method according to claim 5, characterized in that According to the target's unambiguous range-dimensional scattering feature vector and the long pulse signal model, the time-domain echo signal after range-dimensional unfolding is obtained, including: According to s(t M )=Φ·σ and the long pulse signal model Φ is solved to obtain the time domain echo signal s(t M ).

7. The method according to any one of claims 1 to 6, characterized in that According to the demodulated time domain echo signal, a distance dimension sparse observation model is constructed, including: The demodulated time domain echo signal sequence [s(t1),s(t2),…,s(t N )] are arranged vertically to form a one-dimensional column vector as the observation vector s(t); The beat signal data model is used to construct a sparse recovery basis matrix Ψ and a distance dimension sparse observation model; The distance-dimensional sparse observation model includes: Ψ·σ=s(t); Ψ represents a sparse recovery basis matrix composed of multiple sub-matrices arranged vertically based on a linear frequency modulation signal with mutually prime pulse repetition periods, σ represents the unambiguous distance-dimensional scattering feature vector of the reconstructed target, and multiple sub-matrices are expressed as [Ψ1,Ψ2,…,Ψ N ].

8. An echo analysis device for radar detection, comprising a processor and a memory storing program instructions, characterized in that: The processor is configured to execute the echo analysis method for radar detection according to any one of claims 1 to 7 when running the program instructions.

9. A detection device, characterized in that: include: Detection equipment body; The echo analysis device for radar detection according to claim 8 is installed on the detection equipment body.

10. A computer-readable storage medium storing program instructions, characterized in that: When the program instructions are executed, the computer is configured to execute the echo analysis method for radar detection according to any one of claims 1 to 7.

Images