Method and system for suppressing radar interference signal, and vehicle-mounted radar system

Abstract

The invention provides a method for suppressing a radar interference signal. The method comprises the following steps: modeling a mixed signal yL received by a radar; carrying out parallel calculation by using continuous wavelet transform; normalizing the sub-wavelet function psi (t / s) on each scale s into a form with unit energy; selecting continuous wavelet transform domains of different scales; and optimizing the target function f (c) by using a split augmented Lagrange contraction algorithm, and converging the target function to an optimal value by using the split augmented Lagrange contraction algorithm, thereby determining transform domain coefficients c1 and c2. According to the method, the interference signals can be more effectively separated and recovered, and explicit detection and identification are not needed. Therefore, the performance of an interference suppression algorithm can be effectively improved, the performance and reliability of a radar system are improved, and more accurate and reliable information is provided for an advanced driving assistance system.

Description

Technical Field

[0001] The present invention relates to the field of radar signal processing, and in particular to a method and system for suppressing radar interference signals. Background Art

[0002] Considering that when the radar receives a signal, the environmental factors cause the signal received by the radar to usually include the target signal and the interference signal, so it is necessary to suppress the interference signal as much as possible from the received signal. In the prior art, the interference suppression problem is mainly handled by traditional algorithms, compressed sensing algorithms and artificial intelligence-based methods. However, traditional algorithms such as ZF and LPC have low interference detection rates under low signal-to-noise ratio conditions, while compressed sensing algorithms have high sparsity requirements and high computational complexity. Although the interference suppression algorithm based on artificial intelligence has good performance, its application is limited due to the lack of sufficient real data support.

[0003] In order to solve this problem, the closest tunable Q-factor wavelet transform (TQWT)-morphological component analysis (MCA) algorithm is used in the prior art. However, its limitation is that when dealing with interference suppression problems, due to the serial calculation method of TQWT, a high calculation time is required and it cannot meet the real-time requirements.

[0004] Therefore, there is an urgent need in the art for a method for suppressing radar interference signals that can improve the overall operating speed of a radar signal interference suppression algorithm to meet real-time requirements. Summary of the invention

[0005] In view of the above defects of the prior art, the present invention proposes a method for suppressing radar interference signals based on MCA combined with fast continuous wavelet transform (CWT), which solves the problems of high real-time performance and high computational complexity of the interference suppression algorithm in the prior art.

[0006] Specifically, the present invention discloses a method for suppressing radar interference signals, comprising:

[0007] The mixed signal y received by the radar L Modeling: The mixed signal y L Including the target signal y r and interference signal y i , where the target signal y r and interference signal y i The transformation matrices of the sparse domain are denoted as A1 and A2 respectively, and the mixed signal y L The modeling is expressed as: yL =A1c1+A2c2, thereby determining the optimization problem of c1 and c2;

[0008] Use continuous wavelet transform to parallelize computation:

[0009]

[0010] Wherein, δt is the sampling rate of the radar's transmit waveform, is the sampling interval in the time domain, ψ(t / s) is the sub-wavelet function obtained by scaling the base wavelet function ψ0(t), N is the duration of the radar's transmit waveform, is the length of the discrete sequence x(n), and W(n, s) is the convolution of the discrete sequence x(n) with the base wavelet function ψ0(t) after scaling and translation;

[0011] Normalize the sub-wavelet function ψ(t / s) at each scale s to a form with unit energy;

[0012] Selecting different scales of continuous wavelet transform domains: the scales of the two different scales of continuous wavelet transform are s1 and s2 respectively.

[0013] s1=s0·2 nj

[0014] s2=s0·2 mj

[0015] Wherein, s0 is a constant, j is a vector composed of an integer sequence, the dimension of the vector is equal to the number of subbands of the continuous wavelet transform, m and n are large-scale coefficients and small-scale coefficients respectively; according to the different characteristics of the target signal and the interference signal in the large-scale and small-scale continuous wavelet transforms, the large-scale continuous wavelet transform domain and the small-scale continuous wavelet transform domain are respectively selected as the two transform domains of the morphological component analysis algorithm;

[0016] The objective function f(c) is optimized using the split augmented Lagrangian contraction algorithm: The objective function f(c) is:

[0017]

[0018] in, A1 and A2 are the mixed signals y L Transformation matrices in the large-scale continuous wavelet transform domain and the small-scale continuous wavelet transform domain;

[0019] The objective function is converged to an optimal value using a split augmented Lagrangian shrinkage algorithm, thereby determining transform domain coefficients c1 and c2.

[0020] Preferably, the mixed signal y received by the radar L Modeling includes:

[0021] The mixed signal y L By the target signal y r and interference signal y i The vector form is:

[0022] y L =y r +y i Formula (1);

[0023] The target signal y r and interference signal y i The transformation matrices of the sparse domain are A1 and A2 respectively, and the mixed signal y L It is expressed by formula (2):

[0024] y L =A1c1+A2c2 formula (2),

[0025] Among them, the optimization problem of c1 and c2 is defined as:

[0026]

[0027] st||y L -A1c1-A2c2||2≤σ,

[0028] Among them, ρ1 and ρ2 are weight coefficients, and σ is the noise tolerance.

[0029] Preferably, using continuous wavelet transform parallel computing includes:

[0030] The continuous wavelet transform of a discrete sequence x(n) of length N is defined as the convolution of the discrete sequence x(n) with the base wavelet after scaling and translation, as shown in formula (3):

[0031]

[0032] Wherein, the discrete sequence x(n) is the mixed signal y L The sequence after discrete sampling, δt is the sampling rate of the radar's transmit waveform, is the sampling interval in the time domain, ψ(t / s) is the sub-wavelet function, which is obtained by scaling the base wavelet function ψ0(t), N is the duration of the radar's transmit waveform, is the length of the discrete sequence x(n), n and n' are the time index and delay of the discrete time sequence, respectively, and W(n, s) is the convolution of the discrete sequence x(n) with the base wavelet function ψ0(t) after scaling and translation.

[0033] Preferably, the continuous wavelet transform parallel calculation is optimized using fast Fourier transform, comprising:

[0034] The discrete Fourier transform of the discrete sequence x(n)) is expressed as

[0035] Where k=0,...,N-1 is the frequency index;

[0036] According to the scaling property of Fourier transform, the Fourier transform of sub-wavelet function ψ(t / s) corresponds to We can get formula (4):

[0037]

[0038] The angular frequency in formula (4) is defined as

[0039] Preferably, normalizing the sub-wavelet functions at each scale s to a form having unit energy comprises:

[0040] For each of the dimensions s, the following equation is satisfied:

[0041]

[0042] According to formula (5), the normalized sub-wavelet function is obtained:

[0043]

[0044] Wherein, the base wavelet function ψ0(t) is a Morlet wavelet function, and the frequency of the Morlet wavelet function is:

[0045]

[0046] Where S(ω) represents the step function;

[0047] The normalized sub-wavelet function is introduced into the formula (3) for calculation.

[0048] Preferably, the selecting of two transform domains of different scales comprises:

[0049] The target signal y r In the large-scale continuous wavelet transform domain, the interference signal y is sparser. i It is even sparser in the continuous wavelet transform domain at small scales.

[0050] Preferably, the optimization of the objective function f(c) using the split augmented Lagrangian contraction algorithm comprises:

[0051] The objective function f(c) is optimized step by step, and the variable μ is introduced. Then the optimization problem of c1 and c2 is:

[0052]

[0053] stc-μ=0

[0054] The split augmented Lagrangian contraction algorithm is used to solve the problem, and μ and c are updated respectively, so that the objective function gradually converges to the optimal value. After k iterative calculations, the value of c is obtained, and c1 and c2 are calculated accordingly.

[0055] Preferably, the updating of μ and c respectively includes: introducing an intermediate variable d of the kth iteration k ∈C 2N×1 :

[0056] S1: Calculate the current iteration c k The updated value of is recorded as:

[0057]

[0058] in ξ is 0.04;

[0059] S2: Get c k After the update result, update μ k , denoted as

[0060] μ k =S λ / c (c k +d k-1 ) formula (7),

[0061] where λ = [λ1, λ2] T , that is, in μ k The upper part of is filtered by λ1, μ k The lower part of is filtered by λ2;

[0062] S3: for the intermediate variable d k The value of is updated, recorded as:

[0063] d k =d k-1 +c k -μ k Formula (8);

[0064] After 100 iterations, the final result c is obtained. After obtaining the transform domain coefficients c1 and c2, the optimization problem of c1 and c2 is introduced to determine the target signal y r and interference signal y i .

[0065] Another aspect of the present invention provides a system for suppressing radar interference signals, which uses any of the above methods for suppressing radar interference signals.

[0066] Another aspect of the present invention provides a vehicle-mounted radar system, comprising the above-mentioned system for suppressing radar interference signals.

[0067] Compared with the TQWT-MCA algorithm in the prior art, the wavelet continuous transform used in the present invention is optimized and can be parallelized, and can also be accelerated by fast Fourier transform, thereby greatly reducing the computational complexity and meeting the real-time requirements. The method of the present invention can more effectively separate and recover interference signals without explicit detection and identification. This can effectively improve the performance of the interference suppression algorithm, improve the performance and reliability of the radar system, and provide more accurate and reliable information for the advanced driver assistance system (ADAS). BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 is a flow chart of a method for suppressing radar interference signals in accordance with an embodiment of the present invention;

[0069] Figure 2 A schematic diagram of sub-band energy distribution of a received signal in a method for suppressing radar interference signals in accordance with an embodiment of the present invention;

[0070] Figure 3 for Figure 2 Schematic diagram of interference signals and target signals separated in the embodiment;

[0071] Figure 4 for Figure 2 Schematic diagram of the comparison of distance spectrum reconstruction performance between the embodiment and the linear prediction algorithm;

[0072] Figure 5 for Figure 2 Schematic diagram of the mean square error simulation results of the embodiment and the linear prediction algorithm. DETAILED DESCRIPTION

[0073] The advantages of the present invention are further described below in conjunction with the accompanying drawings and specific embodiments.

[0074] The terms used in this disclosure are for the purpose of describing specific embodiments only and are not intended to limit the disclosure. The singular forms of "a", "said" and "the" used in this disclosure and the appended claims are also intended to include plural forms unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more associated listed items.

[0075] It should be understood that although the terms first, second, third, etc. may be used in the present disclosure to describe various information, such information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of the present disclosure, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0076] In the following description, the suffixes such as "module", "component" or "unit" used to represent elements are only used to facilitate the description of the present invention, and have no specific meanings. Therefore, "module" and "component" can be used interchangeably.

[0077] An embodiment of the present invention provides a method for suppressing radar interference signals, such as Figure 1 As shown, including:

[0078] S1: mixed signal y received by the radar L Modeling: The mixed signal y L Including the target signal y r and interference signal y i , where the target signal y r and interference signal y i The transformation matrices of the sparse domain are denoted as A1 and A2 respectively, and the mixed signal y L The modeling is expressed as: y L =A1c1+A2c2, thereby determining the optimization problem of c1 and c2.

[0079] In this step, based on the target signal y r and interference signal y i There are different sparsities in different transformation domains. The transformation matrices in two transformation domains with different sparsities are recorded as A1 and A2, thereby constructing the mixed signal y L model and define the optimization problem accordingly.

[0080] S2: Parallel computation using continuous wavelet transform:

[0081]

[0082] Wherein, δt is the sampling rate of the radar's transmit waveform, is the sampling interval in the time domain, ψ(t / s) is the sub-wavelet function obtained by scaling the base wavelet function ψ0(t), N is the duration of the radar's transmit waveform, is the length of the discrete sequence x(n), and W(n, s) is the convolution of the discrete sequence x(n) with the base wavelet function ψ0(t) after scaling and translation.

[0083] In this step, the continuous wavelet transform can realize parallel computing and can be accelerated by optimization, thereby improving the overall running speed of the interference suppression algorithm to meet the real-time requirements.

[0084] S3: Normalize the sub-wavelet function ψ(t / s) at each scale s to a form with unit energy.

[0085] In this step, in order to ensure the direct comparability between the wavelet transforms at each scale s and the comparability with the transforms of other time series, the sub-wavelet function at each scale s needs to be normalized to a form with unit energy. Before wavelet transform, the sub-wavelet function needs to be normalized first, and the normalized sub-wavelet function is used to process the discrete sequence x(n).

[0086] S4: Selecting continuous wavelet transform domains of different scales: the scales of the two continuous wavelet transforms of different scales are s1 and s2 respectively.

[0087] s1=s0·2 nj

[0088] s2=s0·2 mj

[0089] Among them, s0 is a constant, j is a vector composed of an integer sequence, the dimension of the vector is equal to the number of sub-bands of the continuous wavelet transform, m and n are the large-scale coefficient and the small-scale coefficient, respectively; according to the different characteristics of the target signal and the interference signal in the large-scale and small-scale continuous wavelet transforms, the large-scale continuous wavelet transform domain and the small-scale continuous wavelet transform domain are respectively selected as the two transform domains of the morphological component analysis algorithm.

[0090] Since the necessary condition for the realization of CWT-MCA is that the target signal and the interference signal have different sparse characteristics in different variation domains, for CWT, different continuous wavelet transform domains can be obtained by selecting different scale ranges.

[0091] S5: Optimize the objective function f(c) using the split augmented Lagrangian contraction algorithm: The objective function f(c) is:

[0092]

[0093] in, A1 and A2 are the mixed signals y L Transformation matrices in the large-scale continuous wavelet transform domain and the small-scale continuous wavelet transform domain;

[0094] The objective function is converged to an optimal value using a split augmented Lagrangian shrinkage algorithm, thereby determining transform domain coefficients c1 and c2.

[0095] Based on the above embodiments, this embodiment represents the mixed signal received by the radar as the coefficients of the target signal and the interference signal in different continuous wavelet transform domains, constructs an optimization problem, and solves the optimal solution of the objective function to achieve accurate separation and filtering of the target signal and the interference signal. The split augmented Lagrangian shrinkage algorithm is used for optimization, which effectively solves the optimization problem in the continuous wavelet transform domain.

[0096] In another embodiment of the present invention, the method for suppressing radar interference signals includes:

[0097] S1: mixed signal y received by the radar L Modeling includes:

[0098] The mixed signal y L By the target signal y r and interference signal y i The vector form is:

[0099] y L =y r +y i Formula (1);

[0100] The target signal y r and interference signal y i The transformation matrices of the sparse domain are A1 and A2 respectively, and the mixed signal y L It is expressed by formula (2):

[0101] y L =A1c1+A2c2 formula (2),

[0102] Among them, the optimization problem of c1 and c2 is defined as:

[0103]

[0104] st||y L -A1c1-A2c2||2≤σ,

[0105] Among them, ρ1 and ρ2 are weight coefficients, and σ is the noise tolerance.

[0106] This step constructs an optimization problem. By solving the above optimization problem, c1 and c2 are solved, and then y is solved. r and i Among them, the weight coefficients ρ1 and ρ2 are equal, both are 0.1.

[0107] S2: The continuous wavelet transform of a discrete sequence x(n) of length N is defined as the convolution of the discrete sequence x(n) with the base wavelet after scaling and translation, as shown in formula (3):

[0108]

[0109] Wherein, the discrete sequence x(n) is the mixed signal y L The sequence after discrete sampling.

[0110] δt is the sampling rate of the radar's transmitted waveform, is the sampling interval in the time domain, ψ(t / s) is the sub-wavelet function, which is obtained by scaling the base wavelet function ψ0(t)), N is the duration of the radar's transmitted waveform, is the length of the discrete sequence x(n), n and n' are the time index and delay of the discrete time series, respectively; the above parameters can all be determined based on the actual parameters of the radar.

[0111] W(n, s) is the convolution of the discrete sequence x(n) and the basis wavelet function ψ0(t) after scaling and translation.

[0112] In this embodiment, the continuous wavelet transform (CWT) is further accelerated and optimized by using the fast Fourier transform (FFT), specifically including:

[0113] S21: The discrete Fourier transform of the discrete sequence x(n) is expressed as

[0114] Where k=0,...,N-1 is the frequency index;

[0115] According to the scaling property of Fourier transform, the Fourier transform of sub-wavelet function ψ(t / s) corresponds to We can get formula (4):

[0116]

[0117] The angular frequency in formula (4) is defined as

[0118] S3: Normalizing the sub-wavelet functions at each scale s to a form with unit energy includes:

[0119] For each of the dimensions s, the following equation is satisfied:

[0120]

[0121] According to formula (5), the normalized sub-wavelet function is obtained:

[0122]

[0123] Wherein, the base wavelet function ψ0(t) is a Morlet wavelet function, and the frequency of the Morlet wavelet function is:

[0124]

[0125] Where S(ω) represents the step function;

[0126] The normalized sub-wavelet function is introduced into the formula (3) for calculation.

[0127] To ensure the direct comparability between continuous wavelet transforms at each scale s and with the transforms of other time series, it is necessary to normalize the sub-wavelet functions at each scale s to a form with unit energy.

[0128] S4: The scales of the two different-scale continuous wavelet transforms are s1 and s2 respectively.

[0129] s1=s0·2 nj

[0130] s2=s0·2 mj

[0131] Wherein, s0 is a constant, j is a vector composed of an integer sequence, the dimension of the vector is equal to the number of subbands of the continuous wavelet transform, m and n are large-scale coefficients and small-scale coefficients respectively; according to the different characteristics of the target signal and the interference signal in the large-scale and small-scale continuous wavelet transforms, the large-scale continuous wavelet transform domain and the small-scale continuous wavelet transform domain are respectively selected as the two transform domains of the morphological component analysis algorithm;

[0132] According to experience, the target signal y r In the large-scale continuous wavelet transform domain, the interference signal y is sparser. i It is sparser in the small-scale continuous wavelet transform domain. According to the different characteristics of the two signals in the large-scale and small-scale CWT, the large-scale CWT transform and the small-scale CWT transform are selected as the two transform domains of the MCA algorithm. In this embodiment, m=0.25 and n=1 are selected, such as Figure 2 As shown, the energy distribution of the target signal and the interference signal contained in the mixed signal in the sub-band when the large-scale and small-scale continuous wavelet transform are selected respectively.

[0133] S5: Optimize the objective function f(c) using the split augmented Lagrangian contraction algorithm, including:

[0134] The objective function f(c) is:

[0135]

[0136] in, A1 and A2 are the mixed signals y L Transformation matrices in the large-scale continuous wavelet transform domain and the small-scale continuous wavelet transform domain;

[0137] The objective function is mainly divided into two terms, one is f1(c), which represents the noise tolerance, that is, σ in the optimization problem; the other is f2(c), which represents the sparsity of the two transform domain coefficients. In the actual calculation process, it is difficult to optimize f1(c) and f2(c) at the same time. In order to simplify the problem, a new variable μ is introduced by segmentation. This new variable μ is the same as c in form and is introduced for step-by-step optimization.

[0138] Then the optimization problem of c1 and c2 is expressed as:

[0139]

[0140] stc-μ=0

[0141] Update μ and c separately, including: introducing the intermediate variable d of the kth iteration k ∈C 2N×1 :

[0142] S51: Calculate the current iteration c k The updated value of is recorded as:

[0143]

[0144] in ξ is 0.04;

[0145] S52: Get c k After the update result, update μ k , denoted as

[0146] μ k =S λ / c (c k +d k-1 ) formula (7),

[0147] where λ = [λ1, λ2] T , that is, in μ k The upper part of is filtered by λ1, μ k The lower part of is filtered by λ2;

[0148] S53: For the intermediate variable d k The value of is updated, recorded as:

[0149] d k =d k-1 +ck -μ k Formula (8);

[0150] After k iterations, the final result c is obtained. After obtaining the transform domain coefficients c1 and c2, the optimization problem of c1 and c2 is introduced to determine the target signal y r and interference signal y i In this embodiment, the final result c is obtained by iterating 100 times. After the transform domain coefficients c1 and c2 are obtained, the separation result of the target signal and the interference signal can be obtained through the corresponding domain transform, thereby achieving the suppression of the interference signal.

[0151] The test results of this embodiment are as follows Figure 3-5 shown.

[0152] First, the beat time domain sampling point is set to 512. In this embodiment, the interference coefficient of the interference signal is 0.5, and the interference coefficient is defined as the ratio of the sweep slope of the interference signal to the target echo signal. Since the center frequency deviation only affects the position of the interference strip area, the center frequency deviation of the interference signal from the target echo signal is set to 0.

[0153] The reflection amplitude of the target signal in this embodiment is a random number from 1 to 100, the duration of a snapshot is 1.2×10^(-5)s, the sweep bandwidth is 360MHz, and the sweep slope is 3×10^13Hz / s. The number of repetitions is 1000 times, and the signal-to-noise ratio ranges from -20dB to 20dB. The signal-to-interference ratio is 5dB, and the scales s1 and s2 of the two transform domains are 0.25 and 1 respectively.

[0154] Figure 3 The received signal is given when the signal-to-interference ratio (SIR) is 15dB and the signal-to-noise ratio (SNR) is 10dB. After the method provided in this embodiment, the interference signal and the target signal are separated. Figure 3 It can be seen that the portion of the received signal that is significantly interfered with is separated, and the interference signal component contained in the target signal is significantly reduced.

[0155] Figure 4 The performance comparison of the distance spectrum reconstruction between the method shown in this embodiment and the linear prediction algorithm is given when the signal-to-noise ratio is 10dB. The target real distance is set to 26m. The curve with the legend "Signal with interference" in the figure represents the distance spectrum before interference suppression, which is used for comparison. Figure 4 It can be seen that when SNR=10 dB, the method of the present invention can reduce the background noise, while the linear prediction algorithm cannot suppress interference because the distance spectrum does not show a single peak.

[0156] Figure 5The mean square error simulation results of the method of this embodiment and the linear prediction algorithm when the signal-to-interference ratio is 5dB are given. Figure 5 As shown in the figure, the curve with the legend "Signal without interference" represents the situation without interference signal, which is used for comparison to show the performance of the two interference suppression algorithms. Figure 5 The simulation results show that when SIR=5dB, the mean square error of the method provided in this embodiment decreases the fastest. When the signal-to-noise ratio is 0dB, the mean square error of the method of the present invention is close to the ideal state and has good performance. Under high signal-to-noise ratio conditions, the linear prediction algorithm can also be reduced to a lower mean square error value.

[0157] comprehensive Figure 3-Figure 5 The simulation test results show that the method for suppressing radar interference signals provided in this embodiment has good performance. Under high signal-to-noise ratio conditions, the linear prediction algorithm can also be reduced to a lower mean square error value, but it depends on a very high signal-to-noise ratio.

[0158] In another embodiment of the present invention, a system for suppressing radar interference signals and a vehicle-mounted radar system containing the system are provided. The technical features of the system are the same as the technical features of the above-mentioned method for suppressing radar interference signals, and the present invention will not be elaborated here.

[0159] It should be noted that the embodiments of the present invention have better practicability and do not impose any form of limitation on the present invention. Any technician familiar with the field may use the technical content disclosed above to change or modify it into an equivalent effective embodiment. However, any modification or equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. A method for suppressing radar interference signals, characterized in that: include: The mixed signal y received by the radar L Modeling: The mixed signal y L Including the target signal y r and interference signal y i , where the target signal y r and interference signal y i The transformation matrices of the sparse domain are denoted as A1 and A2 respectively, and the mixed signal y L The modeling is expressed as: y L =A1c1+A2c2, thereby determining the optimization problem of c1 and c2; Use continuous wavelet transform to parallelize computation: Wherein, δt is the sampling rate of the radar's transmit waveform, is the sampling interval in the time domain, ψ(t / s) is the sub-wavelet function obtained by scaling the base wavelet function ψ0(t), N is the duration of the radar's transmit waveform, is the length of the discrete sequence x(n), and W(n, s) is the convolution of the discrete sequence x(n) with the base wavelet function ψ0(t) after scaling and translation; Normalize the sub-wavelet function ψ(t / s) at each scale s to a form with unit energy; Selecting different scales of continuous wavelet transform domains: the scales of the two different scales of continuous wavelet transform are s1 and s2 respectively. s1=s0·2 n) <h2 style=";text-align:left;direction:ltr">s2 = s0·2<h2 style=";text-align:left;direction:ltr"> m <h2 style=";text-align:left;direction:ltr"> ) Wherein, s0 is a constant, j is a vector composed of an integer sequence, the dimension of the vector is equal to the number of subbands of the continuous wavelet transform, m and n are large-scale coefficients and small-scale coefficients respectively; according to the different characteristics of the target signal and the interference signal in the large-scale and small-scale continuous wavelet transforms, the large-scale continuous wavelet transform domain and the small-scale continuous wavelet transform domain are respectively selected as the two transform domains of the morphological component analysis algorithm; The objective function f(c) is optimized using the split augmented Lagrangian contraction algorithm: The objective function f(c) is: in, A1 and A2 are the mixed signals y L Transformation matrices in the large-scale continuous wavelet transform domain and the small-scale continuous wavelet transform domain; The objective function is converged to an optimal value using a split augmented Lagrangian shrinkage algorithm, thereby determining transform domain coefficients c1 and c2.

2. The method for suppressing radar interference signals according to claim 1, characterized in that: The mixed signal y received by the radar L Modeling includes: The mixed signal y L By the target signal y r and interference signal y i The vector form is: y L = y r +y i Formula (1): The target signal y r and interference signal y i The transformation matrices of the sparse domain are A1 and A2 respectively, and the mixed signal y L It is expressed by formula (2): y L = A1c1+A2c2 Formula (2), Among them, the optimization problem of c1 and c2 is defined as: s.t.||y L -A1c1-A2c2||2≤σ, Among them, ρ1 and ρ2 are weight coefficients, and σ is the noise tolerance.

3. The method for suppressing radar interference signals as claimed in claim 2, characterized in that: Parallel computation using continuous wavelet transform includes: The continuous wavelet transform of a discrete sequence x(n) of length N is defined as the convolution of the discrete sequence x(n) with the base wavelet after scaling and translation, as shown in formula (3): Wherein, the discrete sequence x(n) is the mixed signal y L The sequence after discrete sampling, δ t is the sampling rate of the radar's transmitting waveform, is the sampling interval in the time domain, ψ(t / s) is the sub-wavelet function obtained by scaling the base wavelet function ψ0(t), N is the duration of the radar's transmitting waveform, is the length of the discrete sequence x(n), n and n' are the time index and delay of the discrete time series, respectively, and W(n, s) is the convolution of the discrete sequence x(n) with the base wavelet function ψ0(t) after scaling and translation.

4. The method for suppressing radar interference signals as claimed in claim 3, characterized in that: The continuous wavelet transform parallel calculation is optimized using fast Fourier transform, including: The discrete Fourier transform of the discrete sequence x(n) is expressed as Where k=0, ..., N-1 is the frequency index; According to the scaling property of Fourier transform, the Fourier transform of sub-wavelet function ψ(t / s) corresponds to We can get formula (4): The angular frequency in formula (4) is defined as 5. The method for suppressing radar interference signals as claimed in claim 4, characterized in that: The normalizing of the sub-wavelet functions at each scale s to a form having unit energy includes: For each of the dimensions s, the following equation is satisfied: According to formula (5), the normalized sub-wavelet function is obtained: Wherein, the base wavelet function ψ0(t) is a Morlet wavelet function, and the frequency of the Morlet wavelet function is: Where S(ω) represents the step function; The normalized sub-wavelet function is introduced into the formula (3) for calculation.

6. The method for suppressing radar interference signals as claimed in claim 5, characterized in that: The selecting of two transform domains of different scales comprises: The target signal y r In the large-scale continuous wavelet transform domain, the interference signal y is sparser. i It is even sparser in the continuous wavelet transform domain at small scales.

7. The method for suppressing radar interference signals as claimed in claim 6, characterized in that: The optimization of the objective function f(c) using the split augmented Lagrangian contraction algorithm includes: Optimize the objective function f(c) step by step, introduce the variable μ, and then the optimization problem of c1 and c2 is: stc-μ=0 The split augmented Lagrangian contraction algorithm is used to solve the problem, and μ and c are updated respectively, so that the objective function gradually converges to the optimal value. After k iterative calculations, the value of c is obtained, and c1 and c2 are calculated accordingly.

8. The method for suppressing radar interference signals as claimed in claim 7, characterized in that: The updating of μ and c respectively includes: introducing the intermediate variable d of the kth iteration k ∈C 2N×1 : S1: Calculate the current iteration c k The updated value of is recorded as: in ξ is 0.04; S2: Get c k After the update result, update μ k , denoted as μ k =s λ / k (c k +d k-1 ) formula (7), where λ = [λ1, λ2] T , that is, in μ k The upper part of is filtered by λ1, μ k The lower part of is filtered by λ2; S3: for the intermediate variable d k The value of is updated, recorded as: d k = d k-1 + c k - μ k Formula (8); After 100 iterations, the final result c is obtained. After obtaining the transform domain coefficients c1 and c2, the optimization problem of c1 and c2 is introduced to determine the target signal y r and interference signal y i .

9. A system for suppressing radar interference signals, characterized in that: Apply the method for suppressing radar interference signals as described in any one of claims 1-8.

10. A vehicle-mounted radar system, characterized in that: A system for suppressing radar jamming signals comprising the system as claimed in claim 9.

Images