Multi-path interference suppression method and device based on improved VMD-SVD collaborative noise reduction and storage medium

Abstract

The invention discloses a multipath interference suppression method and device during in-pipe detection and a storage medium, and belongs to the technical field of radar signal processing. According to the method, a CFAR detection and clustering algorithm combined method is adopted, and the target number and the center frequency are automatically extracted from a radar amplitude-frequency signal; an original noisy signal is decomposed into a target number of modal components by using an improved variational modal model, and an improved power reference analysis method is proposed to divide the modal components into effective modals and ineffective modals; sVD noise reduction is carried out on the effective mode; and performing signal reconstruction on the plurality of modal components after SVD noise reduction to obtain a signal after multipath interference suppression. According to the method, the problem that the decomposition number and the center frequency of the variational mode decomposition algorithm are difficult to determine is effectively solved, the influence of inaccurate parameters is reduced, multipath interference suppression during complex environment detection is facilitated, and the effectiveness of the algorithm is verified through actual measurement.

Description

Technical Field

[0001] The present invention relates to the field of radar signal processing technology, and in particular to a multipath interference suppression method, device and storage medium based on improved VMD-SVD collaborative noise reduction. Background Art

[0002] Millimeter-wave radar is widely used for internal inspections of oil and gas pipelines, urban underground pipe networks, and industrial containers due to its non-contact and high-precision features. However, the complex inner wall structures of confined spaces such as pipelines cause radar signals to undergo multiple reflections, diffraction, and scattering during propagation, resulting in severe multipath interference. This multipath effect causes delayed versions of multiple paths to be aliased in the received signal, resulting in false peaks, signal distortion, and a decrease in the signal-to-noise ratio during target detection, significantly impacting the accuracy of critical tasks such as pipeline defect location and corrosion monitoring. Especially in long-distance pipeline inspections, weak defect signals are often overwhelmed by multipath noise, making it difficult for traditional signal processing methods to effectively extract valid information.

[0003] Current multipath interference suppression techniques primarily include time-domain filtering, spatial beamforming, and time-frequency decomposition combined with noise reduction. Time-domain filtering requires pre-determining multipath delay parameters, making it difficult to implement in pipeline dynamic detection due to the complexity of the environment. Spatial beamforming relies on array antenna hardware, resulting in high equipment costs and unsuitable for single-antenna radar systems. Time-frequency decomposition methods, such as empirical mode decomposition (EMD) and its improved algorithms, while somewhat adaptable, suffer from modal aliasing and endpoint effects, and lack rigorous mathematical theoretical support. In recent years, variational mode decomposition (VMD) has attracted attention due to its non-recursive characteristics, strong noise resistance and mathematical completeness. However, it still faces significant bottlenecks in engineering applications: First, VMD requires pre-setting the mode number K and center frequency. In pipeline detection, due to the complex target reflection path, manual experience-based setting can easily lead to over-decomposition or under-decomposition, generating false modes or missing key information; second, multipath interference and real signals partially overlap in frequency domain energy distribution. Traditional power threshold methods cannot accurately distinguish between the two, resulting in effective signal loss or residual interference; third, single VMD or singular value decomposition (SVD) cannot completely suppress multipath noise, and the residual components will still interfere with subsequent feature extraction.

[0004] In existing technologies, some research attempts to improve noise reduction by optimizing VMD or combining it with SVD. For example, the traditional singular value difference spectrum method determines the effective order by finding the mutation points of the singular value sequence. However, it is prone to misjudgment in areas where the multipath signal and noise spectrum overlap, resulting in loss of useful signal. Furthermore, existing modal screening methods often rely on the kurtosis metric and fail to comprehensively consider the power attenuation characteristics and signal correlation of multipath interference, resulting in a disconnect between screening criteria and the actual needs of pipeline detection. Summary of the Invention

[0005] In response to the shortcomings of the existing technology, the present invention provides a multipath interference suppression method based on improved VMD-SVD collaborative noise reduction. It addresses the multipath interference problem in millimeter-wave radar detection in narrow and closed spaces such as pipelines, and significantly improves signal purity through parameter adaptive extraction, modal precise screening and hierarchical noise reduction optimization.

[0006] The present invention provides a multipath interference suppression method based on improved VMD-SVD collaborative noise reduction, comprising:

[0007] Step 1: Obtain the echo signal of the detection target and perform discrete sampling on the signal to obtain a discrete time domain signal;

[0008] Step 2: Perform a windowed fast Fourier transform on the discrete time domain signal to obtain an amplitude-frequency domain signal; mark the frequency position of the suspected target point in the amplitude-frequency domain based on the constant false alarm rate detection method. The target point includes the noise point and the detection point;

[0009] Step 3: Use the density-based DBSCAN clustering algorithm to cluster the suspected target points and extract target features to achieve automatic parameter extraction; select the frequency point with the largest signal intensity in each cluster as the target landmark point;

[0010] Step 4: Calculate the target number K and center frequency ω extracted from step 3 k , perform variational mode decomposition on the discrete time domain signal to obtain K eigenmode components μ k [ω];

[0011] Step 5: Use the power benchmark analysis method to screen the effective modes, construct the Hankel matrix for the screened effective modes and perform singular value decomposition to obtain the eigenmode components after singular value reconstruction and noise reduction.

[0012] Step 6: Superimpose the denoised intrinsic mode components to obtain the time domain signal after multipath interference suppression.

[0013] Furthermore, in step 2, the amplitude-frequency domain signal S FFT (f) is:

[0014]

[0015] Among them, S IF [n] is the discrete time domain signal, n is the sampling point number; S FFT (f) is the signal S IF [n] is the Fourier transform result; H is the signal length of the discrete signal; j is the imaginary unit; is a complex exponential function used to convert time domain signals into frequency domain signals.

[0016] Furthermore, in step 2, the detection threshold T(f i ):

[0017]

[0018] Among them, α CFAR is the false alarm probability factor, which is dynamically adjusted according to the statistical characteristics of pipeline environmental noise; N ref is the number of reference cells, which are located in the non-protected area on both sides of the target cell; S FFT is the discrete frequency domain signal obtained by fast Fourier transform of the millimeter wave radar intermediate frequency signal; S FFT (f i ) is the amplitude-frequency domain signal of the i-th reference unit;

[0019] If the unit to be detected D satisfies if|D| 2 >T(f i ), the unit to be detected is the target point.

[0020] Furthermore, in step 3, the neighborhood radius δ of the DBSCAN clustering algorithm is designed according to the radar range resolution:

[0021]

[0022] Among them, f s is the sampling frequency; N is the number of sampling points of FFT; m is the coefficient;

[0023] Furthermore, in step 4, the optimization model of the variational mode decomposition is:

[0024]

[0025] The constraints are:

[0026]

[0027] Where, δ(t) is the impulse response function; ω k is the center frequency of each mode function, ω k =2πf k , f k is the frequency domain frequency of the kth target landmark; K is the number of targets; μ k (t) is the time domain expression of the kth modal component; t is a discrete time point;

[0028] By introducing the Lagrangian function to construct the augmented Lagrangian function, the above constrained problem is transformed into an unconstrained problem and can be solved. The alternating optimization method is used to iteratively solve the problem. The modal update formula is:

[0029]

[0030] Where, s is the number of iterations; λ is the Lagrangian factor; α is the bandwidth adjustment factor; ω is the angular frequency of the modal component; μ q (ω) is the frequency domain expression of the qth modal component;

[0031] The termination conditions are:

[0032]

[0033] Here, ε represents the convergence tolerance.

[0034] Furthermore, in step 5, the power benchmark analysis method is used to screen the effective modes by designing screening conditions based on the original signal and the average power of each modal component, and the mode that meets the conditions is considered to be effective;

[0035] The screening conditions are:

[0036] 10·lg(P 原始 )-10·lg(P k )≤10

[0037]

[0038] Among them, P 原始 is the power of the original signal; P k is the signal power of the kth modal component.

[0039] Furthermore, in step 5, the SVD denoising step is:

[0040] Step 6.1: Construct the Hankel matrix H k :

[0041]

[0042] Where M is the two-dimensional matrix H k The number of columns, L is a two-dimensional matrix H k The number of rows of matrix H, L = H-M+1; Step 6.2: k Perform singular value decomposition:

[0043]

[0044] ∑ k =diag(σ1,σ2,…,σ r )

[0045] Among them, ∑ k is a diagonal matrix; σ l is the Hankel matrix H kThe first r eigenvalues ​​of are non-zero values, which are singular values, and satisfy σ1>σ2>σ3…>σ r ;

[0046] Step 6.3: Truncate the first p singular values ​​and reconstruct the denoised modal components; the truncation threshold p satisfies the following conditions:

[0047]

[0048] Where p is the number of retained singular values, r is the diagonal matrix ∑ k The order of .

[0049] Furthermore, in step 6, the denoised intrinsic mode components are superimposed as follows:

[0050]

[0051] in, is the discrete time series of the kth effective modal component after denoising.

[0052] The present invention also provides a computer device / equipment / system, including a memory, a processor, and a computer program stored in the memory. When the processor executes the computer program, the steps of the multipath interference suppression method based on improved VMD-SVD collaborative noise reduction are implemented in any one of the above items.

[0053] The present invention also provides a computer-readable storage medium having a computer program / instruction stored thereon, which, when executed by a processor, implements the steps of any of the above-mentioned multipath interference suppression methods based on improved VMD-SVD collaborative noise reduction.

[0054] The present invention also provides a computer program product, including a computer program / instruction, which, when executed by a processor, implements the steps of any of the above-mentioned multipath interference suppression methods based on improved VMD-SVD collaborative noise reduction.

[0055] The beneficial effects of the present invention are:

[0056] (1) The method of the present invention automatically extracts the number of targets K and the center frequency ω from the radar amplitude-frequency signal through CFAR detection and DBSCAN clustering algorithm. k , avoiding errors introduced by manual settings. Traditional methods often rely on manual experience or simple algorithms to determine VMD decomposition parameters, which can easily lead to inaccurate parameters due to subjective factors. This present invention uses CFAR detection to mark suspected target points, then uses DBSCAN clustering to merge scattered points and extract target features, achieving automatic and accurate parameter extraction, significantly improving the accuracy and reliability of target detection and laying a solid foundation for subsequent signal processing.

[0057] (2) The method of the present invention proposes a modal screening criterion based on logarithmic power difference to quantify the power difference between multipath signals and true targets, and distinguish between effective modes and interference modes. In a complex pipeline environment, multipath interference signals and true target signals overlap and are difficult to distinguish. The present invention uses the logarithmic power difference criterion to accurately quantify the power difference between the two, comprehensively evaluate the effectiveness of modal components, effectively eliminate multipath interference, retain the true target signal, improve the accuracy and reliability of signal processing, and ensure the accuracy of pipeline detection.

[0058] (3) The method of the present invention proposes a modal screening criterion based on logarithmic power difference, quantifies the power difference between multipath signals and true targets, and combines the mutual information weighted fusion index to distinguish effective modes from interference modes. In a complex pipeline environment, multipath interference signals and true target signals overlap and are difficult to distinguish. The present invention uses the logarithmic power difference criterion to accurately quantify the power difference between the two, and incorporates the mutual information weighted fusion index to comprehensively evaluate the effectiveness of modal components, effectively eliminate multipath interference, retain the true target signal, improve the accuracy and reliability of signal processing, and ensure the accuracy of pipeline detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 Schematic diagram of a process of an embodiment of the present invention;

[0060] Figure 2 A time domain diagram of a multipath interference signal of a radar inside a tube provided by an embodiment of the present invention;

[0061] Figure 3 A frequency domain diagram of a multipath interference signal of a radar inside a tube provided by an embodiment of the present invention;

[0062] Figure 4 CFAR detection result diagram provided by an embodiment of the present invention;

[0063] Figure 5 A time domain diagram of the modal component IMF1 provided in an embodiment of the present invention;

[0064] Figure 6 A time domain diagram of the modal component IMF2 provided in an embodiment of the present invention;

[0065] Figure 7 A time domain diagram of the modal component IMF3 provided in an embodiment of the present invention;

[0066] Figure 8 A time domain diagram of the modal component IMF4 provided in an embodiment of the present invention;

[0067] Figure 9 A time domain diagram of the modal component IMF5 provided in an embodiment of the present invention;

[0068] Figure 10 The singular value decomposition result of the modal component IMF1 provided in the embodiment of the present invention;

[0069] Figure 11 The singular value decomposition result of the modal component IMF3 provided in the embodiment of the present invention;

[0070] Figure 12 The singular value decomposition result of the modal component IMF4 provided in the embodiment of the present invention;

[0071] Figure 13 This is a frequency domain diagram of the in-pipe radar signal after noise reduction according to an embodiment of the present invention. DETAILED DESCRIPTION

[0072] To make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be described in detail below in conjunction with the accompanying drawings in the embodiments of the present invention. It should be understood that the described embodiments only represent a part of the present invention, not all of it. Based on the embodiments disclosed by the present invention, other embodiments obtained by ordinary technicians in this field without creative work are all within the scope of protection of the present invention.

[0073] The present invention innovatively uses the CFAR-DBSCAN algorithm to obtain the target number K and the corresponding center frequency ω, providing a basis for VMD when performing modal decomposition, and effectively solving the problem of unreliable decomposition results caused by incorrect selection of the modal number and failure to lock the center frequency during VMD modal decomposition. In terms of modal screening, an innovative modal screening criterion based on logarithmic power difference is proposed to solve the problem of difficulty in effective modal screening. In pipeline detection, the present invention can effectively solve the serious multipath interference problem caused by the narrow pipeline space, accurately suppress multipath interference, select the real target signal, and complete the precise detection of targets in the pipeline space.

[0074] The present invention provides a multipath interference suppression method based on improved VMD-SVD collaborative noise reduction. Figure 1 As shown, including:

[0075] (1) Acquisition of the original noisy discrete time domain signal: The echo signal of the detected target is obtained by frequency modulated continuous wave millimeter wave radar, and the signal is discretely sampled to obtain the discrete time domain signal S IF [n], where n is the sampling point number;

[0076] (2) Time-frequency conversion and target detection: The discrete time domain signal is subjected to a windowed fast Fourier transform (FFT) to obtain an amplitude-frequency domain signal S FFT (f) Mark the frequency position of the suspected target point in the amplitude-frequency domain based on the constant false alarm rate (CFAR) detection method;

[0077] Perform CFAR detection on the frequency domain information of the original signal to detect all suspected targets, including:

[0078] Calculate the average power of the reference units on both sides of the unit to be detected as an estimate of the background noise:

[0079]

[0080] Where Z CA is the average power of the reference unit; x and y are the signal amplitudes of the reference units on the left and right sides of the unit to be detected; N is the number of reference units.

[0081] Calculate the detection threshold:

[0082] T=α·Z CA

[0083] Among them, α is the threshold factor, which is used to control the false alarm probability P fa , the calculation formula of the threshold factor is:

[0084] α=-ln(P fa )

[0085] This formula assumes that the noise power follows an exponential distribution.

[0086] By comparing the detection threshold T with the power of the unit to be detected D, it can be determined whether the unit to be detected is a suspected target:

[0087] if|D| 2 >T

[0088] If the power of the unit to be detected D is greater than the threshold T, it is considered to contain a target. For discrete frequency domain signals with N units to be detected, a sliding window is set to detect each unit in turn to achieve global frequency domain detection and select all suspected targets. By labeling each suspected target, the number and distribution of suspected targets can be clearly observed.

[0089] (3) Target clustering and feature extraction: The density-based DBSCAN clustering algorithm is used to cluster the suspected target points, merge the scattering points of the same target, and select the frequency point with the largest signal intensity in each cluster as the target landmark point to obtain the target number K and its corresponding center frequency f k ;

[0090] (4) Variational Mode Decomposition (VMD): Based on the target number K and the center frequency ω k =2πf k , perform simplified variational mode decomposition on the original signal to obtain K eigenmode components μ k [ω];

[0091] (5) Modal power screening: Calculate the average power of each eigenmodal component, screen the effective modes through the improved power benchmark analysis method, and eliminate invalid modes with power lower than the preset threshold;

[0092] (6) SVD denoising: Construct the Hankel matrix for the selected effective modes and perform singular value decomposition (SVD), truncate small singular values ​​to suppress residual noise, and reconstruct the denoised modal components.

[0093] (7) Signal reconstruction: superimposing the noise-reduced modal components to generate a time domain signal after multipath interference is suppressed.

[0094] Example 1

[0095] An embodiment of the present invention provides a multipath interference suppression method based on improved VMD-SVD collaborative noise reduction; the method includes:

[0096] S1. Build a frequency modulated continuous wave millimeter wave radar pipeline target acquisition platform. Since the space inside the pipeline is small and not absolutely straight and smooth, a large amount of multipath interference will be generated when collecting object data in the pipeline to form false targets. Collect the original echo signal containing multipath interference; perform discrete sampling to obtain the time domain signal S containing multipath interference. IF [n]; Perform windowed Fast Fourier Transform (FFT) on the signal to obtain the amplitude-frequency domain signal S FFT (f), suppress spectrum leakage and retain effective frequency components;

[0097] S2, based on CFAR target detection and DBSCAN clustering algorithm, the amplitude and frequency domain information of the signal is detected and adaptive parameters are extracted;

[0098] S3, using the improved VMD algorithm to decompose the time domain signal into K modal components;

[0099] S4. Calculate the power of each modal component decomposed by VMD, screen the modes using the power benchmark analysis method, divide them into valid modes and invalid modes, and discard the invalid modes;

[0100] S5. Use the effective modal signal to construct a Hankel matrix, decompose it using the SVD algorithm to obtain the singular value distribution of the signal to be denoised, and set the singular values ​​corresponding to the interference signal to zero;

[0101] S6. Reconstruct the processed modal components to obtain a denoised signal after multipath interference suppression;

[0102] In S1, the time domain signal S IF [n] Perform fast Fourier transform (FFT) to obtain the amplitude-frequency domain signal S FFT(f):

[0103]

[0104] Where S FFT (f) is the signal S IF [n] is the Fourier transform result; H is the signal length of the discrete signal; j is the imaginary unit; is a complex exponential function used to convert time domain signals into frequency domain signals.

[0105] In S2, adaptive parameter extraction includes target detection and clustering, where:

[0106] The CFAR detection method is a unit average CFAR (CA-CFAR) algorithm that marks suspected target points in the amplitude-frequency domain. Specifically, the detection threshold for each frequency unit is calculated by weighting the average energy of adjacent reference units. The detection threshold T(f i ) is calculated as:

[0107]

[0108] Among them, α CFAR is the false alarm probability factor, which is dynamically adjusted according to the statistical characteristics of pipeline environmental noise; N ref is the number of reference cells, which are located in the non-protected area on both sides of the target cell; S FFT is the discrete frequency domain signal obtained by fast Fourier transform of the millimeter wave radar intermediate frequency signal; S FFT (f i ) is the amplitude-frequency domain signal of the i-th reference unit.

[0109] If the unit to be detected D satisfies if|D| 2 >T(f i ), the unit to be detected is the target point.

[0110] The targets detected by CFAR cannot accurately describe the number and amplitude characteristics of the targets, so the clustering algorithm is further used to cluster all suspected targets. The number of clusters indicates the number of targets containing multipath targets, and the point with the highest signal strength in each cluster is selected as the target landmark. The target number K and the center frequency ω are determined based on the landmark. k :

[0111] ω k =2πf k

[0112] Among them, f k is the frequency domain frequency of the kth target landmark point.

[0113] The clustering algorithm uses DBSCAN clustering. The neighborhood radius of the DBSCAN algorithm is dynamically set based on the radar range resolution to perform density clustering on the detected suspected target points. By merging the scattering points of the same target, the false peaks caused by the target sidelobes are eliminated. Finally, the frequency point with the largest signal strength in each cluster is selected as the target landmark point, and the number of targets K and the center frequency ω are determined. k , providing accurate prior parameters for subsequent VMD decomposition.

[0114] The neighborhood radius δ of the DBSCAN clustering algorithm is based on the radar range resolution. Settings, specifically:

[0115] δ=m·Δf

[0116] Among them, f s is the sampling frequency, and N is the number of FFT sampling points.

[0117] In S3, VMD decomposition is performed on the original time domain signal containing multipath interference based on the prior information target number K and center frequency obtained in the previous step. The VMD solution is mainly divided into the construction and solution of the variational model:

[0118] The center frequencies of each mode are known through prior information, so the traditional variational constraint model is simplified to obtain:

[0119]

[0120] The constraints are:

[0121]

[0122] Among them, μ k are the modal components obtained by VMD decomposition; δ(t) is the impulse response function; ω k is the center frequency of each mode function; S IF is the intermediate frequency signal obtained by mixing the millimeter-wave radar receiving signal; K is the target number of modal components obtained by decomposition.

[0123] In this embodiment, VMD decomposition is achieved by solving the optimal solution of the above-mentioned variational constraint model to achieve adaptive decomposition of the signal. Due to the difficulty in solving the constraint problem, this embodiment introduces a quadratic penalty factor α and a Lagrangian multiplier λ to construct an augmented Lagrangian function, converting the constrained problem into an unconstrained problem:

[0124]

[0125] Taking partial derivatives of the constructed augmented Lagrangian function, we can get the modal function μ k And the solution of the Lagrange multiplier λ, the most important modal function μk The solution is:

[0126]

[0127] Among them, S IF is the intermediate frequency signal in the frequency domain; s is the number of iterations in the solution process; λ is the Lagrangian factor; ω k is the center frequency of each modal component; α is the bandwidth adjustment factor; μ q (ω) is the frequency domain expression of the qth modal component.

[0128] In the solution process, the alternating direction multiplication method is used to solve μ k and λ are iteratively solved, and eventually each mode is concentrated near the corresponding center frequency, preliminarily separating multipath interference and valid signals.

[0129] Furthermore, the alternating direction multiplier method is used to alternately update the K modal components and the Lagrange multiplier until the termination condition is met:

[0130]

[0131] Here, ε represents the convergence tolerance.

[0132] The original multipath interference signal is decomposed by VMD to generate several intrinsic mode functions.

[0133] In S4, the modes are screened using the power benchmark analysis method proposed in the present invention, including:

[0134] Calculate the original noisy signal and the average power of each modal component:

[0135]

[0136] Where, P 原始 Represents the average power of the original signal; N represents the number of signal sampling points of the discrete signal; S IF Represents the original intermediate frequency signal; P k represents the average power of the signal of the kth modal component; μ k [ω] represents the kth modal component.

[0137] After calculating the average power of each signal, multipath interference causes severe power attenuation due to multiple transmissions in the confined space of the tube. Based on the power attenuation characteristics of multipath interference (usually more than one order of magnitude greater than the actual signal), a logarithmic power difference screening criterion is proposed to screen out the effective mode and the mode containing multipath interference:

[0138] 10·lg(P 原始 )-10·lg(P k )≤10

[0139] This method can effectively distinguish low-power multipath interference from high-power valid signals, avoiding the over-elimination problem of the traditional threshold method. In S5, the Hankel matrix is ​​constructed and the SVD algorithm is used to remove narrowband interference. The matrix form is:

[0140]

[0141] Matrix H k Is an M*L order matrix, assuming its rank is r, then there must be an M order orthogonal matrix U k , an M*L order diagonal matrix ∑ k and an L-order orthogonal matrix V k , such that:

[0142] ∑ k =diag(σ1,σ2,…,σ r )

[0143] Diagonal matrix ∑ k The first r non-zero values ​​on the diagonal of are singular values, and satisfy σ1>σ2>σ3…>σ r .

[0144] SVD denoising concentrates the energy of the signal on the front singular values ​​as much as possible. The size of the singular value reflects the amount of information of the signal and noise. The present invention determines the retention of the singular value by the energy proportion of the singular value:

[0145]

[0146] Where p is the number of retained singular values ​​and r is the order of the diagonal matrix ∑.

[0147] In S6, the modal components after noise reduction are Superposition generates the final denoised time domain signal, and the signal is reconstructed as:

[0148]

[0149] Example 2

[0150] A frequency-modulated continuous-wave millimeter-wave radar acquisition platform was used to collect radar echo signals during in-pipe measurements, validating the proposed method under typical multipath interference conditions. The measurements were performed on a 20m long, 50mm diameter steel pipe with three unevenly spaced objects inside. The radar was tightly connected to the pipe to ensure that the received signal contained only the measurement data from within the pipe.

[0151] The time domain diagram of the original noisy signal collected is as follows: Figure 2As shown, performing FFT can obtain the frequency domain diagram of the signal as shown Figure 3 As shown by Figure 3 It can be observed that the multipath interference in the signal is very serious at this time, and the signal-to-noise ratio is poor at this time SNR=0.502dB.

[0152] By performing CAFR detection on the frequency domain information of the signal, all suspected targets can be detected, and the suspected targets are set to 1 and the non-target points are set to 0, forming the CFAR detection result as follows: Figure 4 As shown in the figure, there are five clusters of targets, and each cluster has multiple target points. In order to better determine the number of targets and calculate the corresponding center frequency, Figure 4 The detection results use clustering algorithm and select the landmark points of each cluster, and finally determine the prior information K = 5, f1 = 404064Hz, f1 = 623819Hz, f1 = 796314Hz, f1 = 975898Hz, f1 = 1850190Hz.

[0153] According to the prior information, the original signal time domain information is decomposed by VMD, and the Figures 5 to 9 The power benchmark analysis method is used to calculate the power of the original signal and the power of each mode. IMF1, IMF3, and IMF4 are recorded as valid modes, and IMF2 and IMF5 are recorded as invalid signals caused by multipath interference.

[0154] The three effective modes are subjected to singular value decomposition respectively, and the singular value decomposition results are as follows: Figures 10 to 12 As shown in the figure, to determine the effective singular value order of each mode, the energy distribution of the singular values ​​of each mode is calculated. The result shows that the effective order of IMF1 is 2, the effective order of IMF3 is 2, and the effective order of IMF4 is 3. Then, the three singular value sequences are restored to one-dimensional time domain signals, and finally reconstructed to complete the noise reduction.

[0155] The final reconstruction result is as follows Figure 13 As shown, Figure 3 Comparing the frequency domain graphs before and after noise reduction, it can be found that multipath interference has been basically eliminated after noise reduction, and the signal-to-noise ratio has been significantly improved to SNR = 5.376dB without causing any loss of effective components.

[0156] In summary, the method of the present invention innovatively combines CFAR detection and clustering algorithms to effectively solve the problem of difficulty in determining the number of VMD decompositions and center frequency. It includes: detecting suspected targets in the signal based on CFAR; clustering the discrete points of the target using a clustering algorithm, and selecting the signal landmarks in each cluster based on the signal strength of the discrete points; and calculating the prior parameters of VMD decomposition number K and center frequency ω based on CFAR and clustering. kThe method uses simplified VMD to extract K modal components from the original noisy signal. A power benchmark analysis method is proposed to separate valid and invalid modes. SVD is then applied to the selected valid modes for noise reduction. Signal reconstruction of the SVD-denoised modal components yields a signal with multipath interference suppressed. This method is beneficial for suppressing multipath interference during complex environment detection, and field measurements have verified the effectiveness of the algorithm.

[0157] In particular, in some preferred embodiments of the present invention, a computer device is also provided, including a memory and a processor and a computer program stored on the memory. When the processor executes the computer program, the steps of the multipath interference suppression method based on improved VMD-SVD collaborative noise reduction in any of the above embodiments are implemented.

[0158] In other preferred embodiments of the present invention, a computer-readable storage medium is also provided, on which a computer program / instructions are stored. When the computer program is executed by a processor, the steps of the multipath interference suppression method based on improved VMD-SVD collaborative noise reduction described in any of the above embodiments are implemented.

[0159] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiment method can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it may include the process of the above-mentioned multipath interference suppression method embodiment based on the improved VMD-SVD collaborative noise reduction, which will not be repeated here.

[0160] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or N embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine different embodiments or examples described in this specification and the features of different embodiments or examples without contradiction.

[0161] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of technical features indicated. Thus, a feature specified as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of the present invention, "N" means at least two, such as two, three, etc., unless otherwise specifically defined.

[0162] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code comprising one or more executable instructions for implementing a custom logical function or step of a process, and the scope of the preferred embodiments of the invention includes alternative implementations in which functions may be performed out of the order shown or discussed, including performing functions in a substantially simultaneous manner or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of the invention pertain.

[0163] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection with one or N wires (electronic devices), a portable computer disk cartridge (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and programmable read-only memory (EPROM or flash memory), fiber optic devices, and a portable compact disc read-only memory (CDROM). Furthermore, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting or processing it in another suitable manner if necessary, and then storing it in a computer memory.

[0164] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiment, the N steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used: a discrete logic circuit having logic gate circuits for implementing logic functions on data signals, an application-specific integrated circuit having suitable combinational logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0165] Those skilled in the art will understand that all or part of the steps in the method of the above embodiment can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiment.

[0166] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing module, or each unit may exist physically separately, or two or more units may be integrated into a single module. The aforementioned integrated modules may be implemented in the form of hardware or in the form of software functional modules. If the integrated modules are implemented in the form of software functional modules and sold or used as independent products, they may also be stored in a computer-readable storage medium.

[0167] The storage medium mentioned above may be a read-only memory, a magnetic disk, or an optical disk, etc. Although the embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and are not to be construed as limiting the present invention. Persons skilled in the art may make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present invention.

Claims

1. A multipath interference suppression method based on improved VMD-SVD collaborative noise reduction, characterized by: include: Step 1: Obtain the echo signal of the detection target and perform discrete sampling on the signal to obtain a discrete time domain signal; Step 2: Perform a windowed fast Fourier transform on the discrete time domain signal to obtain an amplitude-frequency domain signal; mark the frequency position of the suspected target point in the amplitude-frequency domain based on the constant false alarm rate detection method. The target point includes the noise point and the detection point; Step 3: Use the density-based DBSCAN clustering algorithm to cluster the suspected target points and extract target features to achieve automatic parameter extraction; select the frequency point with the largest signal intensity in each cluster as the target landmark point; Step 4: Calculate the target number K and center frequency ω extracted from step 3 k , perform variational mode decomposition on the discrete time domain signal to obtain K eigenmode components μ k [ω]; Step 5: Use the power benchmark analysis method to screen the effective modes, construct the Hankel matrix for the screened effective modes and perform singular value decomposition to obtain the eigenmode components after singular value reconstruction and noise reduction. Step 6: Superimpose the denoised intrinsic mode components to obtain the time domain signal after multipath interference suppression.

2. The multipath interference suppression method based on improved VMD-SVD collaborative noise reduction according to claim 1, characterized in that: In step 2, the amplitude-frequency domain signal S FFT (f) is: Among them, S IF [n] is the discrete time domain signal, n is the sampling point number; S FFT (f) is the signal S IF [n] is the Fourier transform result; H is the signal length of the discrete signal; j is the imaginary unit; is a complex exponential function used to convert time domain signals into frequency domain signals.

3. The multipath interference suppression method based on improved VMD-SVD collaborative noise reduction according to claim 1, characterized in that: In step 2, the detection threshold T(f i ): Among them, α CFAR is the false alarm probability factor, which is dynamically adjusted according to the statistical characteristics of pipeline environmental noise; N ref is the number of reference cells, which are located in the non-protected area on both sides of the target cell; S FFT is the discrete frequency domain signal obtained by fast Fourier transform of the millimeter wave radar intermediate frequency signal; S FFT (f i ) is the amplitude-frequency domain signal of the i-th reference unit; If the unit to be detected D satisfies if|D| 2 >T(f i ), the unit to be detected is the target point.

4. The multipath interference suppression method based on improved VMD-SVD collaborative noise reduction according to claim 1, characterized in that: In step 3, the neighborhood radius δ of the DBSCAN clustering algorithm is designed according to the radar range resolution: Among them, f s is the sampling frequency; N is the number of FFT sampling points; m is the coefficient.

5. The multipath interference suppression method based on improved VMD-SVD collaborative noise reduction according to claim 1, characterized in that: In step 4, the optimization model of the variational mode decomposition is: The constraints are: Where, δ(t) is the impulse response function; ω k is the center frequency of each mode function, ω k =2πf k , f k is the frequency domain frequency of the kth target landmark; K is the number of targets; μ k (t) is the time domain expression of the kth modal component; t is a discrete time point; By introducing the Lagrangian function to construct the augmented Lagrangian function, the above constrained problem is transformed into an unconstrained problem and can be solved. The alternating optimization method is used to iteratively solve the problem. The modal update formula is: Where, s is the number of iterations; λ is the Lagrangian factor; α is the bandwidth adjustment factor; ω is the angular frequency of the modal component; μ q (ω) is the frequency domain expression of the qth modal component; The termination conditions are: Here, ε represents the convergence tolerance.

6. The multipath interference suppression method based on improved VMD-SVD collaborative noise reduction according to claim 1, characterized in that: In step 5, the power benchmark analysis method is used to screen effective modes by designing screening conditions based on the original signal and the average power of each modal component, and the mode that meets the conditions is considered to be effective; The screening conditions are: 10·lg(P 原始 )-10·lg(P k )≤10 Among them, P 原始 is the power of the original signal; P k is the signal power of the kth modal component.

7. The multipath interference suppression method based on improved VMD-SVD collaborative noise reduction according to claim 1, characterized in that: In step 5, the denoising by singular value decomposition is specifically as follows: Step 5.1: Construct the Hankel matrix H k : Where M is the two-dimensional matrix H k The number of columns, L is a two-dimensional matrix H k The number of rows, L = H - M + 1; Step 5.2: For the matrix H k Perform singular value decomposition: ∑ k =diag(σ1,σ2,…,σ r ) Among them, ∑ k is a diagonal matrix; σ l is the Hankel matrix H k The first r eigenvalues ​​of are non-zero values, which are singular values, and satisfy σ1>σ2>σ3…>σ r ; Step 5.3: Truncate the first p singular values ​​and reconstruct the denoised modal components; the truncation threshold p satisfies the following conditions: Where p is the number of retained singular values, r is the diagonal matrix ∑ k The order of .

8. The multipath interference suppression method based on improved VMD-SVD collaborative noise reduction according to claim 1, characterized in that: In step 6, the denoised intrinsic mode components are superimposed as follows: in, is the kth effective modal component after denoising.

9. A computer device / apparatus / system comprising a memory, a processor, and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program / instruction stored thereon, characterized in that: When the computer program / instructions are executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.

Images