Noise reduction method and system based on millimeter wave radar signal

Abstract

The invention relates to the field of signal processing, and discloses a noise reduction method and system based on millimeter wave radar signals, and the method comprises the following steps: collecting millimeter wave radar signals to be denoised; the method comprises the following steps: constructing a fitness function of a ratio of a mean value to a variance of a multi-scale Kolmogorov entropy based on a millimeter wave radar signal; optimizing a decomposition mode number K and a penalty factor alpha of variational mode decomposition by using a starfish optimization search algorithm; variational mode decomposition is carried out on the millimeter wave radar signal; the sample entropy of each intrinsic mode component is calculated, the mode position with the maximum sample entropy break variable is determined, the intrinsic mode components with the sample entropy smaller than the position serve as signal modes, and the rest serve as noise modes; carrying out wavelet threshold denoising on the noise mode; and performing signal reconstruction by using the signal mode and the de-noised noise mode, and outputting a de-noised radar signal. The millimeter-wave radar gas leakage signal noise reduction method solves the problem that the noise reduction effect of millimeter-wave radar gas leakage signals is poor in the prior art, and has the advantage of being capable of improving the signal-to-noise ratio of the signals.

Description

Technical Field

[0001] The present invention relates to the field of signal processing, and more specifically, to a noise reduction method and system based on millimeter-wave radar signals. Background Art

[0002] Millimeter-wave radar can be used for gas leak detection, enabling all-weather and high-precision detection. However, the radar system generates a large amount of noise within the system due to its manufacturing process and production level. Gas leaks also generate environmental noise such as gas flow noise and equipment operation noise. In addition, the vibration amplitude of gas leakage signals is very small, making detection very difficult. Therefore, it is necessary to perform noise reduction processing on the signal to improve the signal-to-noise ratio of the leakage signal.

[0003] The prior art provides a radar signal denoising method, apparatus, device, storage medium and program product. The method uses an adaptive filter to filter an original radar signal reflected by a human body; decomposes the processed radar signal based on an improved adaptive noise completely integrated empirical mode decomposition (ICEEMDAN) algorithm to obtain multiple intrinsic mode function components; determines multiple high-frequency and low-frequency intrinsic mode function components from the multiple intrinsic mode function components, performs wavelet transform denoising on the multiple high-frequency intrinsic mode function components to obtain multiple denoised high-frequency intrinsic mode function components; and reconstructs the multiple low-frequency intrinsic mode function components and the denoised multiple high-frequency intrinsic mode function components to obtain a reconstructed radar signal.

[0004] However, the existing technology has poor noise reduction effect on millimeter-wave radar gas leakage signals. Therefore, how to invent a noise reduction method based on millimeter-wave radar signals is a technical problem that urgently needs to be solved in this technical field. Summary of the Invention

[0005] In order to solve the problem that the existing technology has poor noise reduction effect on millimeter-wave radar gas leakage signals, the present invention provides a noise reduction method and system based on millimeter-wave radar signals, which has the characteristic of improving the signal-to-noise ratio.

[0006] In order to achieve the above-mentioned purpose of the present invention, the technical solutions adopted are as follows:

[0007] A noise reduction method based on millimeter wave radar signals, comprising the following steps:

[0008] Collect millimeter-wave radar signals to be denoised;

[0009] Based on millimeter-wave radar signals, a fitness function is constructed based on the ratio of the mean to the variance of the multi-scale Kolmogorov entropy.

[0010] The starfish optimization search algorithm is used to optimize the decomposition mode number K and penalty factor α of variational mode decomposition;

[0011] Based on the optimized K and α, the millimeter-wave radar signal is subjected to variational mode decomposition to obtain multiple eigenmode components;

[0012] Calculate the sample entropy of each eigenmode component and sort them from small to large according to the sample entropy value. Determine the modal position with the largest sample entropy mutation. The eigenmode components with sample entropy less than that position are regarded as signal modes, and the rest are regarded as noise modes.

[0013] Perform wavelet threshold denoising on noise modes;

[0014] The signal mode and the denoised noise mode are used to reconstruct the signal and output the denoised radar signal.

[0015] Preferably, the millimeter-wave radar signal to be denoised is collected, and the specific steps are: using the millimeter-wave radar to transmit a frequency-modulated continuous wave signal to monitor the target container; the frequency-modulated continuous wave signal is used to measure the vibration information of the target through linear frequency modulation; the signal emitted by the radar generates an echo when encountering a container or leaking gas, and the receiving antenna captures the echo signal to obtain the millimeter-wave radar signal to be denoised.

[0016] Furthermore, the multi-scale Kolmogorov entropy, wherein the specific expression of Kolmogorov entropy is:

[0017]

[0018] Where n is the number of millimeter wave radar signal information, P i Indicates that information X appears at time τ i probability.

[0019] Going further, to solve the multi-scale Kolmogorov entropy, the specific steps are:

[0020] Reconstruct the time series of millimeter-wave radar signals in phase space to obtain an m-dimensional vector space point set R;

[0021] Calculate the two points r in phase space ij The Euclidean distance C(m,r ij )value;

[0022] Reduce the value of r until C(m,r ij ) remains unchanged, then C(m,r ij ) is recorded as C(m,r);

[0023] Change the value of m to calculate the corresponding C(mx,r), and set the scale factor T to update the sequence. The Kolmogorov entropy of the new sequence is expressed as:

[0024]

[0025] Calculate the multiscale Kolmogorov entropy:

[0026]

[0027] Among them, T satisfies Where N is the set original sequence length.

[0028] Furthermore, the fitness function of the ratio of the mean to the variance of the multi-scale Kolmogorov entropy is:

[0029]

[0030] Among them, Fitness is the fitness function, Mean is the mean, and VAR is the variance.

[0031] Furthermore, the starfish optimization search algorithm is used to optimize the decomposition mode number K and penalty factor α of variational mode decomposition. The specific steps are as follows:

[0032] Select the value range of K and α, and initialize K and α;

[0033] Initialize the SFOA optimization algorithm: determine the population size, that is, the number of candidate solutions in the algorithm, and set the maximum number of iterations;

[0034] The initialized [K, α] parameters are used as the input of variational mode decomposition to decompose the given signal;

[0035] Iteratively calculate the fitness value of the intrinsic mode component obtained by decomposition; if the current fitness value is less than the fitness value of the previous iteration, the current fitness value is used as the new fitness value; if the current fitness value is not less than the fitness value of the previous iteration, the fitness value remains unchanged; repeat the iteration until the preset maximum number of iterations t is reached, and the optimal fitness value and its corresponding parameters [K, α] are obtained.

[0036] Furthermore, based on the optimized K and α, the millimeter-wave radar signal is subjected to variational mode decomposition to obtain multiple intrinsic mode components. The specific steps are as follows:

[0037] Decompose the millimeter-wave radar signal into K IMF intrinsic mode functions u k (t);

[0038] Use Hilbert transform to transform each modal function u k (t) Analyze and obtain the corresponding analytical function;

[0039]

[0040] Utilization Index Correction, so that the spectrum of each modal component is modulated to the corresponding baseband;

[0041]

[0042] The bandwidth of each modal function is obtained by calculating the square L2 norm of the gradient of the analytical function;

[0043]

[0044] The constrained variational mode decomposition model is specifically:

[0045]

[0046] Where, define u k ={u1,u2,...,u k} is the modal component, ω k ={ω1,ω2,...,ω k} is u k The center frequency of , * is the convolution operation, and δ(t) refers to the unit impulse function. Represents the partial derivative of the function with respect to time t;

[0047] Considering the penalty factor α, the constrained variational mode decomposition model is actually:

[0048]

[0049] Here, λ is the Lagrange multiplier.

[0050] Furthermore, when the sample entropy of each modality is arranged in ascending order, the maximum absolute value of the difference between the entropies of two adjacent samples is taken as the mutation point β:

[0051] β=argmax|S(IMF L+1 )-S(IMF L )|

[0052] Among them, argmax is the maximum function, S is the sample entropy, IMF L is the Lth mode;

[0053] The mutation point β is taken as the modal position with the largest mutation amount, the ones before β are signal modes, and the other sample entropies are noise modes.

[0054] Furthermore, the noise mode is subjected to wavelet threshold denoising. Specifically, for the signal classified as the noise mode, the wavelet denoising algorithm is used to remove the noise by decomposing the signal into wavelet subspaces of different frequencies and utilizing the sparsity and statistical characteristics of the wavelet coefficients.

[0055] A system based on millimeter-wave radar signals, comprising a cascaded signal acquisition module, a fitness function module, a factor optimization module, a variational mode decomposition module, a classification module, a denoising module, and a signal output module;

[0056] The signal acquisition module is used to collect the millimeter-wave radar signal to be denoised;

[0057] The fitness function module is used to construct a fitness function based on the ratio of the mean to the variance of the multi-scale Kolmogorov entropy based on the millimeter wave radar signal;

[0058] The factor optimization module is used to optimize the decomposition mode number K and penalty factor α of variational mode decomposition using the starfish optimization search algorithm;

[0059] The variational mode decomposition module is used to calculate the sample entropy of each eigenmode component, and sort the sample entropy values ​​from small to large to determine the modal position with the largest mutation amount. The eigenmode components with sample entropy less than that position are regarded as signal modes, and the rest are regarded as noise modes.

[0060] The classification module is used to perform variational modal decomposition on the millimeter wave radar signal based on the optimized K and α to obtain multiple eigenmode components;

[0061] The denoising module is used to perform wavelet threshold denoising on the noise mode;

[0062] The signal output module is used to reconstruct the signal and output the radar signal after noise reduction.

[0063] The beneficial effects of the present invention are as follows:

[0064] This invention addresses the problem of excessive noise in millimeter-wave radar gas leakage signals, which can affect subsequent signal recognition, caused by existing denoising algorithms. This method provides a denoising method for millimeter-wave radar gas leakage signals. The method uses the ratio of the mean to the variance of the multi-scale Kolmogorov entropy as the fitness function, and employs the Starfish optimization search algorithm to optimize the number of decomposition modes K and the penalty factor α in variational mode decomposition. Sample entropy is used to classify the individual modal signals, and wavelet denoising is used to further remove noise. Multiple modal signals are reconstructed to obtain a reconstructed radar signal, thereby resolving the problem of excessive radar signal noise. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 It is a flow chart of a noise reduction method based on millimeter-wave radar signals according to the present invention.

[0066] Figure 2 This is a flow chart of millimeter wave radar gas leakage signal noise reduction in Example 2.

[0067] Figure 3These are the time domain and frequency domain waveforms of each mode of the simulation signal in Example 3.

[0068] Figure 4 This is a diagram of the simulation signal denoising results in Example 3.

[0069] Figure 5 This is the SFOA-VMD decomposition result diagram of the millimeter-wave radar gas leakage signal in Example 3.

[0070] Figure 6 This is a graph showing the denoising results of the millimeter-wave radar gas leakage signal in Example 3. DETAILED DESCRIPTION

[0071] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0072] Example 1

[0073] like Figure 1 As shown, a noise reduction method based on millimeter wave radar signals includes the following steps:

[0074] Collect millimeter-wave radar signals to be denoised;

[0075] The mean of multi-scale Kolmogorov entropy divided by the variance is used as the fitness function of the optimization algorithm, and this function is used to evaluate the performance and effect of the algorithm;

[0076] Based on the fitness function, the Starfish Optimization Algorithm (SFOA) is used for efficient search, aiming to find the optimal number of decomposition modes and the appropriate penalty factor of variational mode decomposition (VMD), thereby optimizing the decomposition process.

[0077] Based on the optimal number of decomposition modes and penalty factors found in the previous step, the acquired radar signal is decomposed into multiple independent modal signals through variational modal decomposition to facilitate subsequent detailed analysis and processing.

[0078] Using sample entropy as the evaluation criterion, the sample entropy of each modal signal obtained by decomposition is calculated, and these modal signals are sorted from small to large according to the size of the sample entropy. Through this sorting process, the modal position with the largest sample entropy mutation is found, and the modal signal with a sample entropy smaller than this mode is classified as the signal mode, while the other modal signals are classified as noise modes.

[0079] For the signal classified as noise mode, the wavelet denoising algorithm is used to perform further denoising to eliminate the noise components, thereby obtaining the denoised pure noise mode signal;

[0080] Multiple signal modal components and multiple noise modal components that have undergone denoising are comprehensively reconstructed. Through this reconstruction process, the reconstructed radar signal is finally obtained, ensuring that it not only retains the effective information of the original signal but also removes noise interference, thereby improving the overall quality of the signal.

[0081] Example 2

[0082] More specifically, in one embodiment, the millimeter-wave radar signal to be denoised is collected, and the specific steps are as follows: using the millimeter-wave radar to transmit a frequency-modulated continuous wave signal to monitor the target container; the frequency-modulated continuous wave signal is used to measure the vibration information of the target through linear frequency modulation; the signal transmitted by the radar generates an echo when encountering a container or leaking gas, and the receiving antenna captures the echo signal to obtain the millimeter-wave radar signal to be denoised.

[0083] In this embodiment, the specific noise reduction process of the noise reduction method based on millimeter wave radar signal is as follows: Figure 2 shown.

[0084] In this embodiment, K is selected as an integer between [2, 10], and the value range of α is set to [500, 6000]. Such a range selection helps to find the appropriate frequency band and vibration mode in signal decomposition.

[0085] In a specific embodiment, the multi-scale Kolmogorov entropy, wherein the Kolmogorov entropy is specifically expressed as:

[0086]

[0087] Where n is the number of millimeter wave radar signal information, P i Indicates that information X appears at time τ i probability.

[0088] In a specific embodiment, the multi-scale Kolmogorov entropy value is solved by the following steps:

[0089] Reconstruct the time series of millimeter-wave radar signals in phase space to obtain an m-dimensional vector space point set R;

[0090] Calculate the two points r in phase space ij The Euclidean distance C(m,r ij )value;

[0091] Reduce the value of r until C(m,r ij ) remains unchanged, then C(m,r ij ) is recorded as C(m,r);

[0092] Change the value of m to calculate the corresponding C(mx,r), and set the scale factor T to update the sequence. The Kolmogorov entropy of the new sequence is expressed as:

[0093]

[0094] Calculate the multiscale Kolmogorov entropy:

[0095]

[0096] Among them, T satisfies Where N is the set original sequence length.

[0097] In a specific embodiment, the fitness function of the ratio of the mean to the variance of the multi-scale Kolmogorov entropy is specifically:

[0098]

[0099] Among them, Fitness is the fitness function, Mean is the mean, and Var is the variance.

[0100] In a specific embodiment, the starfish optimization search algorithm is used to optimize the decomposition mode number K and the penalty factor α of the variational mode decomposition. The specific steps are:

[0101] Select the value range of K and α, and initialize K and α;

[0102] Initialize the SFOA optimization algorithm: determine the population size, that is, the number of candidate solutions in the algorithm, and set the maximum number of iterations;

[0103] The initialized [K, α] parameters are used as the input of variational mode decomposition to decompose the given signal;

[0104] Iteratively calculate the fitness value of the intrinsic mode component obtained by decomposition; if the current fitness value is less than the fitness value of the previous iteration, the current fitness value is used as the new fitness value; if the current fitness value is not less than the fitness value of the previous iteration, the fitness value remains unchanged; repeat the iteration until the preset maximum number of iterations t is reached, and the optimal fitness value and its corresponding parameters [K, α] are obtained.

[0105] In this embodiment, this process is achieved by continuously adjusting and optimizing the parameters of VMD in order to obtain the optimal signal decomposition result.

[0106] In a specific embodiment, variational modal decomposition is performed on the millimeter-wave radar signal based on the optimized K and α to obtain multiple eigenmode components. The specific steps are as follows:

[0107] Decompose the millimeter-wave radar signal into K IMF intrinsic mode functions u k (t);

[0108] Use Hilbert transform to transform each modal function u k (t) Analyze and obtain the corresponding analytical function;

[0109]

[0110] Utilization Index Correction, so that the spectrum of each modal component is modulated to the corresponding baseband;

[0111]

[0112] The bandwidth of each modal function is obtained by calculating the square L2 norm of the gradient of the analytical function;

[0113]

[0114] The constrained variational mode decomposition model is specifically:

[0115]

[0116] Where, define u k ={u1,u2,...,u k} is the modal component, ω k ={ω1,ω2,...,ω k} is u k The center frequency of , * is the convolution operation, and δ(t) refers to the unit impulse function. Represents the partial derivative of the function with respect to time t;

[0117] Considering the penalty factor α, the constrained variational mode decomposition model is actually:

[0118]

[0119] Where λ is the Lagrange multiplier.

[0120] In a specific embodiment, the sample entropies of each modality are arranged in ascending order, and a mutation point β is introduced. When calculating the sample entropy, the sample entropies of two adjacent IMFs are compared to find the mutation point between the two. That is, the maximum absolute value obtained by subtracting the two adjacent IMFs is the mutation point β:

[0121] β=argmax|K(IMF L+1 )-K(IMF L )|

[0122] Among them, argmax is the maximum function, S is the sample entropy, IMFL is the Lth mode;

[0123] The mutation point β is taken as the modal position with the largest mutation amount, the ones before β are signal modes, and the other sample entropies are noise modes.

[0124] In a specific embodiment, wavelet threshold denoising is performed on the noise mode, specifically: for the signal classified as the noise mode, a wavelet denoising algorithm is used to remove the noise by decomposing the signal into wavelet subspaces of different frequencies and utilizing the sparsity and statistical properties of the wavelet coefficients.

[0125] In this embodiment, after the signal mode and the denoised noise mode are obtained, the reconstructed signal is obtained by adding all the modes.

[0126] Example 3

[0127] In this embodiment, the proposed algorithm is used to denoise the simulation signal and the millimeter-wave radar gas leakage signal. The simulation signal consists of a sine signal, a cosine signal, and a Gaussian noise signal. The three signals are assigned different frequencies, namely 5 Hz, 60 Hz, and 180 Hz. The sampling frequency of the signal is set to 1 kHz, the number of sampling points is set to 1000, and η is Gaussian white noise with different input signal-to-noise ratios. The simulation signal is specifically:

[0128] f(t)=cos(2π·5t)+1.2sin(2π·60t)+0.7cos(2π·180t)+η

[0129] like Figure 3 The SFOA shown in the figure can correctly separate the signal and noise, where IMF1 is the added Gaussian white noise. The denoised signal time domain waveform obtained after the signal is decomposed, denoised and reconstructed is as follows: Figure 4 The millimeter wave radar gas leakage signal is decomposed by SFOA-VMD and the waveforms of each mode are as follows. Figure 5 As shown, the denoised signal time domain waveform obtained after denoising and reconstruction is as follows Figure 6 shown.

[0130] This invention addresses the problem that existing methods for millimeter-wave radar gas leakage signals suffer from excessive noise components, making it ineffective for removing noise and improving the signal-to-noise ratio (SNR) of these signals. This method uses SFOA to optimize VMD, employs the ratio of the mean to variance of the multiscale Kolmogorov entropy as the fitness function, classifies the decomposed modes using sample entropy, performs wavelet denoising on the noise modes, and finally reconstructs the signal. The reconstructed signal achieves a 5.6306dB SNR improvement, a 0.0399 mean average error (MAE), and a 0.1583 mean absolute error (MAE).

[0131] Example 4

[0132] A system based on millimeter-wave radar signals, comprising a cascaded signal acquisition module, a fitness function module, a factor optimization module, a variational mode decomposition module, a classification module, a denoising module, and a signal output module;

[0133] The signal acquisition module is used to collect the millimeter-wave radar signal to be denoised;

[0134] The fitness function module is used to construct a fitness function based on the ratio of the mean to the variance of the multi-scale Kolmogorov entropy based on the millimeter wave radar signal;

[0135] The factor optimization module is used to optimize the decomposition mode number K and penalty factor α of variational mode decomposition using the starfish optimization search algorithm;

[0136] The variational mode decomposition module is used to calculate the sample entropy of each eigenmode component, and sort the sample entropy values ​​from small to large to determine the modal position with the largest mutation amount. The eigenmode components with sample entropy less than that position are regarded as signal modes, and the rest are regarded as noise modes.

[0137] The classification module is used to perform variational modal decomposition on the millimeter wave radar signal based on the optimized K and α to obtain multiple eigenmode components;

[0138] The denoising module is used to perform wavelet threshold denoising on the noise mode;

[0139] The signal output module is used to reconstruct the signal and output the radar signal after noise reduction.

[0140] Obviously, the above embodiments of the present invention are merely examples for the purpose of illustrating the present invention, and are not intended to limit the embodiments of the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the claims of the present invention.

Claims

1. A noise reduction method based on millimeter wave radar signals, characterized by: The following steps are involved: Collect millimeter-wave radar signals to be denoised; Based on millimeter-wave radar signals, a fitness function is constructed based on the ratio of the mean to the variance of the multi-scale Kolmogorov entropy. The starfish optimization search algorithm is used to optimize the decomposition mode number K and penalty factor α of variational mode decomposition; Based on the optimized K and α, the millimeter-wave radar signal is subjected to variational mode decomposition to obtain multiple eigenmode components; Calculate the sample entropy of each eigenmode component and sort them from small to large according to the sample entropy value. Determine the modal position with the largest sample entropy mutation. The eigenmode components with sample entropy less than that position are regarded as signal modes, and the rest are regarded as noise modes. Perform wavelet threshold denoising on noise modes; The signal mode and the denoised noise mode are used to reconstruct the signal and output the denoised radar signal.

2. The method for noise reduction based on millimeter-wave radar signals according to claim 1, wherein: The millimeter-wave radar signal to be denoised is collected. The specific steps are as follows: using the millimeter-wave radar to transmit a frequency-modulated continuous wave signal to monitor the target container; the frequency-modulated continuous wave signal is used to monitor the vibration of the target through linear frequency modulation; The signal emitted by the radar generates an echo when it encounters a container or leaking gas. The receiving antenna captures the echo signal and obtains the millimeter-wave radar signal to be denoised.

3. The method for noise reduction based on millimeter-wave radar signals according to claim 1, wherein: The multi-scale Kolmogorov entropy, wherein the specific expression of Kolmogorov entropy is: Where n is the number of millimeter wave radar signal information, P i Indicates that information X appears at time τ i probability.

4. The method for noise reduction based on millimeter-wave radar signals according to claim 3, wherein: Solve the multi-scale Kolmogorov entropy value. The specific steps are: Reconstruct the time series of millimeter-wave radar signals in phase space to obtain an m-dimensional vector space point set R; Calculate the two points r in phase space ij The Euclidean distance C(m,r ij )value; Reduce the value of r until C(m,r ij ) remains unchanged, then C(m,r ij ) is recorded as C(m,r); Change the value of m to calculate the corresponding C(mx,r), and set the scale factor T to update the sequence. The Kolmogorov entropy of the new sequence is expressed as: Calculate the multiscale Kolmogorov entropy: Among them, T satisfies Where N is the set original sequence length.

5. The method for noise reduction based on millimeter-wave radar signals according to claim 4, characterized in that: The fitness function is the ratio of the mean and variance of the multi-scale Kolmogorov entropy, specifically: Among them, Fitness is the fitness function, Mean is the mean, and VAR is the variance.

6. The method for noise reduction based on millimeter-wave radar signals according to claim 5, characterized in that: The starfish optimization search algorithm is used to optimize the decomposition mode number K and penalty factor α of variational mode decomposition. The specific steps are as follows: Select the value range of K and α, and initialize K and α; Initialize the SFOA optimization algorithm: determine the population size, that is, the number of candidate solutions in the algorithm, and set the maximum number of iterations; The initialized [K, α] parameters are used as the input of variational mode decomposition to decompose the given signal; Iteratively calculate the fitness value of the intrinsic mode component obtained by decomposition; if the current fitness value is less than the fitness value of the previous iteration, the current fitness value is used as the new fitness value; if the current fitness value is not less than the fitness value of the previous iteration, the fitness value remains unchanged; repeat the iteration until the preset maximum number of iterations t is reached, and the optimal fitness value and its corresponding parameters [K, α] are obtained.

7. The method for noise reduction based on millimeter-wave radar signals according to claim 1, wherein: Based on the optimized K and α, the millimeter-wave radar signal is subjected to variational mode decomposition to obtain multiple intrinsic mode components. The specific steps are as follows: Decompose the millimeter-wave radar signal into K IMF intrinsic mode functions u k (t); Use Hilbert transform to transform each modal function u k (t) Analyze and obtain the corresponding analytical function; Utilization Index Correction, so that the spectrum of each modal component is modulated to the corresponding baseband; The bandwidth of each modal function is obtained by calculating the square L2 norm of the gradient of the analytical function; The constrained variational mode decomposition model is specifically: Where, define u k ={u1,u2,...,u k } is the modal component, ω k ={ω1,ω2,...,ω k } is u k The center frequency of , * is the convolution operation, and δ(t) refers to the unit impulse function. Represents the partial derivative of the function with respect to time t; Considering the penalty factor α, the constrained variational mode decomposition model is actually: Where λ is the Lagrange multiplier.

8. The method for noise reduction based on millimeter-wave radar signals according to claim 7, characterized in that: When the sample entropy of each modality is arranged in ascending order, the maximum absolute value of the difference between the entropies of two adjacent samples is taken as the mutation point β: β=argmax|S(IMF L+1 )-S(IMF L )| Among them, argmax is the maximum absolute value function, S is the sample entropy, IMF L is the Lth mode; The mutation point β is taken as the modal position with the largest mutation amount, the ones before β are signal modes, and the other sample entropies are noise modes.

9. The method for noise reduction based on millimeter-wave radar signals according to claim 1, wherein: The noise mode is denoised by wavelet threshold. Specifically, for the signal classified as the noise mode, the wavelet denoising algorithm is used to decompose the signal into wavelet subspaces of different frequencies and remove the noise by utilizing the sparsity and statistical characteristics of the wavelet coefficients.

10. A system based on millimeter wave radar signals, characterized in that: It includes a cascaded signal acquisition module, a fitness function module, a factor optimization module, a variational mode decomposition module, a classification module, a denoising module, and a signal output module; The signal acquisition module is used to collect the millimeter-wave radar signal to be denoised; The fitness function module is used to construct a fitness function based on the ratio of the mean to the variance of the multi-scale Kolmogorov entropy based on the millimeter wave radar signal; The factor optimization module is used to optimize the decomposition mode number K and penalty factor α of variational mode decomposition using the starfish optimization search algorithm; The variational mode decomposition module is used to calculate the sample entropy of each eigenmode component, and sort the sample entropy values ​​from small to large to determine the modal position with the largest mutation amount. The eigenmode components with sample entropy less than that position are regarded as signal modes, and the rest are regarded as noise modes. The classification module is used to perform variational modal decomposition on the millimeter wave radar signal based on the optimized K and α to obtain multiple eigenmode components; The denoising module is used to perform wavelet threshold denoising on the noise mode; The signal output module is used for signal reconstruction and outputting the radar signal after noise reduction.

Images