A method and system for denoising and reconstructing vital signs signals from ultra-wideband radar

Abstract

This invention relates to a method and system for denoising and reconstructing vital signs signals from ultra-wideband radar. The method includes: S1, preprocessing the baseband signal of the vital signs echo to obtain a mixed signal of vital signs to be decomposed; S2, globally adaptively optimizing the core parameters of the ICEEMDAN algorithm based on DBO to obtain the optimal parameter combination; S3, substituting the optimal parameter combination into the ICEEMDAN algorithm to adaptively decompose the mixed signal of vital signs to be decomposed, obtaining multiple IMF components; S4, selecting effective components of the IMF components based on the frequency band characteristics of human physiological signals and Pearson correlation analysis; S5, adaptively weighted fusing and reconstructing the effective IMF components of respiration and heartbeat respectively, outputting respiratory vital signs signals and heartbeat vital signs signals. Combining the DBO algorithm with the ICEEMDAN algorithm significantly improves the algorithm's scene adaptability and decomposition stability.

Description

Technical Field

[0001] This invention belongs to the field of ultra-wideband radar vital sign monitoring and digital signal processing technology, and particularly relates to a method and system for denoising and reconstructing ultra-wideband radar vital sign signals. Background Technology

[0002] With the accelerating global aging process and the rapid growth in demand for telemedicine and emergency rescue, non-contact vital sign monitoring technology has become a research focus in the fields of biomedical engineering and radar signal processing. Ultra-wideband (UWB) radar, with its advantages of strong penetration, high range resolution, low power consumption, and no radiation damage, can capture subtle physiological vibrations in the human chest caused by breathing and heartbeat under non-contact and unrestricted conditions. This overcomes the application limitations of traditional contact-based ECG and pulse oximetry devices, which require close-fitting contact, and demonstrates extremely high application value in scenarios such as home health monitoring, post-disaster search and rescue, and non-intrusive monitoring in intensive care units.

[0003] During UWB radar vital sign monitoring, the amplitude of chest vibrations caused by respiration and heartbeat differs greatly. Respiratory vibration amplitude is approximately 0.1–1 cm, while heartbeat vibration amplitude is only 0.01–0.1 cm. This results in the echo heartbeat signal amplitude being much weaker than the respiratory signal, making it highly susceptible to interference from respiratory harmonics, environmental clutter, and Gaussian white noise. This leads to frequency coupling and aliasing of respiratory and heartbeat signals, significantly increasing the difficulty of extracting effective physiological signals. Therefore, achieving effective denoising and high-fidelity reconstruction of vital sign signals against strong noise backgrounds is a core technical challenge in UWB radar non-contact vital sign monitoring.

[0004] Among existing methods for denoising and decomposing vital signs signals from UWB radar, Empirical Mode Decomposition (EMD) can adaptively handle nonlinear and non-stationary physiological signals, but it is prone to severe mode aliasing in low signal-to-noise ratio scenarios, resulting in the inability to effectively separate effective signals from noise. Ensemble Empirical Mode Decomposition (EEMD) alleviates the mode aliasing problem by introducing Gaussian white noise into the original signal, but its decomposition accuracy is significantly affected by noise characteristics and it suffers from residual reconstructed noise. Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDA) further optimizes the noise addition strategy, reducing the probability of mode aliasing and reconstruction error, but it still does not completely solve the problem of parameter adaptive adaptation in complex scenarios.

[0005] The Improved Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (ICEEMDAN) has significant advantages over the aforementioned algorithms in terms of decomposition stability and mode aliasing suppression, and has become one of the mainstream methods in the field of vital sign signal processing. However, the decomposition effect of the ICEEMDAN algorithm is highly dependent on the selection of two core parameters: the white noise amplitude weight ε and the number of noise additions N. If ε is too small, it will easily cause mode aliasing, while if it is too large, it will reduce the decomposition accuracy; if N is too small, it will lead to insufficient signal decomposition, while if it is too large, it will significantly increase the computational load. In the existing technology, the above parameters all rely on manual experience for adjustment and lack an adaptive selection mechanism. It is impossible to dynamically match the optimal parameter combination according to the signal characteristics under different scenarios, making it difficult to guarantee the decomposition effect and signal reconstruction accuracy in complex environments.

[0006] Furthermore, while Variational Mode Decomposition (VMD) can theoretically avoid mode aliasing, the combination of its mode number K and penalty factor α still relies on manual experience for setting, resulting in insufficient adaptability to non-stationary vital sign signals. Simple filtering-based denoising methods cannot effectively separate respiratory and heartbeat signals, and their denoising effect is very limited in noisy environments. Therefore, developing a method that can adaptively optimize ICEEMDAN parameters, effectively suppress mode aliasing, and achieve high signal-to-noise ratio vital sign signal reconstruction has significant theoretical and engineering application value. Summary of the Invention

[0007] To address the aforementioned technical problems, this invention provides a method and system for denoising and reconstructing vital signs signals from ultra-wideband radar.

[0008] The technical solution adopted in this invention is: Firstly, a method for denoising and reconstructing vital sign signals from ultra-wideband radar is provided, including: S1, acquire the baseband signal of vital signs echo collected by ultra-wideband radar, preprocess the vital signs echo baseband signal, and extract the one-dimensional slow time domain mixed signal of vital signs to be decomposed. S2, based on the Dung Beetle Optimizer (DBO) algorithm, performs global adaptive optimization of the core parameters of the ICEEMDAN algorithm to obtain the optimal parameter combination; S3, Substitute the optimal parameter combination into the ICEEMDAN algorithm to adaptively decompose the mixed signal of vital signs to be decomposed, and obtain multiple IMF components arranged from high to low frequency. S4. Based on the frequency band characteristics of human physiological signals and Pearson correlation analysis, the effective components of IMF are screened to obtain the effective IMF components of the corresponding respiratory signals and the effective IMF components of the heartbeat signals. S5 performs adaptive weighted fusion reconstruction on the effective IMF components of the respiratory signal and the effective IMF components of the heartbeat signal, and outputs respiratory vital signs signals and heartbeat vital signs signals.

[0009] Further, S1, acquire the baseband signal of vital signs echo from the ultra-wideband radar, preprocess the vital signs echo baseband signal, and extract the one-dimensional slow-time domain mixed signal of vital signs to be decomposed, including: S1.1, Read the baseband signal of vital signs echo acquired by the ultra-wideband radar, and separate the in-phase component I and the quadrature component in the vital signs echo baseband signal. Construct a complex signal matrix Complex signal matrix The behavior is a distance gate with a fixed length value, and the columns are slow-time sampling points with fixed time intervals, matrix elements. This represents the signal value at the Xth distance gate and the Yth slow-time sampling point; S1.2, for Perform SVD decomposition. The SVD decomposition formula is: ; Where U is the left singular matrix and S is the singular value diagonal matrix. To obtain the static clutter matrix, we take the conjugate transpose of the right singular matrix V, set all singular values ​​except the first maximal singular value in both the left singular matrix U and the right singular matrix V to zero, and then reconstruct the static clutter matrix. The clean signal matrix after clutter suppression is calculated. ; S1.3, for the clean signal matrix Median filtering is performed on all signal values ​​of distance gates within 0.1m, and the energy sum of all signal values ​​corresponding to each distance gate in the range of 0.1m to 1.0m is calculated. The distance gate with the largest energy sum is selected as the target distance gate corresponding to the micro-movement of the human chest cavity. S1.4, from the clean signal matrix Extract the entire row of signal values ​​corresponding to the target distance gate and use them as a slow-time one-dimensional signal; S1.5 performs phase demodulation, DC component removal, amplitude normalization, and physiological bandpass filtering on the slow-time one-dimensional signal to obtain the mixed signal of vital signs to be decomposed, x(t).

[0010] Furthermore, the core parameters of the ICEEMDAN algorithm include white noise amplitude weights. and the number of times noise was added ; S2, based on the Dung Beetle Optimization (DBO) algorithm, performs global adaptive optimization of the core parameters of the ICEEMDAN algorithm to obtain the optimal parameter combination, including: S2.1, Pre-set white noise amplitude weights and the number of times noise was added The search range is determined, and the population size and maximum number of iterations of the Dymplocan optimization algorithm (DBO) are initialized. S2.2, Randomly generate the initial positions of the dung beetle population within the optimization range, with each initial position corresponding to a set of parameter combinations. ; S2.3, Substitute each set of parameters into the ICEEMDAN algorithm to decompose the mixed signal of vital signs to be decomposed, and obtain multiple IMF components; S2.4 Construct a fitness function and calculate the fitness value of each dung beetle population with the goal of minimizing the envelope entropy value of each IMF component. Record the global optimum and local optimum positions. S2.5, simulate dung beetle behavior to iteratively update the global and local optimal positions of the population; S2.6, Determine if the population iteration has reached the maximum number of iterations. If not, return to step S2.5 to continue iteration; if it has, stop iteration and output the optimal parameter combination corresponding to the final global optimum and local optimum. .

[0011] Furthermore, the expression for the fitness function is: ; in, Let k be the fitness value, and k be the total number of IMF components obtained from the decomposition of the mixed vital sign signal x(t). Let be the envelope entropy value of the i-th IMF component; The envelope weighting coefficient is... , The values ​​of are linearly distributed from low to high frequencies of the k IMF components; This is a parameter boundary penalty term, applied when the parameter combination... Triggered when parameters exceed the optimization range. When parameter combinations When there are no parameters outside the optimization range, The value of is 0; The calculation formula is: ; Where L is the number of sampling points of the mixed signal of vital signs x(t) to be decomposed. To preset the minimum value, Hilbert envelope of the i-th IMF component The normalized probability distribution.

[0012] Furthermore, dung beetle behavior includes rolling behavior, reproductive behavior, foraging behavior, and predator avoidance behavior; The expression for population iterative updates based on the rolling ball behavior is: ; in, For the first The position of an individual dung beetle in the (v+1)th iteration. For the first The individual position of a dung beetle population in the current iteration v. This is the worst position for the current population. This represents the globally optimal position for the current population. For direction coefficients, ; The expression for population iteration and updating based on reproductive behavior is: ; in, This represents the local optimum position for the current population. A random vector with 1 row and 1 column. Each element in the set has a value range of (0, 1). This represents the maximum number of iterations.

[0013] Furthermore, in S3, the optimal parameter combination is substituted into the ICEEMDAN algorithm to adaptively decompose the mixed signal of vital signs to be decomposed, resulting in multiple IMF components arranged from high to low frequency, including:

[0014] Combine the optimal parameters Substituting into the ICEEMDAN algorithm, we perform adaptive decomposition on the mixed signal x(t) of vital signs to be decomposed. The expression for adaptive decomposition is: ; in, Let x(t) be the d-th IMF component obtained from the decomposition of the mixed signal of vital signs to be decomposed. K represents the final residual components, and K is the total number of IMF components obtained from the decomposition of the mixed vital signs signal x(t). Arrange the K IMF components in descending order of frequency.

[0015] Furthermore, in S4, based on the frequency band characteristics of human physiological signals and Pearson correlation analysis, effective IMF components are screened to obtain the effective IMF components of the corresponding respiratory signals and heartbeat signals, including: S4.1, Based on the frequency band characteristics of human physiological signals, perform spectral analysis on each IMF component, extract the peak frequency, and screen out the respiratory candidate IMF components with peak frequencies located in the respiratory physiological frequency band of 0.1Hz~0.6Hz and the heartbeat candidate IMF components located in the heartbeat physiological frequency band of 0.8Hz~2.5Hz. S4.2, calculate the Pearson correlation coefficients of the respiratory candidate IMF component, the heartbeat candidate IMF component, and the mixed signal x(t) of the vital signs to be decomposed, respectively. The formula for calculating the Pearson correlation coefficient is: ; in, For covariance, Let h be the standard deviation of the h-th candidate IMF component for respiration or heartbeat. Let x(t) be the standard deviation. S4.3, Pearson correlation coefficients between respiratory candidate IMF components and heartbeat candidate IMF components. In the middle, Pearson correlation coefficients were selected respectively. The highest C1 respiratory candidate IMF components and C2 heartbeat candidate IMF components are used as the effective IMF components of the respiratory signal and the effective IMF components of the heartbeat signal, respectively. C1 is the initial respiratory selection value and C2 is the initial heartbeat selection value.

[0016] Furthermore, the calculation formula for adaptive weighted fusion reconstruction is as follows: ; Where C is either C1 or C2, The c-th effective IMF component of the respiratory or heartbeat signal. This refers to the adaptively weighted fusion and reconstruction of respiratory vital signs or cardiac vital signs signals. For adaptive weighting coefficients; adaptive weighting coefficients The calculation formula is: ; in, This represents the energy percentage of the c-th effective IMF component of the respiratory or heartbeat signal within the target physiological frequency band. is the Pearson correlation coefficient between the c-th effective IMF component and the mixed signal x(t) of the vital signs to be decomposed.

[0017] Furthermore, the methods also include: Calculate the reconstructed respiratory rate from respiratory vital signs signals and obtain the measured respiratory rate using a human contact device; Determine whether the difference between the reconstructed respiratory rate and the measured respiratory rate exceeds a preset difference threshold; If the difference does not exceed the preset threshold, the respiratory vital signs signal is determined to meet the accuracy requirements. If the difference exceeds the preset threshold, the first respiratory selection value is obtained by adjusting the initial respiratory selection value C1 by C1±1. ,according to Return to step S4.3 to obtain... The first respiratory candidate IMF component is then processed, and step S5 is executed to obtain the first adjusted first respiratory signal. When the difference between the reconstructed respiratory rate of the first respiratory signal and the measured respiratory rate does not exceed a preset difference threshold, the first respiratory signal is determined to meet the accuracy requirements. When the difference between the reconstructed respiratory rate and the measured respiratory rate of the first respiratory signal exceeds a preset difference threshold, the value selected in the first respiratory signal is... Based on Adjust to obtain the second breath selection value ,according to Return to step S4.3 to obtain... The second respiratory signal is obtained by performing step S5 after identifying the candidate IMF components for each breath. Until the The adjusted number The respiratory signal meets the accuracy requirements. Less than or equal to the preset number of adjustments.

[0018] Secondly, a system for denoising and reconstructing ultra-wideband radar vital sign signals is provided, including: The mixed signal extraction module is used to acquire the baseband signal of vital signs echo collected by ultra-wideband radar, preprocess the vital signs echo baseband signal, and extract the one-dimensional slow time domain mixed signal of vital signs to be decomposed. The parameter optimization module is used to perform global adaptive optimization of the core parameters of the ICEEMDAN algorithm based on the Dung Beetle Optimization Algorithm (DBO) to obtain the optimal parameter combination. The adaptive signal decomposition module is used to substitute the optimal parameter combination into the ICEEMDAN algorithm to adaptively decompose the mixed vital signs signal to be decomposed, and obtain multiple IMF components arranged from high to low frequency. The effective component screening module is used to screen the effective components of IMF based on the frequency band characteristics of human physiological signals and Pearson correlation analysis, and obtain the effective IMF components of the corresponding respiratory signals and heartbeat signals respectively. The adaptive weighted fusion reconstruction module is used to adaptively weightedly fused and reconstruct the effective IMF components of the respiratory signal and the effective IMF components of the heartbeat signal, and output respiratory vital signs signals and heartbeat vital signs signals.

[0019] The beneficial effects achieved by this invention are as follows: By combining the Dung Beetle Optimization (DBO) algorithm with the ICEEMDAN algorithm, and taking the minimization of the envelope entropy of the fitness function as the optimization objective, the global adaptive optimization of the core parameters of the ICEEMDAN algorithm is realized. This breaks through the technical bottleneck of the traditional ICEEMDAN algorithm relying on manual experience for parameter tuning. It can dynamically match the optimal parameter combination according to the characteristics of radar echo signals in different scenarios, which greatly improves the algorithm's scenario adaptability and decomposition stability. The combination of the Dung Beetle Optimization (DBO) algorithm and the ICEEMDAN algorithm achieves adaptive decomposition of mixed vital signs signals, effectively suppressing the mode aliasing problem of traditional EMD-type algorithms. By combining the frequency band characteristics of human physiological signals with Pearson correlation analysis, the effective IMF components of respiratory and heartbeat signals are accurately selected. With the help of an adaptive weighted fusion reconstruction strategy, the effective separation and high-fidelity reconstruction of respiratory and heartbeat signals under strong noise background are achieved, significantly improving the signal-to-noise ratio of the reconstructed signal and the accuracy of vital sign parameter extraction. The preprocessing workflow of the vital signs echo baseband signal combines SVD static clutter filtering, target range gate energy screening, phase demodulation and bandpass filtering to effectively remove static clutter, equipment self-reflection interference and out-of-band noise in the echo baseband signal, providing a high-quality input signal for subsequent decomposition and reconstruction. In summary, the ultra-wideband radar vital sign signal denoising and reconstruction method and system of the present invention, while ensuring signal processing accuracy, have good engineering applicability and can provide a highly robust technical solution for UWB radar non-contact vital sign monitoring in complex environments. Attached Figure Description

[0020] Figure 1 This is a flowchart of the ultra-wideband radar vital sign signal denoising and reconstruction method of the present invention; Figure 2 This is a structural diagram of the ultra-wideband radar vital sign signal denoising and reconstruction system of the present invention. Detailed Implementation

[0021] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and should not be used to limit the scope of protection of the present invention.

[0022] like Figure 1 As shown, this invention provides a method for denoising and reconstructing ultra-wideband radar vital sign signals, comprising: S1, acquire the baseband signal of vital signs echo collected by ultra-wideband radar, preprocess the vital signs echo baseband signal, and extract the one-dimensional slow time domain mixed signal of vital signs to be decomposed. In this embodiment, the baseband signal of vital signs of personnel in the monitoring scene is collected by ultra-wideband radar. The ultra-wideband radar adopts a pulse ultra-wideband system, and its center frequency, bandwidth, sampling rate and detection range can be adapted and adjusted in advance according to the actual monitoring scene.

[0023] S1.1, Read the baseband signal of vital signs echo acquired by the ultra-wideband radar, and separate the in-phase component I and the quadrature component in the vital signs echo baseband signal. Construct a complex signal matrix Complex signal matrix The behavior is a distance gate with a fixed length value, and the columns are slow-time sampling points with fixed time intervals, matrix elements. This represents the signal value at the Xth distance gate and the Yth slow-time sampling point; S1.2, for Perform singular value decomposition (SVD). The SVD decomposition formula is: ; Where U is the left singular matrix and S is the singular value diagonal matrix. To obtain the static clutter matrix, we take the conjugate transpose of the right singular matrix V, set all singular values ​​except the first maximal singular value in both the left singular matrix U and the right singular matrix V to zero, and then reconstruct the static clutter matrix. The clean signal matrix after clutter suppression is calculated. The purpose of SVD decomposition is to eliminate static clutter interference caused by reflections from stationary objects within the monitoring scene. S1.3, for the clean signal matrix Median filtering is applied to all signal values ​​within a range of 0.1m, and the energy sum of all signal values ​​corresponding to each range gate in the range of 0.1m to 1.0m is calculated. The range gate with the largest energy sum is selected as the target range gate corresponding to the micro-movement of the human chest cavity. Median filtering of range gates within 0.1m can suppress self-reflection clutter of radar equipment. S1.4, from the clean signal matrix Extract the entire row of signal values ​​corresponding to the target distance gate and use them as a slow-time one-dimensional signal; S1.5 performs phase demodulation, DC component removal, amplitude normalization, and physiological bandpass filtering on the slow-time one-dimensional signal to obtain the mixed signal of vital signs to be decomposed, x(t).

[0024] Physiological bandpass filtering typically involves using a Butterworth bandpass filter (0.1Hz~2.5Hz) to remove out-of-band noise.

[0025] S2, Based on the Dung Beetle Optimization Algorithm (DBO), the core parameters of the ICEEMDAN algorithm are globally adaptively optimized to obtain the optimal parameter combination; In this embodiment, the core parameters of the ICEEMDAN algorithm include the white noise amplitude weight. and the number of times noise was added .

[0026] S2.1, Pre-set white noise amplitude weights and the number of times noise was added The search range is determined, and the population size and maximum number of iterations of the Dymplocan optimization algorithm (DBO) are initialized. Specifically, white noise amplitude weighting The optimization range is 0.1~0.5, and the number of noise additions is [not specified]. The optimization range is 20~250, the initial population size of DBO is 40, and the maximum number of iterations is 30.

[0027] S2.2, Randomly generate the initial positions of the dung beetle population within the optimization range, with each initial position corresponding to a set of parameter combinations. ; In step S2.1, the initial positions of the dung beetle population are randomly generated within the preset optimization range, and each initial position corresponds to a set of parameter combinations. .

[0028] S2.3, Substitute each set of parameters into the ICEEMDAN algorithm to decompose the mixed signal of vital signs to be decomposed, and obtain multiple IMF components; The detailed process of this step is described in detail in step S3.

[0029] S2.4 Construct a fitness function and calculate the fitness value of each dung beetle population with the goal of minimizing the envelope entropy value of each IMF component. Record the global optimum and local optimum positions. The fitness function is expressed as follows: ; in, Let k be the fitness value, and k be the total number of IMF components obtained from the decomposition of the mixed vital sign signal x(t). Let be the envelope entropy value of the i-th IMF component; The envelope weighting coefficient is... , The values ​​of are linearly distributed from low to high frequencies of the k IMF components; This is a parameter boundary penalty term, applied when the parameter combination... Triggered when parameters exceed the optimization range. When parameter combinations When there are no parameters outside the optimization range, The value of is 0; The calculation formula is: ; Where L is the number of sampling points of the mixed signal of vital signs x(t) to be decomposed. The purpose of setting a minimum value is to avoid making logarithmic operations meaningless. Hilbert envelope of the i-th IMF component The normalized probability distribution.

[0030] That is, the envelope entropy value of each IMF component. Minimize the fitness of individual dung beetles in the population as the objective, calculate the fitness value of each individual based on the fitness function, and record the current global optimum and local optimum.

[0031] S2.5, simulate dung beetle behavior to iteratively update the global and local optimal positions of the population; Dung beetle behavior includes rolling behavior, reproductive behavior, foraging behavior, and predator avoidance behavior; The expression for population iterative updates based on the rolling ball behavior is: ; in, For the first The position of an individual dung beetle in the (v+1)th iteration. For the first The individual position of a dung beetle population in the current iteration v. This is the worst position for the current population. This represents the globally optimal position for the current population. For direction coefficients, ; The expression for population iteration and updating based on reproductive behavior is: ; in, This represents the local optimum position for the current population. A random vector with 1 row and 1 column. Each element in the set has a value range of (0, 1). This represents the maximum number of iterations.

[0032] Foraging and predator avoidance behavior updates: The position of individual dung beetles in the population is updated by combining the global optimal position with Gaussian random perturbation. The updated position must be constrained within the parameter optimization range.

[0033] S2.6, Determine if the population iteration has reached the maximum number of iterations. If not, return to step S2.5 to continue iteration; if it has, stop iteration and output the optimal parameter combination corresponding to the final global optimum and local optimum. .

[0034] S3, Substitute the optimal parameter combination into the ICEEMDAN algorithm to adaptively decompose the mixed signal of vital signs to be decomposed, and obtain multiple IMF components arranged from high to low frequency. In this embodiment, the optimal parameter combination obtained in step S2.6 above is used. Substituting into the ICEEMDAN algorithm, the mixed signal of vital signs to be decomposed, x(t), is adaptively decomposed. The expression for the adaptive decomposition is: ; in, Let x(t) be the d-th IMF component obtained from the decomposition of the mixed signal of vital signs to be decomposed. K represents the final residual components, and K is the total number of IMF components obtained from the decomposition of the mixed vital signs signal x(t). Arrange the K IMF components in descending order of frequency.

[0035] S4. Based on the frequency band characteristics of human physiological signals and Pearson correlation analysis, the effective components of IMF are screened to obtain the effective IMF components of the corresponding respiratory signals and the effective IMF components of the heartbeat signals. In this embodiment, the process of filtering effective components from IMF components specifically includes: S4.1, Based on the frequency band characteristics of human physiological signals, perform spectral analysis on each IMF component, extract the peak frequency, and screen out the respiratory candidate IMF components with peak frequencies located in the respiratory physiological frequency band of 0.1Hz~0.6Hz and the heartbeat candidate IMF components located in the heartbeat physiological frequency band of 0.8Hz~2.5Hz. S4.2, calculate the Pearson correlation coefficients of the respiratory candidate IMF component, the heartbeat candidate IMF component, and the mixed signal x(t) of the vital signs to be decomposed, respectively. The formula for calculating the Pearson correlation coefficient is: ; in, For covariance, Let h be the standard deviation of the h-th candidate IMF component for respiration or heartbeat. Let x(t) be the standard deviation. S4.3, Pearson correlation coefficients between respiratory candidate IMF components and heartbeat candidate IMF components. In the middle, Pearson correlation coefficients were selected respectively. The highest C1 respiratory candidate IMF components and C2 heartbeat candidate IMF components are used as the effective IMF components of the respiratory signal and the effective IMF components of the heartbeat signal, respectively. C1 is the initial respiratory selection value and C2 is the initial heartbeat selection value.

[0036] S5 performs adaptive weighted fusion reconstruction on the effective IMF components of the respiratory signal and the effective IMF components of the heartbeat signal, and outputs respiratory vital signs signals and heartbeat vital signs signals.

[0037] In this embodiment, the calculation formula for adaptive weighted fusion reconstruction is: ; Where C is either C1 or C2, The c-th effective IMF component of the respiratory or heartbeat signal. This refers to the adaptively weighted fusion and reconstruction of respiratory vital signs or cardiac vital signs signals. For adaptive weighting coefficients; adaptive weighting coefficients The calculation formula is: ; in, This represents the energy percentage of the c-th effective IMF component of the respiratory or heartbeat signal within the target physiological frequency band. is the Pearson correlation coefficient between the c-th effective IMF component and the mixed signal x(t) of the vital signs to be decomposed.

[0038] Based on the above calculation formula for adaptive weighted fusion reconstruction, the effective IMF components of the respiratory signal and heartbeat signal can be adaptively weighted and reconstructed separately, outputting... .

[0039] It should be noted that, in the above Figure 1 In the ultra-wideband radar vital sign signal denoising and reconstruction method described in the illustrated embodiment, the final respiratory vital sign signals and heartbeat vital sign signals are calculated and require further accuracy assessment. The specific process is as follows: Calculate the reconstructed respiratory rate from respiratory vital signs signals and obtain the measured respiratory rate using a human contact device; Determine whether the difference between the reconstructed respiratory rate and the measured respiratory rate exceeds a preset difference threshold; If the difference does not exceed the preset threshold, the respiratory vital signs signal is determined to meet the accuracy requirements. If the difference exceeds the preset threshold, the first respiratory selection value is obtained by adjusting the initial respiratory selection value C1 by C1±1. ,according to Return to step S4.3 to obtain... The first respiratory candidate IMF component is then processed, and step S5 is executed to obtain the first adjusted first respiratory signal. When the difference between the reconstructed respiratory rate of the first respiratory signal and the measured respiratory rate does not exceed a preset difference threshold, the first respiratory signal is determined to meet the accuracy requirements. When the difference between the reconstructed respiratory rate and the measured respiratory rate of the first respiratory signal exceeds a preset difference threshold, the value selected in the first respiratory signal is... Based on Adjust to obtain the second breath selection value ,according to Return to step S4.3 to obtain... The second respiratory signal is obtained by performing step S5 after identifying the candidate IMF components for each breath. Until the The adjusted number The respiratory signal meets the accuracy requirements. Less than or equal to the preset number of adjustments.

[0040] The adjustment process for the initial heart rate selection value C2 is similar to the adjustment process for the initial respiration selection value C1. In practical applications, 1 to 2 effective IMF components are usually selected for respiration, and 2 to 3 effective IMF components are usually selected for heart rate. However, under different signal-to-noise ratio environments, too many or too few effective IMF components are not good. Too few will lead to unevenness, while too many will introduce interference.

[0041] The difference between the reconstructed respiratory rate and the measured respiratory rate is compared with a preset difference threshold. This preset threshold can be a range [-10Hz, 10Hz]. A difference within this range indicates that the accuracy requirement is met; a difference outside the range indicates insufficient accuracy and requires adjustment. Since the number of effective IMF components cannot be zero, the number of adjustments... The limit for the number of preset adjustments should be: .

[0042] like Figure 2 As shown, this embodiment of the invention provides a denoising and reconstruction system for ultra-wideband radar vital sign signals, comprising: The mixed signal extraction module 201 is used to acquire the baseband signal of vital signs echo collected by ultra-wideband radar, preprocess the vital signs echo baseband signal, and extract the one-dimensional slow time domain mixed signal of vital signs to be decomposed. The parameter optimization module 202 is used to perform global adaptive optimization of the core parameters of the ICEEMDAN algorithm based on the Dung Beetle Optimization Algorithm (DBO) to obtain the optimal parameter combination. The adaptive signal decomposition module 203 is used to substitute the optimal parameter combination into the ICEEMDAN algorithm to adaptively decompose the mixed vital signs signal to be decomposed, and obtain multiple IMF components arranged from high to low frequency. The effective component screening module 204 is used to screen the effective components of IMF based on the frequency band characteristics of human physiological signals and Pearson correlation analysis, and obtain the effective IMF components of the corresponding respiratory signals and the effective IMF components of the heartbeat signals respectively. The adaptive weighted fusion reconstruction module 205 is used to perform adaptive weighted fusion reconstruction on the effective IMF components of the respiratory signal and the effective IMF components of the heartbeat signal, and output respiratory vital signs signals and heartbeat vital signs signals.

[0043] The beneficial effects achieved by the embodiments of the present invention are as follows: By combining the Dung Beetle Optimization (DBO) algorithm with the ICEEMDAN algorithm, and taking the minimization of the envelope entropy of the fitness function as the optimization objective, the global adaptive optimization of the core parameters of the ICEEMDAN algorithm is realized. This breaks through the technical bottleneck of the traditional ICEEMDAN algorithm relying on manual experience for parameter tuning. It can dynamically match the optimal parameter combination according to the characteristics of radar echo signals in different scenarios, which greatly improves the algorithm's scenario adaptability and decomposition stability. The combination of the Dung Beetle Optimization (DBO) algorithm and the ICEEMDAN algorithm achieves adaptive decomposition of mixed vital signs signals, effectively suppressing the mode aliasing problem of traditional EMD-type algorithms. By combining the frequency band characteristics of human physiological signals with Pearson correlation analysis, the effective IMF components of respiratory and heartbeat signals are accurately selected. With the help of an adaptive weighted fusion reconstruction strategy, the effective separation and high-fidelity reconstruction of respiratory and heartbeat signals under strong noise background are achieved, significantly improving the signal-to-noise ratio of the reconstructed signal and the accuracy of vital sign parameter extraction. The preprocessing workflow of the vital signs echo baseband signal combines SVD static clutter filtering, target range gate energy screening, phase demodulation and bandpass filtering to effectively remove static clutter, equipment self-reflection interference and out-of-band noise in the echo baseband signal, providing a high-quality input signal for subsequent decomposition and reconstruction. In summary, the ultra-wideband radar vital sign signal denoising and reconstruction method and system of the present invention, while ensuring signal processing accuracy, have good engineering applicability and can provide a highly robust technical solution for UWB radar non-contact vital sign monitoring in complex environments.

[0044] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0045] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0046] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0047] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0048] The above are merely embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention are included within the scope of the claims of the present invention pending approval.

Claims

1. A method for denoising and reconstructing vital sign signals from ultra-wideband radar, characterized in that, include: S1, acquire the baseband signal of vital signs echo collected by ultra-wideband radar, preprocess the vital signs echo baseband signal, and extract the one-dimensional slow time domain mixed signal of vital signs to be decomposed. S2, Based on the Dung Beetle Optimization Algorithm (DBO), the core parameters of the ICEEMDAN algorithm are globally adaptively optimized to obtain the optimal parameter combination; S3, Substitute the optimal parameter combination into the ICEEMDAN algorithm to adaptively decompose the mixed signal of vital signs to be decomposed, and obtain multiple IMF components arranged from high to low frequency. S4. Based on the frequency band characteristics of human physiological signals and Pearson correlation analysis, the effective components of IMF are screened to obtain the effective IMF components of the corresponding respiratory signals and the effective IMF components of the heartbeat signals. S5 performs adaptive weighted fusion reconstruction on the effective IMF components of the respiratory signal and the effective IMF components of the heartbeat signal, and outputs respiratory vital signs signals and heartbeat vital signs signals.

2. The method for denoising and reconstructing ultra-wideband radar vital sign signals according to claim 1, characterized in that, S1, acquire the baseband echo signal of vital signs from the ultra-wideband radar, preprocess the baseband echo signal of vital signs, and extract the one-dimensional slow-time domain mixed signal of vital signs to be decomposed, including: S1.1, Read the baseband signal of vital signs echo acquired by the ultra-wideband radar, and separate the in-phase component I and the quadrature component in the vital signs echo baseband signal. Construct a complex signal matrix Complex signal matrix The behavior is a distance gate with a fixed length value, and the columns are slow-time sampling points with fixed time intervals, matrix elements. This represents the signal value at the Xth distance gate and the Yth slow-time sampling point; S1.2, for Perform SVD decomposition. The SVD decomposition formula is: ； Where U is the left singular matrix and S is the singular value diagonal matrix. To obtain the static clutter matrix, we take the conjugate transpose of the right singular matrix V, set all singular values ​​except the first maximal singular value in both the left singular matrix U and the right singular matrix V to zero, and then reconstruct the static clutter matrix. The clean signal matrix after clutter suppression is calculated. ; S1.3, for the clean signal matrix Median filtering is performed on all signal values ​​of distance gates within 0.1m, and the energy sum of all signal values ​​corresponding to each distance gate in the range of 0.1m to 1.0m is calculated. The distance gate with the largest energy sum is selected as the target distance gate corresponding to the micro-movement of the human chest cavity. S1.4, from the clean signal matrix Extract the entire row of signal values ​​corresponding to the target distance gate and use them as a slow-time one-dimensional signal; S1.5 performs phase demodulation, DC component removal, amplitude normalization, and physiological bandpass filtering on the slow-time one-dimensional signal to obtain the mixed signal of vital signs to be decomposed, x(t).

3. The method for denoising and reconstructing ultra-wideband radar vital sign signals according to claim 1, characterized in that, The core parameters of the ICEEMDAN algorithm include white noise amplitude weights. and the number of times noise was added ; S2, based on the Dung Beetle Optimization (DBO) algorithm, performs global adaptive optimization of the core parameters of the ICEEMDAN algorithm to obtain the optimal parameter combination, including: S2.1, Pre-set white noise amplitude weights and the number of times noise was added The search range is determined, and the population size and maximum number of iterations of the Dymplocan optimization algorithm (DBO) are initialized. S2.2, Randomly generate the initial positions of the dung beetle population within the optimization range, with each initial position corresponding to a set of parameter combinations. ; S2.3, Substitute each set of parameters into the ICEEMDAN algorithm to decompose the mixed signal of vital signs to be decomposed, and obtain multiple IMF components; S2.4 Construct a fitness function and calculate the fitness value of each dung beetle population with the goal of minimizing the envelope entropy value of each IMF component. Record the global optimum and local optimum positions. S2.5, simulate dung beetle behavior to iteratively update the global and local optimal positions of the population; S2.6, Determine if the population iteration has reached the maximum number of iterations. If not, return to step S2.5 to continue iteration; if it has, stop iteration and output the optimal parameter combination corresponding to the final global optimum and local optimum. .

4. The method for denoising and reconstructing ultra-wideband radar vital sign signals according to claim 3, characterized in that, The fitness function is expressed as follows: ； in, Let k be the fitness value, and k be the total number of IMF components obtained from the decomposition of the mixed vital sign signal x(t). Let be the envelope entropy value of the i-th IMF component; The envelope weighting coefficient is... , The values ​​of are linearly distributed from low to high frequencies of the k IMF components; This is a parameter boundary penalty term, applied when the parameter combination... Triggered when parameters exceed the optimization range. When parameter combinations When there are no parameters outside the optimization range, The value of is 0; The calculation formula is: ； Where L is the number of sampling points of the mixed signal of vital signs x(t) to be decomposed. To preset the minimum value, Hilbert envelope of the i-th IMF component The normalized probability distribution.

5. The method for denoising and reconstructing ultra-wideband radar vital sign signals according to claim 3, characterized in that, Dung beetle behavior includes rolling behavior, reproductive behavior, foraging behavior, and predator avoidance behavior; The expression for population iterative updates based on the rolling ball behavior is: ； in, For the first The position of an individual dung beetle in the (v+1)th iteration. For the first The individual position of a dung beetle population in the current iteration v. This is the worst position for the current population. This represents the globally optimal position for the current population. For direction coefficients, ; The expression for population iteration and updating based on reproductive behavior is: ； in, This represents the local optimum position for the current population. A random vector with 1 row and 1 column. Each element in the set has a value range of (0, 1). This represents the maximum number of iterations.

6. The method for denoising and reconstructing ultra-wideband radar vital sign signals according to claim 3, characterized in that, S3, substituting the optimal parameter combination into the ICEEMDAN algorithm, adaptively decomposes the mixed signal of vital signs to be decomposed, obtaining multiple IMF components arranged from high to low frequency, including: Combine the optimal parameters Substituting into the ICEEMDAN algorithm, we perform adaptive decomposition on the mixed signal x(t) of vital signs to be decomposed. The expression for adaptive decomposition is: ； in, Let x(t) be the d-th IMF component obtained from the decomposition of the mixed signal of vital signs to be decomposed. K represents the final residual components, and K is the total number of IMF components obtained from the decomposition of the mixed vital signs signal x(t). Arrange the K IMF components in descending order of frequency.

7. The method for denoising and reconstructing ultra-wideband radar vital sign signals according to claim 1, characterized in that, S4, based on the frequency band characteristics of human physiological signals and Pearson correlation analysis, effective IMF components are screened to obtain the effective IMF components of the corresponding respiratory signals and heartbeat signals, including: S4.1, Based on the frequency band characteristics of human physiological signals, perform spectral analysis on each IMF component, extract the peak frequency, and screen out the respiratory candidate IMF components with peak frequencies located in the respiratory physiological frequency band of 0.1Hz~0.6Hz and the heartbeat candidate IMF components located in the heartbeat physiological frequency band of 0.8Hz~2.5Hz. S4.2, calculate the Pearson correlation coefficients of the respiratory candidate IMF component, the heartbeat candidate IMF component, and the mixed signal x(t) of the vital signs to be decomposed, respectively. The formula for calculating the Pearson correlation coefficient is: ； in, For covariance, Let h be the standard deviation of the h-th candidate IMF component for respiration or heartbeat. Let x(t) be the standard deviation. S4.3, Pearson correlation coefficients between respiratory candidate IMF components and heartbeat candidate IMF components. In the middle, Pearson correlation coefficients were selected respectively. The highest C1 respiratory candidate IMF components and C2 heartbeat candidate IMF components are used as the effective IMF components of the respiratory signal and the effective IMF components of the heartbeat signal, respectively. C1 is the initial respiratory selection value and C2 is the initial heartbeat selection value.

8. The method for denoising and reconstructing ultra-wideband radar vital sign signals according to claim 7, characterized in that, The calculation formula for adaptive weighted fusion reconstruction is as follows: ； Where C is either C1 or C2, The c-th effective IMF component of the respiratory or heartbeat signal. This refers to the adaptively weighted fusion and reconstruction of respiratory vital signs or cardiac vital signs signals. For adaptive weighting coefficients; adaptive weighting coefficients The calculation formula is: ； in, This represents the energy percentage of the c-th effective IMF component of the respiratory or heartbeat signal within the target physiological frequency band. is the Pearson correlation coefficient between the c-th effective IMF component and the mixed signal x(t) of the vital signs to be decomposed.

9. The method for denoising and reconstructing ultra-wideband radar vital sign signals according to claim 8, characterized in that, The method also includes: Calculate the reconstructed respiratory rate from respiratory vital signs signals and obtain the measured respiratory rate using a human contact device; Determine whether the difference between the reconstructed respiratory rate and the measured respiratory rate exceeds a preset difference threshold; If the difference does not exceed the preset threshold, the respiratory vital signs signal is determined to meet the accuracy requirements. If the difference exceeds the preset threshold, the first respiratory selection value is obtained by adjusting the initial respiratory selection value C1 by C1±1. ,according to Return to step S4.3 to obtain... The first respiratory candidate IMF component is then processed, and step S5 is executed to obtain the first adjusted first respiratory signal. When the difference between the reconstructed respiratory rate of the first respiratory signal and the measured respiratory rate does not exceed a preset difference threshold, the first respiratory signal is determined to meet the accuracy requirements. When the difference between the reconstructed respiratory rate and the measured respiratory rate of the first respiratory signal exceeds a preset difference threshold, the value selected in the first respiratory signal is... Based on Adjust to obtain the second breath selection value ,according to Return to step S4.3 to obtain... The second respiratory signal is obtained by performing step S5 after identifying the candidate IMF components for each breath. Until the The adjusted number The respiratory signal meets the accuracy requirements. Less than or equal to the preset number of adjustments.

10. A denoising and reconstruction system for ultra-wideband radar vital sign signals, characterized in that, include: The mixed signal extraction module is used to acquire the baseband signal of vital signs echo collected by ultra-wideband radar, preprocess the vital signs echo baseband signal, and extract the one-dimensional slow time domain mixed signal of vital signs to be decomposed. The parameter optimization module is used to perform global adaptive optimization of the core parameters of the ICEEMDAN algorithm based on the Dung Beetle Optimization Algorithm (DBO) to obtain the optimal parameter combination. The adaptive signal decomposition module is used to substitute the optimal parameter combination into the ICEEMDAN algorithm to adaptively decompose the mixed vital signs signal to be decomposed, and obtain multiple IMF components arranged from high to low frequency. The effective component screening module is used to screen the effective components of IMF based on the frequency band characteristics of human physiological signals and Pearson correlation analysis, and obtain the effective IMF components of the corresponding respiratory signals and heartbeat signals respectively. The adaptive weighted fusion reconstruction module is used to adaptively weightedly fused and reconstruct the effective IMF components of the respiratory signal and the effective IMF components of the heartbeat signal, and output respiratory vital signs signals and heartbeat vital signs signals.

Images