Characteristic extraction method for abnormal vital signs based on sign signals

Abstract

The invention relates to the technical field of vital sign anomaly detection, in particular to a vital sign anomaly feature extraction method based on a sign signal, which comprises the following steps: firstly, dynamically adjusting the size of a time cabin through variational optimization, taking both transient and steady feature capture into account, and avoiding feature loss caused by a fixed window length; in combination with principal component analysis, Lyapunov exponential chaos quantization and a high-order differential operator, nonlinear features are decoupled from multiple dimensions of a time domain, a frequency domain and a phase domain, the feature interpretability is improved, further, through phase-speed mapping and attention weighted fusion, self-adaptive enhancement and decoupling of spatial-temporal features are achieved, noise interference is suppressed, and the spatial-temporal features are enhanced. The finally generated multi-dimensional feature cube integrates time resolution, frequency precision and phase differential information, the anomaly detection robustness of complex vital signs is remarkably improved, and the defects that a traditional method is single in dimension and insufficient in adaptability in dynamic signal processing are overcome.

Description

Technical Field

[0001] The present invention relates to the technical field of vital sign abnormality detection, and in particular to a feature extraction method for vital sign abnormalities based on vital sign signals. Background Art

[0002] With the widespread application of FMCW (frequency modulated continuous wave) millimeter-wave radar technology in medical monitoring, non-contact detection of vital signs (such as breathing, heartbeat, and body movement) has become a research hotspot. Traditional vital sign monitoring relies primarily on contact sensors (such as ECG electrodes and breathing belts), which suffer from poor comfort and susceptibility to interference. While radar-based non-contact monitoring can overcome these limitations, its signal processing faces the following technical challenges: Transient feature loss: Fixed window length framing makes it difficult to take into account both the transient and steady-state features of the signal, resulting in the smoothing or omission of key phase change information; Nonlinear characteristics are difficult to explain: Existing methods focus on linear characteristics such as amplitude and energy, but lack the ability to effectively model and explain the phase behavior of chaos and nonlinear perturbations; Limited feature fusion dimensions: Multi-scale feature fusion methods mainly target modalities such as amplitude and energy, and lack structured encoding methods for high-order phase differentials and dynamic evolution processes; Insufficient spatiotemporal decoupling capabilities: Existing technologies make it difficult to achieve spatiotemporal separation and independent modeling of phase information, affecting the accuracy of subsequent identification and tracking.

[0003] In summary, in vital sign monitoring technology, traditional time-frequency analysis methods (such as short-time Fourier transform and wavelet transform) have problems such as loss of transient features, difficulty in interpreting nonlinear features, limited feature fusion dimensions, and insufficient time-space decoupling capabilities when processing dynamic and nonlinear signals. Summary of the Invention

[0004] The purpose of the present invention is to provide a feature extraction method for vital sign abnormalities based on vital sign signals, so as to solve the problems in vital sign monitoring technology that traditional time-frequency analysis methods have transient feature loss, difficulty in interpreting nonlinear features, limited feature fusion dimension and insufficient time-space decoupling capability when processing dynamic and nonlinear signals.

[0005] To achieve the above object, the present invention provides a method for extracting features of abnormal vital signs based on vital sign signals, the method comprising the following steps: The FMCW millimeter wave radar collects the radar echo signal reflected by the human body and converts it into a continuous phase sequence ; Continuous radar echo phase sequence Divided into time capsule units , each cabin carries The phase information of the duration forms a space-time matrix; Adaptive adjustment of time capsule size through variational optimization principle ; The optimal window length is solved by the Lagrange multiplier method to achieve complete encapsulation of transient characteristics, and then the data is multi-dimensionally deconstructed and the phase characteristics are decoupled; Time capsule unit Perform principal component analysis to extract the dominant phase mode , retain the principal components with cumulative contribution rate ≥ 95%; Analyze the integer frequency phase structure by fast Fourier transform; Based on the Lyapunov exponent, calculate the maximum Lyapunov exponent of nonlinear phase perturbations , quantify the chaotic nature of phase evolution; The time-varying features are extracted by high-order differential operators to construct the phase gradient vector , characterizes the time-frequency domain coupling characteristics; Establish phase-velocity mapping relationship; The features of time, frequency and phase differential orders are integrated to generate a multidimensional feature cube, which serves as the feature basis for vital sign abnormality detection.

[0006] Specifically, in step "converting the continuous radar echo phase sequence Divided into time capsule units , each cabin carries The phase information of the duration is used to form a space-time matrix. The space-time matrix formed is specifically as follows: ; in, Represents the nth-order time derivative of the phase, constructing a multidimensional phase matrix containing time domain differential features, is the sampling interval in the time capsule, m is the number of sampling points and satisfies m≥n+1.

[0007] Specifically, in step "Adaptively adjust the time capsule size through variational optimization principle ", the objective function is specifically: ; in, is the phase characteristic variance in the cabin, used to ensure feature richness; for Norm, used to constrain the smoothness of adjacent cabins; is the information entropy, which is used to suppress redundant features.

[0008] Specifically, the step "time capsule unit Perform principal component analysis to extract the dominant phase mode The core algorithm of "retaining principal components with cumulative contribution rate ≥ 95%" is: ; is the covariance matrix, is the characteristic value.

[0009] Specifically, the core algorithm of the step "analyzing the integer frequency multiplication phase structure by fast Fourier transform" is: Harmonic Chamber (FFT): ; Extracting harmonic components , As the fundamental frequency, construct the harmonic phase spectrum .

[0010] Specifically, the step "calculate the maximum Lyapunov exponent of the nonlinear phase perturbation based on the Lyapunov exponent" , the specific content of "quantifying the chaotic characteristics of phase evolution" is: The Wolf algorithm is used to calculate the maximum Lyapunov exponent of nonlinear phase perturbations. : ; in, is the Jacobian matrix of the phase flow, is the tangent space vector, which is used to quantify the chaotic characteristics of the phase evolution.

[0011] Specifically, the steps are "extracting time-varying features through high-order differential operators and constructing phase gradient vectors , the specific content of "characterizing the time-frequency domain coupling characteristics" is: The time-varying features are extracted by high-order differential operators, and the n-order phase gradient vector is defined: ; Among them, the mixed partial derivatives Characterize the time-frequency domain coupling characteristics.

[0012] Specifically, in the step “establishing phase-velocity mapping relationship”, the target radial velocity needs to be considered Time-varying characteristics of: ; in, is the nonlinear phase perturbation term, which can be expressed by Taylor expansion as follows: .

[0013] Specifically, in the step “establishing phase-velocity mapping relationship”, the nonlinear phase perturbation term When performing calculations, it is necessary to obtain the time attention and frequency attention weights; Calculate the temporal attention weight and use an exponential decay model to enhance recent phase features: ; in, is the characteristic peak moment, is the time window standard deviation; Calculate the frequency attention weight and highlight the feature frequency band based on the Gaussian kernel function: ; in, is the target frequency, is the standard deviation of the frequency window.

[0014] Specifically, in the step of "fusing the features of time, frequency, and phase differential orders to generate a multidimensional feature cube as the feature basis for vital sign abnormality detection", the feature cube that fuses the time axis, frequency axis, and phase differential order is expressed as: ; in, T is the time point, F is the frequency point number, N is the upper limit of the differential order.

[0015] The present invention provides a feature extraction method for abnormal vital signs based on vital sign signals. First, the time capsule size is dynamically adjusted through variational optimization, taking into account both transient and steady-state feature capture and avoiding feature loss caused by fixed window length. Secondly, principal component analysis, Lyapunov exponent chaos quantization and high-order differential operators are combined to decouple nonlinear features from multiple dimensions in the time domain, frequency domain and phase domain to improve feature interpretability. Furthermore, through phase-velocity mapping and attention weighted fusion, adaptive enhancement and decoupling of spatiotemporal features are achieved, and noise interference is suppressed. The resulting multidimensional feature cube integrates time resolution, frequency accuracy and phase differential information, significantly improving the robustness of abnormality detection for complex vital signs and overcoming the defects of traditional methods in dynamic signal processing, such as single dimension and insufficient adaptability. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0017] Figure 1 This is a flowchart of the steps of the feature extraction method of abnormal vital signs based on vital sign signals provided by the present invention. DETAILED DESCRIPTION

[0018] The following describes embodiments of the present invention in detail, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and are not to be construed as limiting the present invention.

[0019] See also Figure 1 The present invention provides a method for extracting features of abnormal vital signs based on vital sign signals, the method comprising the following steps: S1: Based on FMCW millimeter wave radar, the radar echo signal reflected by the human body is collected and converted into a continuous phase sequence ; S2: Convert continuous radar echo phase sequence Divided into time capsule units , each cabin carries The phase information of the duration forms a space-time matrix; S3: Adaptive adjustment of time capsule size through variational optimization principle ; S4: After solving the optimal window length by the Lagrange multiplier method and achieving complete encapsulation of transient features, the data is multi-dimensionally deconstructed and the phase features are decoupled; S5: Time capsule unit Perform principal component analysis to extract the dominant phase mode , retain the principal components with cumulative contribution rate ≥ 95%; S6: Resolving integer frequency phase structure by fast Fourier transform; S7: Based on the Lyapunov exponent, calculate the maximum Lyapunov exponent of the nonlinear phase perturbation , quantify the chaotic nature of phase evolution; S8: Extract time-varying features through high-order differential operators and construct phase gradient vectors , characterizes the time-frequency domain coupling characteristics; S9: establishing phase-velocity mapping relationship; S10: Fuse the features of time, frequency and phase differential orders to generate a multidimensional feature cube as the feature basis for vital sign abnormality detection.

[0020] In this embodiment, first, the time capsule size is dynamically adjusted through variational optimization, taking into account both transient and steady-state feature capture and avoiding feature loss caused by fixed window length. Secondly, principal component analysis, Lyapunov exponent chaos quantization and high-order differential operators are combined to decouple nonlinear features from multiple dimensions in the time, frequency and phase domains to improve feature interpretability. Furthermore, through phase-velocity mapping and attention weighted fusion, adaptive enhancement and decoupling of spatiotemporal features are achieved, and noise interference is suppressed. The multidimensional feature cube finally generated integrates time resolution, frequency accuracy and phase differential information, significantly improving the robustness of abnormality detection of complex vital signs and overcoming the defects of traditional methods in dynamic signal processing, such as single dimension and insufficient adaptability.

[0021] Further, specifically, in the step of "converting the continuous radar echo phase sequence Divided into time capsule units , each cabin carries The phase information of the duration is used to form a space-time matrix. The space-time matrix formed is specifically as follows: ; in, Represents the nth-order time derivative of the phase, constructing a multidimensional phase matrix containing time domain differential features, is the sampling interval in the time capsule, m is the number of sampling points and satisfies m≥n+1.

[0022] Furthermore, in step “Adaptively adjust the time capsule size by variational optimization principle ", the objective function is specifically: ; in, is the phase characteristic variance in the cabin, used to ensure feature richness; for Norm, used to constrain the smoothness of adjacent cabins; is the information entropy, which is used to suppress redundant features.

[0023] Further, the step "time capsule unit Perform principal component analysis to extract the dominant phase mode The core algorithm of "retaining principal components with cumulative contribution rate ≥ 95%" is: ; is the covariance matrix, is the characteristic value.

[0024] Furthermore, the core algorithm of the step “analyzing the integer frequency multiplication phase structure by fast Fourier transform” is: Harmonic Chamber (FFT): ; Extracting harmonic components , As the fundamental frequency, construct the harmonic phase spectrum .

[0025] Furthermore, the step "calculate the maximum Lyapunov exponent of the nonlinear phase perturbation based on the Lyapunov exponent" , the specific content of "quantifying the chaotic characteristics of phase evolution" is: The Wolf algorithm is used to calculate the maximum Lyapunov exponent of nonlinear phase perturbations. : ; in, is the Jacobian matrix of the phase flow, is the tangent space vector, which is used to quantify the chaotic characteristics of the phase evolution.

[0026] Further, the step "extracts time-varying features through high-order differential operators and constructs phase gradient vectors , the specific content of "characterizing the time-frequency domain coupling characteristics" is: The time-varying features are extracted by high-order differential operators, and the n-order phase gradient vector is defined: ; Among them, the mixed partial derivatives Characterize the time-frequency domain coupling characteristics.

[0027] Furthermore, in the step “Establishing Phase-Velocity Mapping Relationship”, the target radial velocity needs to be considered. Time-varying characteristics of: ; in, is the nonlinear phase perturbation term, which can be expressed by Taylor expansion as follows: .

[0028] Furthermore, in the step “establishing phase-velocity mapping relationship”, the nonlinear phase perturbation term When performing calculations, it is necessary to obtain the time attention and frequency attention weights; Calculate the temporal attention weight and use an exponential decay model to enhance recent phase features: ; in, is the characteristic peak moment, is the time window standard deviation; Calculate the frequency attention weight and highlight the feature frequency band based on the Gaussian kernel function: ; in, is the target frequency, is the standard deviation of the frequency window.

[0029] Furthermore, in the step of "fusing the features of time, frequency, and phase differential orders to generate a multidimensional feature cube as the feature basis for vital sign abnormality detection", the feature cube that fuses the time axis, frequency axis, and phase differential order is expressed as: ; in, T is the time point, F is the frequency point number, N is the upper limit of the differential order.

[0030] In summary, this technical solution finally obtains structured and interpretable phase feature data through the above-mentioned multi-level "time capsule" framing and multi-dimensional phase feature decoupling and fusion method. This feature data can comprehensively characterize the time-varying, frequency-varying and high-order differential characteristics in the radar echo signal, providing a highly robust and high-resolution feature foundation for the subsequent diagnosis of respiratory disorders, chest motion analysis and vital sign abnormality detection.

[0031] Specifically, this technical solution has the following beneficial effects: Transient features are fully preserved: The adaptive "time capsule" framing method can dynamically adjust the analysis window length to accurately capture transient changes in the signal, avoiding the feature loss caused by traditional fixed window length methods; Nonlinear and chaotic characteristic modeling: Using chaos analysis methods such as the Lyapunov exponent, we can effectively extract and quantify nonlinear perturbations in phase sequences, providing theoretical support for complex motion recognition. Multi-dimensional feature fusion and decoupling: Combining principal component analysis, harmonic analysis, and high-order differential operators, it achieves multimodal feature fusion in the time domain, frequency domain, and differential domain, improving feature expression capabilities; Spatiotemporal attention mechanism: Introducing time and frequency attention weights to highlight key moments and characteristic frequency bands, enhancing the discriminability and robustness of features.

[0032] The above disclosure is only a preferred embodiment of the present invention, and certainly cannot be used to limit the scope of the rights of the present invention. Ordinary technicians in this field can understand that all or part of the processes of the above embodiment and equivalent changes made in accordance with the claims of the present invention are still within the scope of the invention.

Claims

1. A feature extraction method for abnormal vital signs based on vital sign signals, characterized in that: The steps include: The FMCW millimeter wave radar collects the radar echo signal reflected by the human body and converts it into a continuous phase sequence ; Continuous radar echo phase sequence Divided into time capsule units , each cabin carries The phase information of the duration forms a space-time matrix; Adaptive adjustment of time capsule size through variational optimization principle ; The optimal window length is solved by the Lagrange multiplier method to achieve complete encapsulation of transient characteristics, and then the data is multi-dimensionally deconstructed and the phase characteristics are decoupled; Time capsule unit Perform principal component analysis to extract the dominant phase mode , retain the principal components with cumulative contribution rate ≥ 95%; Analyze the integer frequency phase structure by fast Fourier transform; Based on the Lyapunov exponent, calculate the maximum Lyapunov exponent of nonlinear phase perturbations , quantify the chaotic nature of phase evolution; The time-varying features are extracted by high-order differential operators to construct the phase gradient vector , characterizes the time-frequency domain coupling characteristics; Establish phase-velocity mapping relationship; The features of time, frequency and phase differential orders are integrated to generate a multidimensional feature cube, which serves as the feature basis for vital sign abnormality detection.

2. The feature extraction method for abnormal vital signs based on vital sign signals according to claim 1, characterized in that: In step 1, the continuous radar echo phase sequence Divided into time capsule units , each cabin carries The phase information of the duration forms a space-time matrix, and the space-time matrix formed is specifically: ； in, Represents the nth-order time derivative of the phase, constructing a multidimensional phase matrix containing time domain differential features, is the sampling interval in the time capsule, m is the number of sampling points and satisfies m≥n+1.

3. The feature extraction method for abnormal vital signs based on vital sign signals according to claim 2, characterized in that: In the step, the time capsule size is adaptively adjusted by the variational optimization principle. In the above example, the objective function is: ； in, is the phase characteristic variance in the cabin, used to ensure feature richness; for Norm, used to constrain the smoothness of adjacent cabins; is the information entropy, which is used to suppress redundant features.

4. The feature extraction method for abnormal vital signs based on vital sign signals according to claim 3, characterized in that: Steps to the Time Capsule Unit Perform principal component analysis to extract the dominant phase mode The core algorithm for retaining principal components with cumulative contribution rates ≥ 95% is: ； is the covariance matrix, is the characteristic value.

5. The feature extraction method for abnormal vital signs based on vital sign signals according to claim 4, characterized in that: The core algorithm for analyzing the integer frequency multiplication phase structure through fast Fourier transform is: Harmonic Chamber: ； Extracting harmonic components , As the fundamental frequency, construct the harmonic phase spectrum .

6. The feature extraction method for abnormal vital signs based on vital sign signals according to claim 5, characterized in that: Step 1: Calculate the maximum Lyapunov exponent of the nonlinear phase perturbation based on the Lyapunov exponent. , the specific content of the chaotic characteristics of quantified phase evolution is: The Wolf algorithm is used to calculate the maximum Lyapunov exponent of nonlinear phase perturbations. : ； in, is the Jacobian matrix of the phase flow, is the tangent space vector, which is used to quantify the chaotic characteristics of the phase evolution.

7. The feature extraction method for abnormal vital signs based on vital sign signals according to claim 6, characterized in that: Step 1: Extract time-varying features by high-order differential operators and construct phase gradient vectors , the specific content of characterizing the time-frequency domain coupling characteristics is: The time-varying features are extracted by high-order differential operators, and the n-order phase gradient vector is defined: ； Among them, the mixed partial derivatives Characterize the time-frequency domain coupling characteristics.

8. The feature extraction method for abnormal vital signs based on vital sign signals according to claim 7, characterized in that: In the step of establishing the phase-velocity mapping relationship, the target radial velocity needs to be considered Time-varying characteristics of: ； Among them, is the nonlinear phase perturbation term, which can be expressed by Taylor expansion as follows: 。 9. The feature extraction method for abnormal vital signs based on vital sign signals according to claim 8, characterized in that: In the step of establishing the phase-velocity mapping relationship, the nonlinear phase disturbance term When performing calculations, it is necessary to obtain the time attention and frequency attention weights; Calculate the temporal attention weight and use an exponential decay model to enhance recent phase features: ； in, is the characteristic peak moment, is the time window standard deviation; Calculate the frequency attention weight and highlight the feature frequency band based on the Gaussian kernel function: ； in, is the target frequency, is the standard deviation of the frequency window.

10. The feature extraction method for abnormal vital signs based on vital sign signals according to claim 9, characterized in that: In the step, the features of time, frequency and phase differential order are fused to generate a multidimensional feature cube, which is used as the feature basis for vital sign abnormality detection. The feature cube that integrates the time axis, frequency axis and phase differential order is expressed as: ； in, T is the time point, F is the frequency point number, N is the upper limit of the differential order.

Images