VMD and SSR combined FMCW radar vital sign detection method

Abstract

The invention discloses an FMCW radar vital sign detection method combining VMD and SSR. The method comprises the following steps: firstly, transmitting a chirp signal through an FMCW radar, capturing a radar echo signal, preprocessing the radar echo signal, and extracting a human chest wall displacement signal; secondly, decomposing the human chest wall displacement signal by adopting a VMD algorithm, determining an optimal decomposition parameter by utilizing variable step size double-stage search through a vital sign judgment mode, and decomposing the human chest wall displacement signal by utilizing the optimal decomposition parameter to obtain a heartbeat signal and breathing signal modal component set; and finally, judging a breathing harmonic modal component in a heartbeat signal frequency range, and reconstructing a heartbeat signal. And constructing an SSR model, performing sparse spectrum reconstruction on the heartbeat signal based on a ZA-EFRLS algorithm, solving the heartbeat signal, and extracting the heartbeat frequency. The method solves the problems of difficulty in heartbeat signal extraction, clutter interference, insufficient frequency resolution under a short-time observation window, long heartbeat frequency extraction time and the like.

Description

Technical Field

[0001] This invention belongs to the field of non-contact vital sign detection technology, specifically relating to an FMCW radar vital sign detection method that combines VMD and SSR. Background Technology

[0002] Cardiovascular disease has become a major threat to human health, and electrocardiogram (ECG) monitoring is considered one of the standards for clinical diagnosis of cardiovascular disease, possessing extremely high clinical value in early disease diagnosis and subsequent treatment. However, due to the inconvenience of traditional ECG monitoring, long-term continuous monitoring is often difficult to implement in daily life, which may lead to the loss of abnormal ECG records and delay in disease diagnosis. Wearable sensors or adhesive electrodes, such as breathing straps, photoelectric volume pulse wave sensors, and ECG sensors, may bring uncomfortable experiences and additional burdens to subjects, especially for some patients with skin allergies or burns.

[0003] In non-contact vital sign detection methods, although Wi-Fi-based non-contact sensing has gained increasing attention in numerous applications such as respiration detection, location estimation, and gesture recognition, its performance is limited by narrow bandwidth, few antennas, and large wavelengths. This is particularly true in heart rate detection, where the 2.4- / 5- GHz Wi-Fi wavelength (60-120 mm) is much larger than the chest wall displacement caused by heartbeat (0.2-0.5 mm), making it difficult to capture the minute phase changes caused by heartbeats. Acoustic signals are another potential solution for vital sign detection, but their sensing range is limited. Furthermore, computer vision-based vital sign detection is sensitive to lighting and line-of-sight conditions, performing poorly in smoke, low light, or obstructed conditions, and may also lead to privacy breaches. Currently, continuous wave (CW) Doppler radar, ultra-wideband (UWB) pulse radar, and frequency modulated continuous wave (FMCW) radar are three commonly used radars for vital sign detection. Continuous wave Doppler radar has a simple radio structure and low power consumption, but it cannot provide target distance information. Therefore, its performance is easily affected by clutter interference, resulting in poor accuracy in vital sign detection. In contrast, UWB pulse radar and FMCW radar can measure the distance between the target and the device and occupy a wider bandwidth. However, for UWB pulse radar, the wide bandwidth depends on precise control of the pulse width and the peak signal strength of the radar, which leads to higher hardware costs and system complexity. Millimeter-wave radar is a radar technology with an electromagnetic spectrum corresponding to the 30-300 GHz frequency band, featuring wide bandwidth, narrow beamwidth, and small size. FMCW millimeter-wave radar combines the advantages of FMCW and millimeter-wave technologies. Due to its wider bandwidth, FMCW millimeter-wave radar has significantly improved range resolution and helps isolate reflections from different objects. At the same time, FMCW millimeter-wave radar has the potential for miniaturization due to its higher carrier frequency. In addition, compared with cameras, WiFi routers, and acoustic sensors, the propagation of radar signals is less affected by changes in light / temperature / sound and does not infringe on privacy. Therefore, FMCW millimeter-wave radar can be used for target HR detection.

[0004] Radar senses the surrounding environment through reflected signals from chest wall displacement caused by respiration and cardiac activity, as well as various environmental noises. The heartbeat cycle cannot be directly observed in radar signals, requiring appropriate algorithms to further extract potential cardiac features. However, accurate heart rate (HR) detection is difficult due to interference from respiratory harmonics, noise, and clutter. Specifically, the environment may contain a variety of objects (e.g., walls, doors, tables, and furniture), resulting in reflected signals with much clutter that can mask the heartbeat signal. Furthermore, chest wall displacement caused by respiration can be an order of magnitude greater than that caused by heartbeat, and it is not a pure sine wave, containing several significant harmonic components. Respiratory harmonics can approach or even mask the heartbeat signal, leading to incorrect peak selection. Therefore, achieving accurate and reliable HR detection remains a challenge. Summary of the Invention

[0005] To address the shortcomings of existing technologies, the technical problem this invention aims to solve is to provide a combined VMD and SSR FMCW radar vital sign detection method. This method aims to solve problems such as difficulty in extracting heartbeat signals, clutter interference, insufficient frequency resolution under short observation windows, and long heartbeat frequency extraction time in existing technologies, thereby improving the accuracy and robustness of non-contact vital sign detection.

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: A method for detecting vital signs using FMCW radar combining VMD and SSR, characterized by the following steps: Step 1: Transmit chirped signals and capture radar echo signals using FMCW radar, preprocess the radar echo signals, and extract human chest wall displacement signals. Step 2: The VMD algorithm is used to decompose the human chest wall displacement signal. The optimal decomposition parameters are determined by using a variable step size two-stage search through the vital signs discrimination mode. The optimal decomposition parameters are then used to decompose the human chest wall displacement signal to obtain the modal component sets of heartbeat signal and respiratory signal. Step 3: Identify the respiratory harmonic mode components within the frequency range of the heartbeat signal and reconstruct the heartbeat signal; The respiratory harmonic mode component set was obtained through harmonic detection. : (4) in, , These are the modal component sets of heartbeat and respiratory signals, respectively. For integers ranging from 0 to 2, , The first , The center frequency of each modal component; The intersection of the respiratory signal modal component set and the respiratory harmonic modal component set is used to obtain the pure respiratory signal modal component set. The difference between the heartbeat signal modal component set and the respiratory harmonic modal component set is used to obtain the pure heartbeat signal modal component set. The modal components in the pure respiratory signal modal component set and the heartbeat signal modal component set are superimposed in the time domain to obtain the reconstructed respiratory signal and heartbeat signal. Step 4: Construct an SSR model, perform sparse spectrum reconstruction of the heartbeat signal based on the ZA-EFRLS algorithm, solve the heartbeat signal, and extract the heartbeat frequency.

[0007] Furthermore, in the second step, the energy loss rate is calculated according to equation (6): (6) In the formula, The total energy of all modal components. For the first Modal components Total energy, The number of modal components; Based on the energy proportion of respiratory and heartbeat signals in each modal component, the modal components belonging to respiratory and heartbeat signals are selected to obtain the respiratory signal modal component set and the heartbeat signal modal component set; (7) (8) In the formula, For the first The energy of the respiratory signal in each modal component For the first The energy of the center jump signal of each modal component and The preset threshold; The optimal decomposition parameters are selected based on the minimum energy loss rate and the fact that neither the respiratory signal modal component set nor the heartbeat signal modal component set is empty.

[0008] Compared with the prior art, the beneficial effects of the present invention are: This invention addresses the problem of the VMD algorithm's over-reliance on decomposition parameters by introducing the energy loss rate. Furthermore, it utilizes a variable-step-size two-stage search to determine the optimal decomposition parameters, thus reducing the algorithm's complexity. Relying solely on the energy loss rate as the sole criterion can lead to problems within the potential range. and The minimum value judgment is a misjudgment of the optimal decomposition parameters. Therefore, the introduced vital sign discrimination mode can ensure that the decomposition contains respiratory signal set and heartbeat signal set, thus solving the mode mixing problem. Harmonic detection and signal reconstruction solve the interference of respiratory harmonic signals and obtain pure heartbeat signals. Sparse spectrum reconstruction solves the picket fence effect of traditional FFT algorithm when sampling short sequences, thus achieving high resolution and high robustness of heartbeat frequency extraction under short observation window. Attached Figure Description

[0009] Figure 1 This is an overall flowchart of the present invention; Figure 2 This is a flowchart of the radar signal preprocessing and human chest wall displacement signal extraction of the present invention; Figure 3 The flowchart shows the combined improved VMD algorithm and SSR of this invention; Figure 4 This is a flowchart of the sparse spectrum reconstruction and heart rate extraction based on the ZA-EFRLS algorithm of the present invention. Detailed Implementation

[0010] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0011] like Figure 1 As shown, this invention provides a method for detecting vital signs using FMCW radar that combines VMD and SSR, comprising the following steps: Step 1: Transmit FMCW radar chirp signal and receive radar echo signal, preprocess the radar echo signal to obtain human chest wall displacement signal; like Figure 2 As shown, the starting frequency of the FMCW radar is bandwidth is , indicating the frequency variation range of the signal; The amplitude of the transmitted chirped signal is such that the frequency modulation slope of the radar wave increases linearly with time as... ;exist At any given time, the FMCW radar's transmitted signal is represented in complex signal form as follows: (1) in, for The transmission signal of the FMCW radar at all times. For the chirping period; A stationary human target is located at the range radar. The instantaneous distance between the human target and the FMCW radar is expressed as: (2) in, for The instantaneous distance between the human target and the FMCW radar at any given moment. and They are respectively Chest displacement caused by constant breathing and heartbeat; When the transmitted signal hits a human target, the signal is reflected back to the radar receiver. The received signal of the FMCW radar at any time Represented as: (3) in, The amplitude of the received signal. The round-trip time delay of signals reflected from the human body. The speed of light; The received signal is mixed with a copy of the transmitted signal, and then subjected to quadrature demodulation and low-pass filtering to obtain the intermediate frequency signal. , can be represented as: (4) in The third term of the amplitude and phase of the intermediate frequency signal. It is relatively small and can be ignored in close-up scenarios; therefore... It can be represented as (5) in, The frequency of the intermediate frequency signal. This represents the time-varying phase of the intermediate frequency signal. Because the chest wall displacement caused by respiration and heartbeat is very small, The chest wall displacement, which is approximately constant in each chirp cycle, can be inferred from the phase signal when multiple chirps are emitted in succession. Next, the intermediate frequency signal is converted from analog to digital, and a fast Fourier transform is performed on the fast time dimension of the sampled data matrix to obtain the range-slow time matrix. Generally, the amplitude of the reflected signal from a stationary target within the radar illumination range does not change in the slow time dimension. However, due to small physiological movements caused by breathing and heartbeat, the amplitude of the reflected signal from a stationary human target varies significantly. Therefore, the range bin with the most prominent amplitude variation is determined using maximum variance, and DC bias calibration is performed to remove the DC bias of the slow time signal. Finally, the slow-time signal after DC bias calibration was subjected to arctangent demodulation to extract the phase, thus obtaining the human chest wall displacement signal. .

[0012] Step 2: Use an improved VMD algorithm to analyze the human chest wall displacement signal. The decomposition process is performed, and the optimal decomposition parameters are determined through the vital sign discrimination mode. Based on the optimal decomposition parameters, the human chest wall displacement signal is decomposed to obtain the respiratory signal modal component set and the heartbeat signal modal component set. like Figure 3 As shown, the VMD algorithm uses ADMM to iteratively update each modal component and its associated center frequency. This process gradually demodulates each modal component to its respective baseband, ultimately extracting each modal component. and its center frequency ; through adaptive parameter energy loss rate Select decomposition parameters (including the number of modal components) and penalty coefficient ): (6) Based on the energy proportions of respiratory and heartbeat signals in each modal component, modal components belonging to respiratory and heartbeat signals are selected to obtain the respiratory signal modal component set. and heartbeat signal modal component set ; (7) (8) in, The total energy of all modal components. For the first Modal components Total energy, For the first The energy of the respiratory signal in each modal component For the first The energy of the center jump signal of each modal component and The preset threshold; The optimal decomposition parameters are selected based on the minimum energy loss rate and the fact that both the respiratory signal modal component set and the heartbeat signal modal component set are non-empty. These optimal parameters are then used to decompose the human chest wall displacement signal to obtain the respiratory signal modal component set. and heartbeat signal modal component set .

[0013] Step 3: Identify the harmonic signal set and reconstruct the heartbeat signal; Since respiratory harmonics and heartbeat signals may have similar amplitudes and frequencies, they can interfere with heartbeat frequency extraction. Therefore, it is necessary to separate the respiratory harmonic signals within the heartbeat signal modal component set as much as possible. Based on the respiratory and heartbeat frequency bands of healthy adults, it can be concluded that the frequencies of the respiratory harmonics that cause interference are mostly 2 to 4 times the respiratory signal frequency. Therefore, the respiratory harmonic mode component set within the heartbeat signal frequency range can be represented as: (9) After obtaining the respiratory harmonic mode component set, the respiratory signal mode component set... With respiratory harmonic mode component set Taking the intersection yields the pure respiratory signal modal component set, and the heartbeat signal modal component set... With respiratory harmonic mode component set The pure heartbeat signal modal component set can be obtained by taking the difference between the two sets; the modal components in the pure respiratory signal modal component set and the heartbeat signal modal component set are then superimposed in the time domain to obtain the reconstructed respiratory signal and heartbeat signal, represented as follows: (10) (11) in, It is a reconstructed respiratory signal. It is a reconstructed heartbeat signal.

[0014] Step 4: Construct an SSR model and perform sparse spectrum reconstruction of the heartbeat signal based on the ZA-EFRLS algorithm to extract the heartbeat frequency; like Figure 4 As shown, a sparse frequency domain representation model of the heartbeat signal is established; the respiratory and heartbeat spectra are reconstructed based on the following undetermined linear equations: (12) In the formula, It is a length of The column vectors are the reconstructed heartbeat signals; , It is a length of The column vectors are unknown solutions to the intrinsic sparsity of the heartbeat spectrum; These are noise signals from the environment and the body; It is The basis matrix is ​​represented by the following elements: (13) The cost function of the SSR model is: (14) in, Forgetting factor, The recursive error of the SSR is used; since the standard RLS algorithm cannot directly generate sparse solutions, a sparse penalty function is introduced, forcing the unknown solution of the heartbeat spectrum. The goal is to make as many elements as possible approach 0, thus obtaining a sparse coefficient solution. Norms are the most commonly used sparsity penalty functions, which... of Introducing the norm into the cost function of SSR yields a new cost function: (15) Introduce an iteration step size This controls the update magnitude at each step. The gradient descent of the heartbeat signal spectral vector is recursively calculated as follows: (16) (17) (18) (19) (20) (twenty one) yes The soft thresholding operation of the norm sparsity penalty term sets the function value to 1 when the value is positive and to -1 when the value is negative. Since noise can cause impulse interference and instability in gradient descent, to suppress sudden impulse interference, [the following is omitted as it's not relevant to the context]. Increased Soft threshold operation; The gradient descent of the updated heartbeat signal spectrum vector is recursively described as follows: (twenty two) Finally, by iteratively calculating according to the above formula, the reconstructed heartbeat spectrum is obtained. The precise heart rate can be obtained by searching for spectral peaks.

[0015] Any aspects not covered in this invention are applicable to existing technologies.

Claims

1. A method for detecting vital signs using FMCW radar combining VMD and SSR, characterized in that, Includes the following steps: Step 1: Transmit chirped signals and capture radar echo signals using FMCW radar, preprocess the radar echo signals, and extract human chest wall displacement signals. Step 2: The VMD algorithm is used to decompose the human chest wall displacement signal. The optimal decomposition parameters are determined by using a variable step size two-stage search through the vital signs discrimination mode. The optimal decomposition parameters are then used to decompose the human chest wall displacement signal to obtain the modal component sets of heartbeat signal and respiratory signal. Step 3: Identify the respiratory harmonic mode components within the frequency range of the heartbeat signal and reconstruct the heartbeat signal; The respiratory harmonic mode component set was obtained through harmonic detection. : （4） in, , These are the modal component sets of heartbeat and respiratory signals, respectively. For integers ranging from 0 to 2, , The first , The center frequency of each modal component; The intersection of the respiratory signal modal component set and the respiratory harmonic modal component set is used to obtain the pure respiratory signal modal component set. The difference between the heartbeat signal modal component set and the respiratory harmonic modal component set is used to obtain the pure heartbeat signal modal component set. The modal components in the pure respiratory signal modal component set and the heartbeat signal modal component set are superimposed in the time domain to obtain the reconstructed respiratory signal and heartbeat signal. Step 4: Construct an SSR model, perform sparse spectrum reconstruction of the heartbeat signal based on the ZA-EFRLS algorithm, solve the heartbeat signal, and extract the heartbeat frequency.

2. The FMCW radar vital sign detection method combining VMD and SSR according to claim 1, characterized in that, In the second step, the energy loss rate is calculated according to equation (6): （6） In the formula, The total energy of all modal components. For the first Modal components Total energy, The number of modal components; Based on the energy proportion of respiratory and heartbeat signals in each modal component, the modal components belonging to respiratory and heartbeat signals are selected to obtain the respiratory signal modal component set and the heartbeat signal modal component set; （7） （8） In the formula, For the first The energy of the respiratory signal in each modal component For the first The energy of the center jump signal of each modal component and The preset threshold; The optimal decomposition parameters are selected based on the minimum energy loss rate and the fact that neither the respiratory signal modal component set nor the heartbeat signal modal component set is empty.

Images