A multi-person vital sign monitoring method based on millimeter-wave radar

Abstract

The present invention discloses a multi-person vital sign monitoring method based on millimeter-wave radar, comprising: using a millimeter-wave radar to transmit a frequency-modulated continuous wave signal to a human body and extracting an intermediate frequency signal; estimating the azimuth angle of the human body relative to the radar and using digital beamforming; generating a distance-azimuth heat map; estimating the elevation angle of the human body relative to the radar and extracting a distance-azimuth heat map corresponding to the obtained elevation angle; determining detection points of different human bodies and extracting intermediate frequency signals of different human bodies based on the distance-azimuth-elevation angle; and estimating heart rate and respiratory rate. The present invention realizes non-contact multi-person vital sign monitoring. By jointly estimating the distance-azimuth-elevation angle, the echo signals of different human bodies in the radar monitoring scene can be effectively separated, thereby removing interference signals in the monitoring scene. In addition, the heart rate and respiratory rate are jointly estimated using multiple algorithms, thereby improving monitoring accuracy and ensuring good monitoring effects.

Description

Technical Field

[0001] The present invention belongs to the field of millimeter-wave radar vital signs monitoring, relates to radar signal processing and array signal processing technology, and specifically relates to a multi-person vital signs monitoring method based on millimeter-wave radar. Background Art

[0002] With the rapid development of 5G communications and the Internet of Things (IoT) technology, smart health monitoring devices are becoming increasingly commonplace in people's daily lives. By monitoring various physiological parameters, the health of various organs and muscle activity can be managed. Vital signs such as breathing and heartbeat are crucial for determining the normality of cardiopulmonary activity. Accurate, real-time monitoring of these characteristic parameters plays a crucial role in promoting the development of clinical disease diagnosis, health monitoring, and other fields.

[0003] Traditional vital signs monitoring systems are mostly contact devices. In recent years, products such as electrocardiogram monitoring and information-based monitors have been developed. These products have rich functions, high measurement accuracy, and good stability, but they all require users to wear them all the time to obtain vital signs information in real time.

[0004] To address the pain points of contact-based vital signs monitoring systems, non-contact vital signs monitoring technology has emerged. Millimeter-wave radar has the characteristics of low power consumption, strong anti-interference ability, and high resolution. It can accurately and efficiently monitor breathing and heart rate in real time without direct contact with the human body, and has good comfort. Therefore, vital signs monitoring technology based on millimeter-wave radar has broad application prospects in the field of intelligent health monitoring.

[0005] Breathing and heartbeats cause the human chest to rise and fall, which in turn causes changes in the radial distance between the radar and the person being measured. Measuring the change in radial distance from the radar echo signal and estimating the human breathing and heartbeat rate is the key to radar detection of vital signs. Frequency Modulated Continuous Wave (FMCW) in the millimeter wave band can achieve a large bandwidth, which enables the radar to obtain better distance resolution. The literature "G. Sacco, E. Piuzzi, E. Pittella, et al. An FMCW Radar for Localization and Vital Signs Measurement for Different Chest Orientations [J]. Sensors, 2020, 20 (12): 3489." uses FMCW radar with a high-gain antenna to study the problem of vital sign detection at different angles of the chest relative to the radar radiation direction. For FMCW radar, accurate spectrum estimation algorithm determines the accuracy and reliability of human vital sign detection. The paper "H. Chang, C. Lin, Y. Lin, et al. DL-Aided NOMP: a Deep Learning-Based Vital Sign Estimating Scheme Using FMCW Radar[C] / / 2020IEEE 91st Vehicular Technology Conference (VTC2020-Spring). IEEE, 2020: 1-7." uses the matching pursuit algorithm combined with deep learning methods to enhance the estimation accuracy of respiration and heartbeat under low signal-to-noise ratio. The paper "C. Huang, G. Fang, H. Chuang, et al. Clutter-Resistant Vital Sign Detection Using Amplitude-Based Demodulation by EEMD-PCA-Correlation Algorithm for FMCW Radar Systems[C] / / 201949thEuropean Microwave Conference(EuMC).IEEE,2019:928-931." uses the amplitude change of radar echo on the slow time axis to characterize vital sign signals, and adopts ensemble empirical mode decomposition combined with principal component analysis to further improve the accuracy of heart rate and respiratory rate estimation.The paper "SUN L, HUANG S, LI Y, et al. Remote Measurement of Human Vital Signs Based on Joint-Range Adaptive EEMD[J]. IEEE Access, IEEE Access, 2020, 8: 68514–68524." proposed an adaptive ensemble empirical mode decomposition method, and simultaneously used phase information at multiple distances to achieve adaptive separation and estimation of heartbeat and respiratory signals, limiting the estimation error to less than 6 bpm. The paper "LEE H, KIM BH, PARK JK, et al. A Novel Vital-Sign Sensing Algorithm for Multiple Subjects Based on 24-GHz FMCW Doppler Radar[J]. Remote Sensing, Remote Sensing, 2019, 11(10): 1237." uses super-resolution spectrum estimation technology to make up for the range resolution of the radar, and combines the phase information extracted by the FFT algorithm to achieve vital sign detection, which can better distinguish multiple targets at distance and achieve multi-target detection.

[0006] However, most studies focus only on single-person vital sign monitoring. In contrast, the difficulty of multi-person respiratory and heartbeat monitoring lies in separating and extracting radar echoes from multiple individuals. In this scenario, multiple non-ideal factors exist, such as signal arrival angle mismatch, steering vector error, and the presence of desired signal components in the sample covariance matrix, which reduce the efficiency of vital sign signal separation. Summary of the Invention

[0007] Purpose of the invention: In order to overcome the problem of lack of multi-person vital signs monitoring technology in the existing technology, a multi-person vital signs monitoring method based on millimeter-wave radar is provided, which realizes non-contact multi-person vital signs monitoring. Through the joint estimation of distance-azimuth-elevation angle, the echo signals of different human bodies are effectively separated, and the interference signals in the monitoring scene are removed. The heart rate and respiratory rate are jointly estimated by multiple algorithms, which improves the monitoring accuracy and ensures good monitoring effect.

[0008] Technical solution: To achieve the above objectives, the present invention provides a method for monitoring vital signs of multiple people based on millimeter-wave radar, comprising the following steps:

[0009] S1: Use millimeter-wave radar to transmit frequency modulated continuous wave (FMCW) signals to the human body, and mix the echo signals to extract the intermediate frequency signals;

[0010] S2: Use the Multiple Signal Classification (Music) algorithm to estimate the azimuth of the human body relative to the radar and perform receive beamforming;

[0011] S3: Perform a range-dimensional Fast Fourier Transform (FFT) on the extracted intermediate frequency signal to obtain the position of the range gates where different human bodies are located, and remove static interference from each range gate to generate a range-azimuth heat map;

[0012] S4: Use the Music algorithm to estimate the elevation angle of the human body relative to the radar and extract the range-azimuth heat map corresponding to the obtained elevation angle;

[0013] S5: A two-dimensional constant false-alarm rate (2D-CFAR) detection algorithm is used to determine the detection points of different human bodies, and the intermediate frequency signals of different human bodies are extracted based on the distance-azimuth-elevation angle.

[0014] S6: Extract the phase difference signal from the intermediate frequency signals after separation of different human bodies, and use FFT, peak spacing, cross-correlation and other algorithms to estimate their heart rate and respiratory rate.

[0015] Furthermore, the steps of extracting the intermediate frequency signal in step S1 are:

[0016] A1: Assume that the number of transmitting antennas is M T , the number of receiving antennas is M R , the array configuration is a one-dimensional uniform linear array (ULA), the array element spacing is half a wavelength, and the baseband waveform is a linear frequency modulation (LFM) signal. The transmitted signal of each antenna is:

[0017]

[0018] Among them, f c is the carrier frequency; B is the signal bandwidth; T is the pulse width, and the far-field transmission signal of the radar at the azimuth angle θ is

[0019] s T (t;θ)=a H (θ)s(t) (2)

[0020] in, is the transmit signal matrix; (·) H represents the conjugate transpose of the matrix; θ represents the azimuth angle; is the launch steering vector, and its expression is:

[0021] a(θ)=[1,exp(jπsinθ),...,exp(jπ(M T -1)sinθ)] H (3)

[0022] A2: Assume that there are L, L≤M in the scene R If there are a number of monitored persons, the radar echo signal is:

[0023]

[0024] in, τ l =R l / c is the time delay of the lth monitored person relative to the radar; R l is the distance of the lth monitored person relative to the radar; c is the speed of light; To receive noise; To receive the steering vector, its expression is:

[0025] b(θ)=[1,exp(jπsinθ),...,exp(jπ(M R -1)sinθ)] H (5)

[0026] A3: After mixing the echo signal with the transmitted signal, it can be expressed as

[0027]

[0028] in,(·) * represents the conjugate operation; is the intermediate frequency signal matrix of the lth human target, and the expression of each element is

[0029]

[0030] Where λ represents the wavelength; and They represent the frequency and phase of the intermediate frequency signal of the lth human target, and their expressions are

[0031]

[0032]

[0033] Furthermore, the step of using the Music algorithm to estimate the azimuth angle of the human body relative to the radar in step S2 is:

[0034] B1: Calculate the covariance matrix of the received signal as:

[0035] R=y(t)y H (t) (9)

[0036] B2: Calculate the eigenvalue decomposition of the covariance matrix as:

[0037] R=UΣU H (10)

[0038] in, Represented by the eigenvalue ξ m ,m=1,...,M R The symmetric matrix formed; Represented by the eigenvector The unitary matrix formed by

[0039] B3: Sort the eigenvalues ​​by size, regard the eigenvectors corresponding to the L largest eigenvalues ​​as the signal subspace, and the remaining M R The eigenvectors corresponding to the -L small eigenvalues ​​are regarded as noise subspaces, and the covariance matrix can be further expressed as:

[0040]

[0041] Wherein, the subscripts s and n represent signal and noise, respectively;

[0042] B4: Calculate the peak function as:

[0043]

[0044] B5: The azimuth of the first human body relative to the radar can be estimated as:

[0045]

[0046] Furthermore, the receive beamforming step in step S2 is as follows:

[0047]

[0048] Among them, A l =b H (θ l )b(θ l ) is the gain of the first monitored person after receiving beamforming; n l (t) = b H (θ l )n(t) is the noise of the echo of the lth monitored person.

[0049] Furthermore, in step S3, a range-dimensional FFT is performed on the intermediate frequency signal to obtain a median frequency, and the time delay of the human body relative to the radar is estimated using the obtained median frequency to obtain the position of the range gate where different human bodies are located. After ignoring the transmitting and receiving steering vectors, the FFT of the intermediate frequency signal of L human bodies is calculated as:

[0050]

[0051] in, represents a vector of all 1s.

[0052] Furthermore, the step of removing static interference in step S3 is:

[0053] C1: Assume the number of radar pulses is N c , the number of sampling points for a single pulse is N s , then the data of the cth pulse after the distance dimension FFT is:

[0054] z c =[z c (1),...,z c (N s )] (16)

[0055] Among them, z n,c (n),n=1,...,N s Represents the value of the data of the nth sampling point of the cth pulse after the distance dimension FFT;

[0056] C2: The average value of the FFT of all pulse distances of the nth receiving antenna is:

[0057]

[0058] C3: The data of the cth pulse after removing static interference is:

[0059]

[0060] Furthermore, the calculation formula of the range-azimuth heat map in step S3 is:

[0061]

[0062] in,(·) -1 represents the inverse of a matrix; M R The antennas receive N c The covariance matrix of the pulse data after the range dimension FFT is expressed as:

[0063]

[0064] in, represents the Kronecker product; Indicates N c The FFT data of a pulse is expressed as:

[0065]

[0066] Furthermore, the step of using the Music algorithm to estimate the elevation angle of the human body relative to the radar in step S4 is as follows: Assume that the vertical direction of the receiving array has M K antennas, calculate the one-dimensional elevation spectrum as:

[0067]

[0068] in, M R Planar antennas and M K Vertical receiving N c The covariance matrix of the pulse data after the distance dimension FFT is expressed as

[0069]

[0070] in, is the two-dimensional steering vector of azimuth-elevation, expressed as:

[0071]

[0072] in, is the vertical direction steering vector, expressed as:

[0073] c(θ)=[1,exp(jπsinθ),...,exp(jπ(M K -1)sinθ)] H (25)

[0074] Furthermore, the steps of using the 2D-CFAR detection algorithm to determine the detection points of different human bodies in step S5 are: setting the false alarm rate and the distance-azimuth window size, estimating the noise parameters based on the Gaussian distribution probability density, calculating the threshold, judging whether the entire heat map has been detected, and determining the detection points.

[0075] Furthermore, the process of estimating the human heart rate and respiratory rate in step S6 is as follows:

[0076] D1: For the intermediate frequency signal, the phase is extracted once within each frame period. If the distance, azimuth, and elevation of the human body and the radar change, the distance, azimuth, and elevation of the human body need to be tracked to re-determine the detection point at that time. The phase is then extracted and the target phase is transmitted repeatedly for N frames. This curve of the target phase changing with the number of frames can be obtained, which can also be viewed as the relationship between the target phase and time. The extracted phase signal is phase unwrapped to ensure that the phase value is between [-π, π]. Whenever the phase difference between consecutive values ​​is greater than / less than ±π, 2π is subtracted / added from the phase. The phase signal is subtracted frame by frame to obtain a phase difference signal.

[0077] D2: Based on the difference in breathing and heartbeat frequencies, a bandpass filter is used to filter and separate the phase difference signals of breathing and heartbeat;

[0078] D3: Filter out motion interference from the heartbeat phase difference signal. (Heart rate measurement is based on the distance difference caused by the tiny movements of the heart's contraction and relaxation, which causes phase changes. According to the micro-Doppler principle, large body sway will affect its accuracy.) Here, the sample is segmented and an energy threshold is set to determine whether it meets the heart rate variation range. High-energy data segments are discarded and stable data segments are integrated for further estimation.

[0079] D4: Use multiple algorithms such as FFT in the frequency domain and peak spacing and cross-correlation in the time domain to process the phase difference signals of breathing and heartbeat, and take the average of the results of the three algorithms to obtain the current breathing and heartbeat frequencies;

[0080] D5: Record the respiratory and heart rate over a period of time, determine the actual frequency at that time based on the confidence index, and output the relationship between the respiratory and heart rate changes over time;

[0081] D6: Filter out the harmonics of the heart rate and the respiratory rate (the heart rate here is likely to be a harmonic of the respiratory rate. For example, if the respiratory rate is 30 and the heart rate is 60, 90, 120, etc., then the heart rate here is likely to be the harmonic interference of the respiratory rate rather than the actual heart rate).

[0082] The present invention provides a multi-person vital sign monitoring method based on millimeter-wave radar. First, a millimeter-wave radar is used to transmit an FMCW signal to the human body, and the radar echo signal is mixed with the transmitted signal to obtain an intermediate frequency signal. Then, an FFT algorithm and a MUSIC algorithm are used respectively to estimate the distance, azimuth, and elevation of the human body relative to the radar, and a range-azimuth-elevation profile is generated. A 2D-CFAR algorithm is used to determine the detection points of different human bodies. Finally, phase signals of different human bodies are extracted from the target detection points to estimate the respiratory and heart rates.

[0083] The above solution can be summarized into the following two steps:

[0084] (1) Millimeter-wave radar is used to collect the intermediate frequency signals of human vital signs, and algorithms such as FFT, MUSIC, and 2D-CFAR are used to find the distance-azimuth-elevation parameters of different human bodies. The intermediate frequency signals of different human bodies are separated and extracted according to the different parameters.

[0085] (2) Extract phase information from the intermediate frequency signal and use FFT, peak spacing, cross-correlation, confidence level and other algorithms to estimate the heart rate and respiratory rate of each person and obtain vital signs information.

[0086] Beneficial effects: Compared with the existing technology, the present invention realizes non-contact multi-person vital signs monitoring. Through the joint estimation of distance, azimuth and elevation, it effectively separates the echo signals of different human bodies, removes the interference signals in the monitoring scene, and uses multiple algorithms to jointly estimate the heart rate and respiratory rate, thereby improving the monitoring accuracy and ensuring good monitoring effects. BRIEF DESCRIPTION OF THE DRAWINGS

[0087] Figure 1 It is a schematic flow diagram of the present invention;

[0088] Figure 2 This is a flow chart of the algorithm for estimating heart rate and respiratory rate;

[0089] Figure 3 It is the millimeter wave radar regional monitoring map;

[0090] Figure 4 is the breathing and heartbeat waveform of person 1, where Figure 4 (a) is the respiratory waveform diagram, Figure 4 (b) is a heartbeat waveform diagram;

[0091] Figure 5 is the breathing and heartbeat waveform of person 2, where Figure 5 (a) is the respiratory waveform diagram, Figure 5 (b) is a heartbeat waveform diagram. DETAILED DESCRIPTION

[0092] The present invention is further illustrated below with reference to the accompanying drawings and specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and are not used to limit the scope of the present invention. After reading the present invention, modifications of various equivalent forms of the present invention made by those skilled in the art all fall within the scope defined by the claims attached to this application.

[0093] 1. The present invention provides a method for monitoring the vital signs of multiple people based on millimeter wave radar, such as Figure 1 As shown, the following steps are included:

[0094] S1: Use millimeter-wave radar to transmit FMCW signals to the human body and mix the echo signals to extract the intermediate frequency signals;

[0095] S2: Use the Music algorithm to estimate the azimuth angle of the human body relative to the radar and perform receive beamforming;

[0096] S3: Performing range-dimensional FFT on the intermediate frequency signal and removing static interference from each range gate to generate a range-azimuth heat map;

[0097] S4: Use the Music algorithm to estimate the elevation angle of the human body relative to the radar and extract the range-azimuth heat map corresponding to the obtained elevation angle;

[0098] S5: Use the 2D-CFAR detection algorithm to determine the detection points of different human bodies, and extract the intermediate frequency signals of different human bodies based on distance-azimuth-elevation;

[0099] S6: extracting phase difference signals from the intermediate frequency signals after separation of different human bodies, and estimating their heart rates and respiratory rates using algorithms such as FFT, peak spacing, and cross-correlation.

[0100] The steps for extracting the intermediate frequency signal in step S1 are:

[0101] A1: Assume that the number of transmitting antennas is M T , the number of receiving antennas is M R , the array configuration is a one-dimensional uniform linear array (ULA), the array element spacing is half a wavelength, and the baseband waveform is a linear frequency modulation (LFM) signal. The transmitted signal of each antenna is:

[0102]

[0103] Among them, f c is the carrier frequency; B is the signal bandwidth; T is the pulse width, and the far-field transmission signal of the radar at the azimuth angle θ is

[0104] s T (t;θ)=a H (θ)s(t) (2)

[0105] in, is the transmit signal matrix; (·) H represents the conjugate transpose of the matrix; θ represents the azimuth angle; is the launch steering vector, and its expression is:

[0106] a(θ)=[1,exp(jπsinθ),...,exp(jπ(M T -1)sinθ)]H (3)

[0107] A2: Assume that there are L, L≤M in the scene R If there are a number of monitored persons, the radar echo signal is:

[0108]

[0109] in, τ l =R l / c is the time delay of the lth monitored person relative to the radar; R l is the distance of the lth monitored person relative to the radar; c is the speed of light; To receive noise; To receive the steering vector, its expression is:

[0110] b(θ)=[1,exp(jπsinθ),...,exp(jπ(M R -1)sinθ)] H (5)

[0111] A3: After mixing the echo signal with the transmitted signal, it can be expressed as

[0112]

[0113] in,(·) * represents the conjugate operation; is the intermediate frequency signal matrix of the lth human target, and the expression of each element is

[0114]

[0115] Where λ represents the wavelength; and They represent the frequency and phase of the intermediate frequency signal of the lth human target, and their expressions are

[0116]

[0117]

[0118] The steps of using the Music algorithm to estimate the azimuth of the human body relative to the radar in step S2 are:

[0119] B1: Calculate the covariance matrix of the received signal as:

[0120] R=y(t)y H (t) (9)

[0121] B2: Calculate the eigenvalue decomposition of the covariance matrix as:

[0122] R=UΣU H (10)

[0123] in, Represented by the eigenvalue ξ m ,m=1,...,M R The symmetric matrix formed; Represented by the eigenvector The unitary matrix formed by

[0124] B3: Sort the eigenvalues ​​by size, regard the eigenvectors corresponding to the L largest eigenvalues ​​as the signal subspace, and the remaining M R The eigenvectors corresponding to the -L small eigenvalues ​​are regarded as noise subspaces, and the covariance matrix can be further expressed as:

[0125]

[0126] Wherein, the subscripts s and n represent signal and noise, respectively;

[0127] B4: Calculate the peak function as:

[0128]

[0129] B5: The azimuth of the first human body relative to the radar can be estimated as:

[0130]

[0131] The steps of receiving beam forming in step S2 are:

[0132]

[0133] Among them, A l =b H (θ l )b(θ l ) is the gain of the first monitored person after receiving beamforming; n l (t) = b H (θ l )n(t) is the noise of the echo of the lth monitored person.

[0134] In step S3, the intermediate frequency signal is subjected to a range-dimensional FFT to obtain the median frequency. The time delay of the human body relative to the radar is estimated using the obtained median frequency to obtain the position of the range gate where different human bodies are located. After ignoring the transmitting and receiving steering vectors, the FFT of the intermediate frequency signal of L human bodies is calculated as:

[0135]

[0136] in, represents a vector of all 1s.

[0137] The steps for removing static interference in step S3 are:

[0138] C1: Assume the number of radar pulses is N c , the number of sampling points for a single pulse is N s , then the data of the cth pulse after the distance dimension FFT is:

[0139] z c =[z c (1),...,z c (N s )] (16)

[0140] Among them, z n,c (n),n=1,...,N s Represents the value of the data of the nth sampling point of the cth pulse after the distance dimension FFT;

[0141] C2: The average value of the FFT of all pulse distances of the nth receiving antenna is:

[0142]

[0143] C3: The data of the cth pulse after removing static interference is:

[0144]

[0145] The calculation formula of the range-azimuth heat map in step S3 is:

[0146]

[0147] in,(·) -1 represents the inverse of a matrix; M R The antennas receive N c The covariance matrix of the pulse data after the range dimension FFT is expressed as:

[0148]

[0149] in, represents the Kronecker product; Indicates N c The FFT data of a pulse is expressed as:

[0150]

[0151] The steps of using the Music algorithm to estimate the elevation angle of the human body relative to the radar in step S4 are as follows: Assume that the vertical direction of the receiving array has M K antennas, calculate the one-dimensional elevation spectrum as:

[0152]

[0153] in, M R Planar antennas and M K Vertical receiving N c The covariance matrix of the pulse data after the distance dimension FFT is expressed as

[0154]

[0155] in, is the two-dimensional steering vector of azimuth-elevation, expressed as:

[0156]

[0157] in, is the vertical direction steering vector, expressed as:

[0158] c(θ)=[1,exp(jπsinθ),...,exp(jπ(M K -1)sinθ)] H (25)

[0159] In step S5, the steps of using the 2D-CFAR detection algorithm to determine the detection points of different human bodies are as follows: setting the false alarm rate and the range-azimuth window size, estimating the noise parameters based on the Gaussian distribution probability density, calculating the threshold, judging whether the entire heat map has been detected, and determining the detection points.

[0160] like Figure 2 As shown, the estimation process of the human heart rate and respiratory rate in step S6 is as follows:

[0161] D1: For the intermediate frequency signal, the phase is extracted once within each frame period. If the distance, azimuth, and elevation of the human body and the radar change, the distance, azimuth, and elevation of the human body need to be tracked to re-determine the detection point at that time. The phase is then extracted and the target phase is transmitted repeatedly for N frames. This curve of the target phase changing with the number of frames can be obtained, which can also be viewed as the relationship between the target phase and time. The extracted phase signal is phase unwrapped to ensure that the phase value is between [-π, π]. Whenever the phase difference between consecutive values ​​is greater than / less than ±π, 2π is subtracted / added from the phase. The phase signal is subtracted frame by frame to obtain a phase difference signal.

[0162] D2: Based on the difference between the breathing and heartbeat frequencies, the phase difference signals of breathing and heartbeat are filtered and separated using a bandpass filter;

[0163] D3: Filter out motion interference from the heartbeat phase difference signal. (Heart rate measurement is based on the distance difference caused by the tiny movements of the heart's contraction and relaxation, which causes phase changes. According to the micro-Doppler principle, large body sway will affect its accuracy.) Here, the sample is segmented and an energy threshold is set to determine whether it meets the heart rate variation range. High-energy data segments are discarded and stable data segments are integrated for further estimation.

[0164] D4: Use multiple algorithms such as FFT in the frequency domain and peak spacing and cross-correlation in the time domain to process the phase difference signals of breathing and heartbeat, and take the average of the results of the three algorithms to obtain the current breathing and heartbeat frequencies;

[0165] D5: Record the respiratory and heart rate over a period of time, determine the actual frequency at that time based on the confidence index, and output the relationship between the respiratory and heart rate changes over time;

[0166] D6: Filter out the harmonics of the heart rate and the respiratory rate (the heart rate here is likely to be a harmonic of the respiratory rate. For example, if the respiratory rate is 30 and the heart rate is 60, 90, 120, etc., then the heart rate here is likely to be the harmonic interference of the respiratory rate rather than the actual heart rate).

[0167] 2. Based on the above technical solution, in order to verify the effect of the method of the present invention, this embodiment conducts experimental verification, which is as follows:

[0168] A simulation example is given below to verify the effectiveness of the present invention. This example uses a 3-transmitter and 4-receiver co-located millimeter-wave radar system, and the simulation environment is MATLAB R2021a.

[0169] The specific experimental process is:

[0170] Step 1: Initialize system parameters

[0171] The configuration of the radar system is given in Table 1.

[0172] Table 1 Radar sensor parameters

[0173]

[0174]

[0175] Step 2: The millimeter-wave radar transmits FMCW to the monitoring area;

[0176] Step 3: Generate a range-azimuth-elevation heatmap

[0177] Figure 3 A range-azimuth-elevation heat map is given. Figure 3As can be seen, the monitoring range of millimeter-wave radar is 0-3 meters, and the monitoring azimuth is -60° to 60°. The two red boxes in the figure indicate the detection points where the two monitored people are currently located. By extracting the corresponding intermediate frequency signals from the target detection points, the radar echoes of different monitored people can be separated.

[0178] Step 4: Display the respiratory and heartbeat phase difference waveforms and estimate the respiratory and heartbeat rates

[0179] Figure 4 and Figure 5 The phase difference waveforms of the breathing and heartbeat of the two monitored persons for 110 frames are given, in which the large interference and large noise data are removed, and the harmonic interference of the breathing signal is removed from the heartbeat waveform. Figure 4 and Figure 5 Given the phase difference waveform, it can be calculated that the respiratory and heart rates of monitored person 1 are 22 times / minute and 77 times / minute, and the respiratory and heart rates of monitored person 2 are 18 times / minute and 83 times / minute.

Claims

1. A method for monitoring multiple vital signs based on millimeter wave radar, characterized in that: The steps include: S1: Use millimeter-wave radar to transmit frequency-modulated continuous wave signals to the human body, and mix the echo signals to extract intermediate frequency signals; S2: Use the multiple signal classification algorithm to estimate the azimuth of the human body relative to the radar and perform receive beamforming; S3: Performing a range-dimensional fast Fourier transform on the extracted intermediate frequency signal to obtain the position of the range gates where different human bodies are located, and removing static interference from each range gate to generate a range-azimuth heat map; S4: Use the Music algorithm to estimate the elevation angle of the human body relative to the radar and extract the range-azimuth heat map corresponding to the obtained elevation angle; S5: A two-dimensional constant false alarm rate (CFAR) detection algorithm is used to determine the detection points of different human bodies, and the intermediate frequency signals of different human bodies are extracted based on the distance-azimuth-elevation angle. S6: extracting phase difference signals from the intermediate frequency signals separated from different human bodies, and estimating the heart rate and respiratory rate of the human body based on the phase difference signals; The steps of using the Music algorithm to estimate the azimuth of the human body relative to the radar in step S2 are: B1: Calculate the covariance matrix of the received signal as: R=y(t)y H (t) (9) B2: Calculate the eigenvalue decomposition of the covariance matrix as: R=UΣU H (10) in, Represented by the eigenvalue ξ m ,m=1,...,M R The symmetric matrix formed; Represented by the eigenvector The unitary matrix formed by B3: Sort the eigenvalues ​​by size, regard the eigenvectors corresponding to the L largest eigenvalues ​​as the signal subspace, and the remaining M R The eigenvectors corresponding to the -L small eigenvalues ​​are regarded as noise subspaces, and the covariance matrix can be further expressed as: Wherein, the subscripts s and n represent signal and noise, respectively; B4: Calculate the peak function as: B5: The azimuth of the first human body relative to the radar can be estimated as: The steps of receiving beam forming in step S2 are: Among them, A l =b H (θ l )b(θ l ) is the gain of the first monitored person after receiving beamforming; n l (t) = b H (θ l )n(t) is the noise of the echo of the lth monitored person; In step S3, the intermediate frequency signal is subjected to a range-dimensional FFT to obtain the median frequency. The time delay of the human body relative to the radar is estimated using the obtained median frequency to obtain the position of the range gate where different human bodies are located. After ignoring the transmitting and receiving steering vectors, the FFT of the intermediate frequency signal of L human bodies is calculated as: in, represents a vector of all 1s; The steps of removing static interference in step S3 are: C1: Assume the number of radar pulses is N c , the number of sampling points for a single pulse is N s , then the data of the cth pulse after the distance dimension FFT is: With c =[z c (1),...,of c (N s )] (16) Among them, z n,c (n),n=1,…,N s Represents the value of the data of the nth sampling point of the cth pulse after the range dimension FFT; C2: The average value of the FFT of all pulse distance dimensions of the nth receiving antenna is: C3: The data of the cth pulse after removing static interference is: The calculation formula of the range-azimuth heat map in step S3 is: in,(·) -1 represents the inverse of a matrix; M R The antennas receive N c The covariance matrix of the pulse data after the range dimension FFT is expressed as: in, represents the Kronecker product; Indicates N c The FFT data of a pulse is expressed as: The steps of using the Music algorithm to estimate the elevation angle of the human body relative to the radar in step S4 are as follows: Assume that the vertical direction of the receiving array has M K antennas, calculate the one-dimensional elevation spectrum as: in, M R Planar antennas and M K Vertical receiving N c The covariance matrix of the pulse data after the distance dimension FFT is expressed as in, is the two-dimensional steering vector of azimuth-elevation, expressed as: in, is the vertical direction steering vector, expressed as:

2. The method for monitoring multiple vital signs based on millimeter wave radar according to claim 1, characterized in that: The steps of extracting the intermediate frequency signal in step S1 are: A1: Assume that the number of transmitting antennas is M T , the number of receiving antennas is M R , the array configuration is a one-dimensional uniform linear array, the array element spacing is half a wavelength, and the baseband waveform is a linear frequency modulation signal. Then the transmitted signal of each antenna is: Among them, f c is the carrier frequency; B is the signal bandwidth; T is the pulse width, and the far-field transmission signal of the radar at the azimuth angle θ is s T (t;θ)=a H (θ)s(t) (2) in, is the transmit signal matrix; (·) H represents the conjugate transpose of the matrix; θ represents the azimuth angle; is the launch steering vector, and its expression is: a(θ)=[1,exp(jπsinθ),...,exp(jπ(M T -1)sinθ)] H (3) A2: Assume that there are L, L≤M in the scene R If there are a number of monitored persons, the radar echo signal is: in, τ l =R l / c is the time delay of the lth monitored person relative to the radar; R l is the distance of the lth monitored person relative to the radar; c is the speed of light; To receive noise; To receive the steering vector, its expression is: b(θ)=[1,exp(jπsinθ),...,exp(jπ(M R -1)sinθ)] H (5) A3: After mixing the echo signal with the transmitted signal, it can be expressed as in,(·) * represents the conjugate operation; is the intermediate frequency signal matrix of the lth human target, and the expression of each element is Where λ represents the wavelength; and They represent the frequency and phase of the intermediate frequency signal of the lth human target, and their expressions are 3. The method for monitoring multiple vital signs based on millimeter wave radar according to claim 1, characterized in that: The steps of using the 2D-CFAR detection algorithm to determine the detection points of different human bodies in step S5 are: setting the false alarm rate and the range-azimuth window size, estimating the noise parameters based on the Gaussian distribution probability density, calculating the threshold, judging whether the entire heat map has been detected, and determining the detection points.

4. The method for monitoring multiple vital signs based on millimeter wave radar according to claim 1, characterized in that: The estimation process of the human heart rate and respiratory rate in step S6 is as follows: D1: For the intermediate frequency signal, the phase is extracted once within each frame period. If the distance, azimuth, and elevation of the human body and the radar change, the distance, azimuth, and elevation of the human body need to be tracked to re-determine the detection point at that time. The phase is then extracted and the target phase is transmitted repeatedly for N frames. This curve of the target phase changing with the number of frames can be obtained, which can also be viewed as the relationship between the target phase and time. The extracted phase signal is phase unwrapped to ensure that the phase value is between [-π, π]. Whenever the phase difference between consecutive values ​​is greater than / less than ±π, 2π is subtracted / added from the phase. The phase signal is subtracted frame by frame to obtain a phase difference signal. D2: Based on the difference in breathing and heartbeat frequencies, a bandpass filter is used to filter and separate the phase difference signals of breathing and heartbeat; D3: Filter out motion interference from the heartbeat phase difference signal. This involves segmenting the sample and setting an energy threshold to determine whether it falls within the heart rate range. High-energy data segments are discarded, and stable data segments are integrated for further estimation. D4: Use multiple algorithms to process the phase difference signals of breathing and heartbeat, and take the average of the results of the three algorithms to obtain the current breathing and heartbeat frequencies; D5: Record the respiratory and heart rate over a period of time, determine the actual frequency at that time based on the confidence index, and output the relationship between the respiratory and heart rate changes over time; D6: Filter out the harmonics of the heart rate and respiratory rate.

Images