Millimeter-wave radar life signal extraction method based on VMD algorithm

Abstract

A method for extracting life signals from millimeter-wave radar based on VMD algorithm, the method specifically comprises the following steps: obtaining distance information D[P k ,M]; pre-process the distance information containing the target human breathing and heartbeat signals to obtain the vital sign phase information y r (t); The vital sign phase information y is decomposed by variational mode decomposition algorithm r (t) is decomposed to obtain the reconstructed life breathing signal y R (t) and heartbeat signal y H (t); The reconstructed heartbeat signal y is subjected to variational mode decomposition algorithm H (t) is decomposed to obtain the heartbeat signal y′ H (t). The present invention performs variational modal decomposition on the extracted vital sign phase information to separate the respiratory signal and the heartbeat signal. Since the human heartbeat signal is extremely weak, the extracted heartbeat signal contains a small amount of respiratory signal harmonic components, which greatly reduces the accuracy of the extracted heartbeat signal. This solves the problem of low accuracy in heartbeat signal extraction in the existing technology.

Description

Technical Field

[0001] The present invention belongs to the field of radar technology and relates to a variational mode decomposition (VMD) algorithm, in particular to a millimeter wave radar life signal extraction method based on the VMD algorithm. Background Art

[0002] With an aging population and the prevalence of various diseases at younger ages, we need to pay more attention to our own and our families' health, leading to increased interest in human health monitoring. Respiration and heartbeat are two crucial indicators of a person's vital signs. In medicine, doctors often use these vital signs to assess a patient's condition. Traditional clinical life detectors are primarily contact-based, requiring sensors to contact the skin to measure vital signs. However, this contact-based measurement method has limitations in both its application environment and scope.

[0003] For patients with severe skin burns, contact-based measurement methods can cause secondary infections and discomfort. In natural disasters like earthquakes, mudslides, and tsunamis, people in need of rescue are likely to be trapped under collapsed buildings or buried in the mud. Contact-based detectors are unable to detect life signs, delaying rescue efforts. Therefore, non-contact vital sign detection technology has emerged. Non-contact monitoring technology improves monitoring efficiency to a certain extent, allowing monitoring equipment to monitor vital signs without making contact with the human body.

[0004] Non-contact vital sign monitoring is typically achieved through infrared, sound waves, and radar. However, these methods also have limitations. For example, infrared detection is susceptible to temperature fluctuations and has poor penetration. If a trapped person is obscured by a wall or buried, infrared detection capabilities are weakened. When the ambient temperature approaches human body temperature, the sensitivity of the infrared sensor is affected, thereby affecting detection accuracy. Audio detection, which detects faint audio vibrations such as cries and groans, can determine the location of trapped people. However, in the event of natural disasters such as earthquakes and mudslides, detection probes are required and are susceptible to interference from ambient noise, resulting in poor stability and low efficiency. Summary of the Invention

[0005] In view of the shortcomings of the existing technology, the purpose of the present invention is to provide a millimeter-wave radar life signal extraction method based on the VMD algorithm to solve the technical problems of low accuracy, poor stability and low efficiency of the life signal extraction method in the existing technology.

[0006] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0007] A method for extracting life signals from millimeter-wave radar based on a VMD algorithm, the method specifically comprising the following steps:

[0008] Step 1: Get the distance information D[P k ,M];

[0009] Step 2: Preprocess the distance information containing the target human breathing and heartbeat signals to obtain the vital sign phase information y r (t);

[0010] Step 3: The phase information y of the vital signs is decomposed by the variational mode decomposition algorithm. r (t) is decomposed to obtain the reconstructed life breathing signal y R (t) and heartbeat signal y H (t);

[0011] Step 4: The reconstructed heartbeat signal y is decomposed by variational mode decomposition algorithm. H (t) is decomposed to obtain the heartbeat signal y′ H (t).

[0012] The present invention also includes the following technical features:

[0013] The first step is to use a millimeter-wave radar life detection system to obtain distance information containing the target human body's breathing and heartbeat signals. The millimeter-wave radar life detection system includes a transmitting end and a receiving end; the transmitting end includes a voltage-controlled oscillator, a waveform generator, and a transmitting antenna connected together; the receiving end includes a receiving antenna, a mixer, a bandpass filter, an analog-to-digital converter, and a baseband signal processing module connected in sequence; the voltage-controlled oscillator is connected to the mixer; the method specifically includes the following sub-steps:

[0014] Step 1.1: The voltage-controlled oscillator receives the RF signal from the waveform generator and modulates it. Part of the modulated signal is transmitted by the transmitting antenna, and the other part enters the mixer and serves as the mixer's local oscillator signal.

[0015] Step 1.2: The signal transmitted by the transmitting antenna is reflected by the target part, and the receiving antenna receives the reflected signal and sends it to the mixer;

[0016] Step 1.3: The mixer mixes the signal sent by the receiving antenna and the local oscillator signal, and sends the mixed signal to the bandpass filter for processing to obtain an intermediate frequency signal;

[0017] Step 1.4: The intermediate frequency signal enters the analog-to-digital converter for A / D sampling to obtain the data matrix of the echo signal, which is sent to the baseband signal processing module;

[0018] Step 1.5: The baseband signal processing module performs distance-dimensional Fourier transform on the data matrix of the echo signal to obtain the distance information matrix D[P k ,M];

[0019] in:

[0020] M represents the number of RF signals emitted by the waveform generator;

[0021] P k Indicates the sequence number of the distance unit, 1≤k≤256.

[0022] The pretreatment in step 2 specifically includes the following steps:

[0023] Step 2.1: The distance information D[P k ,M] is processed by the moving target display technology shown in formula 2-1 to obtain the distance information matrix D′[P k ,M];

[0024]

[0025] in:

[0026] M represents the number of RF signals emitted by the waveform generator;

[0027] P k Indicates the sequence number of the distance unit, 1≤k≤256;

[0028] D[P k ,M] represents the distance information matrix containing the target human breathing and heartbeat signals;

[0029] ∑D[P k ,:] represents the matrix D[P k ,M] k The sum of all data within a distance unit;

[0030] Step 2.2, the distance information matrix D′[P k ,M] perform the autocorrelation processing shown in formula 2-2 to obtain the data matrix S′[P k ,];

[0031] S′[P k ,N]=xcorr(D′[P k ,M]) (2-2)

[0032] in:

[0033] xcorr(·) is the autocorrelation function;

[0034] N = 2M-1, which represents the number of RF signals after autocorrelation processing;

[0035] Step 2.3: According to formulas (2-3) and (2-4), the vital sign phase information y is extracted. r (t);

[0036] [a,P a ]=max(max(S′[P k ,N] T )) (2-3)

[0037] y r (t) = phase (S′ [P a ,:]) (2-4)

[0038] in:

[0039] S′[P k ,N] T Denote the matrix S′[P k ,N];

[0040] max(S′[P k ,N] T ) represents the matrix S′[P k ,N] T The maximum value of each column forms a row vector;

[0041] max(max(S′[P k ,N])) means getting the maximum value a in this row vector and the distance unit P where the maximum value is located a ;

[0042] S′[P a ,:] represents the matrix S′[P k ,N] a All data within a distance unit;

[0043] phase(·) represents the phase function.

[0044] The step three specifically includes the following steps:

[0045] Step 3.1: Use the variational mode decomposition algorithm to transform the vital sign phase information y r (t) decomposed into K1 modal components IMF;

[0046] Step 3.2: Perform fast Fourier transform on each modal component IMF to obtain the spectrum corresponding to each modal component IMF. Calculate the total energy E() of each modal component IMF and the energy E of each modal component IMF in the heartbeat frequency band on the spectrum.H () and the energy E in the respiratory frequency band R ();

[0047] Step 3.3: Determine whether the following equation is true. If equation (3-1) is true, it is considered that this modal component IMF contains the heartbeat signal component and can be used to reconstruct the heartbeat signal. If not, it is considered that this modal component IMF does not contain the spectral characteristics of the human heartbeat signal.

[0048]

[0049] in:

[0050] i represents the serial number of the modal component, i=1, 2, ... K1;

[0051] δ represents the threshold value, and the value range is δ not less than 0.5;

[0052] Step 3.4, determine whether the following formula is true: Formula (3-2), then it is considered that this modal component IMF contains respiratory signal components and can be used to reconstruct the respiratory signal. If it is not satisfied, it is considered that this modal component IMF does not contain the spectral characteristics of the human respiratory signal;

[0053]

[0054] in:

[0055] i represents the decomposed modal component, i = 1, 2, ... K1;

[0056] δ represents the threshold value, and the value range is δ not less than 0.5;

[0057] Step 3.5: Add all the modal components IMF that satisfy formula (3-1) according to formula (3-3) to obtain the reconstructed heartbeat signal y H (t); add all the modal components IMF that satisfy formula (3-2) according to formula (3-4) to obtain the reconstructed respiratory signal y R (t);

[0058] y H (t)=∑IMF(t) (3-3)

[0059] y R (t)=∑IMF(t) (3-4)

[0060] t represents time, ranging from 0 to 50s.

[0061] The step 4 specifically includes the following steps:

[0062] Step 4.1, reconstruct the heartbeat signal y H(t) Perform variational mode decomposition algorithm to obtain K2 modal components IMF1;

[0063] Step 4.2: Perform fast Fourier transform on each modal component IMF1 to obtain the spectrum corresponding to each modal component IMF1. Calculate the total energy E1(j) of each modal component IMF1 on the spectrum. The energy of each modal component IMF1 in the heartbeat frequency band is

[0064] Step 4.3: Determine whether the following equation is true. If so, it is considered that the modal component IMF1 contains a heartbeat signal component and can be used to reconstruct the heartbeat signal. If not, it is considered that the modal component IMF1 does not contain the spectral characteristics of a human heartbeat signal.

[0065]

[0066] in:

[0067] j represents the serial number of the modal component, j = 1, 2, ... K2;

[0068] δ represents the threshold value, and the value range is δ not less than 0.5;

[0069] Step 4.4, add the modal components IMF1 that satisfy formula (4-1) according to formula (4-2) to obtain the reconstructed heartbeat signal y′ H (t):

[0070] y′ H (t) = ∑IMF1(t) (4-2).

[0071] Compared with the prior art, the present invention has the following beneficial technical effects:

[0072] (I) The present invention performs a second variational modal decomposition on the separated heartbeat signal, effectively suppressing the respiratory harmonic component, completing the accurate extraction of the heartbeat signal, and solving the problem of low accuracy in the extraction of heartbeat signals in the prior art.

[0073] (II) The present invention adopts millimeter wave radar detection technology. Millimeter wave radar has strong penetration, high sensitivity, high detection accuracy, high distance resolution, and good ranging capability. It can effectively filter out noise interference within other distances, making the method have good stability and improving the accuracy of extracting heartbeat signals, solving the technical problem of high stability of life signal extraction methods in the existing technology.

[0074] (III) The present invention performs variational modal decomposition on the extracted vital sign phase information to separate the respiratory signal and the heartbeat signal. Since the human heartbeat signal is extremely weak, the separated and extracted heartbeat signal contains a small amount of respiratory signal harmonic components, which greatly reduces the accuracy of the extracted heartbeat signal. Compared with other methods, the method of the present invention can accurately complete the extraction of the heartbeat signal while avoiding the situation of using a large number of lumped averages to extract accurate heartbeat signals, reducing the impact of the lumped average number on the algorithm efficiency, and solving the technical problem of low efficiency of the vital signal extraction method in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] Figure 1 is an overall flow chart of the method of the present invention;

[0076] Figure 2 is a flow chart of step three of the method of the present invention;

[0077] Figure 3 Schematic diagram of the structure of the millimeter wave radar life detection system of the present invention;

[0078] Figure 4 This is a one-dimensional distance pulse pressure result diagram in the embodiment;

[0079] Figure 5 This is a diagram showing the processing result of the moving target display technology in the embodiment;

[0080] Figure 6 This is a result diagram obtained after performing the autocorrelation algorithm in the embodiment;

[0081] FIG7( a ) shows the vital sign phase information extracted in the embodiment;

[0082] FIG7( b ) is a spectrum diagram of the vital sign phase information extracted in the embodiment;

[0083] FIG8( a ) shows the modal components obtained by decomposing the vital sign phase information by VMD in an embodiment;

[0084] FIG8( b ) is a spectrum diagram corresponding to the modal components obtained by decomposing the vital sign phase information by VMD in an embodiment;

[0085] Figure 9 is the energy percentage of each modal component in the embodiment;

[0086] FIG10( a ) is a reconstructed respiratory signal in an embodiment;

[0087] FIG10( b ) is a spectrum diagram of the reconstructed respiratory signal in the embodiment;

[0088] FIG11( a ) is a reconstructed heartbeat signal in an embodiment;

[0089] FIG11( b ) is a spectrum diagram of a reconstructed heartbeat signal in an embodiment;

[0090] FIG12( a ) shows the modal components obtained by performing VMD decomposition on the reconstructed heartbeat signal in an embodiment;

[0091] FIG12( b ) is a spectrum diagram corresponding to the modal components obtained by performing VMD decomposition on the reconstructed heartbeat signal in an embodiment;

[0092] Figure 13 is the energy percentage of each modal component in the embodiment;

[0093] FIG14( a ) is a heartbeat signal obtained by secondary VMD decomposition in an embodiment;

[0094] FIG14( b ) is a heartbeat signal spectrum diagram obtained by secondary VMD decomposition in an embodiment;

[0095] Figure 15(a) shows the heartbeat signal obtained by performing EEMD decomposition on the heartbeat signal obtained by VMD decomposition in the verification example;

[0096] Figure 15(b) is the heartbeat signal spectrum obtained by performing EEMD decomposition on the heartbeat signal obtained by VMD decomposition in the verification example.

[0097] Figure 16 This is a comparison chart of the signal-to-noise ratio of the heartbeat signals separated by the two algorithms in the verification example;

[0098] Figure 17 The following is a comparison chart of the running time of the two algorithms in the verification example.

[0099] The specific contents of the present invention are further explained in detail below with reference to the embodiments. DETAILED DESCRIPTION

[0100] It should be noted that the variational mode decomposition (VMD) algorithm is different from the ensemble empirical mode decomposition (EEMD) algorithm:

[0101] First, under normal circumstances, the human heart rate is approximately 0.8Hz to 1.5Hz, the respiratory rate is approximately 0.1Hz to 0.4Hz, and the harmonic frequency of the respiratory signal is roughly concentrated between 1Hz and 2Hz, which is exactly between the heartbeat frequency band. Therefore, under normal breathing, the heartbeat signal is affected by the harmonic components of the respiratory signal and the human heartbeat signal is extremely weak, so the heartbeat signal will be submerged by the harmonic components of the respiratory signal, making it difficult for the EEMD algorithm to separate the heartbeat signal; the variational mode decomposition (VMD) algorithm can separate the harmonic components and can detect and separate the heartbeat signal submerged in the harmonic components.

[0102] Secondly, the essence of the EEMD method is a multiple empirical mode decomposition (EMD) of superimposed Gaussian white noise. The extreme point characteristics of the signal are changed by adding different white noises of the same amplitude each time. The corresponding IMFs obtained from the multiple EMDs are then averaged to offset the added white noise. However, when the number of lumped averages is large, the algorithm is very time-consuming and inefficient. A limited number of lumped averages cannot completely eliminate the added white noise, resulting in large algorithm reconstruction errors and poor decomposition completeness. The variational mode decomposition (VMD) algorithm adopted in the present invention is a non-recursive variational mode signal decomposition method that decomposes the signal into a set of IMFs with a limited bandwidth, avoiding the redundant components of the EEMD algorithm.

[0103] Specific embodiments of the present invention are given below. It should be noted that the present invention is not limited to the following specific embodiments, and all equivalent modifications made on the basis of the technical solution of this application fall within the protection scope of the present invention.

[0104] Example:

[0105] This embodiment provides a method for extracting life signals from millimeter-wave radar based on VMD algorithm. Figures 1 to 3 As shown, the method specifically includes the following steps:

[0106] Step 1: Get the distance information D[P k ,M];

[0107] Step 2: Preprocess the distance information containing the target human breathing and heartbeat signals to obtain the vital sign phase information y r (t);

[0108] Step 3: The phase information y of the vital signs is decomposed by the variational mode decomposition algorithm. r (t) is decomposed to obtain the reconstructed life breathing signal y R (t) and heartbeat signal y H (t);

[0109] Step 4: The reconstructed heartbeat signal y is decomposed by variational mode decomposition algorithm. H (t) is decomposed to obtain the heartbeat signal y′ H (t).

[0110] In this embodiment, the radar is facing the front of the target person, illuminating the stomach and chest of the person. The target person is breathing normally and is about 1 meter away from the radar.

[0111] Table 1 The main parameters of the radar system are shown in the following table

[0112] <![CDATA[Signal carrier frequency f c > 77GHz Bandwidth B 1.7408GHz Pulse repetition period PRT 0.02s Total processing time t 50s Range resolution bin-r 0.0885m Total number of frames N 2500 <![CDATA[Sampling rate f s > 2.5MHz FM slope γ 17MHz / μs <![CDATA[Number of sampling points N for a single pulse soc > 256

[0113] In the above technical solution, the separated heartbeat signal undergoes a second variational modal decomposition, which effectively suppresses the respiratory harmonic component and completes the accurate extraction of the heartbeat signal, solving the problem of low accuracy in heartbeat signal extraction in the existing technology.

[0114] Specifically, step 1 is to use a millimeter-wave radar life detection system to obtain distance information containing the target human body's breathing and heartbeat signals. The millimeter-wave radar life detection system includes a transmitting end and a receiving end; the transmitting end includes a voltage-controlled oscillator, a waveform generator, and a transmitting antenna connected together; the receiving end includes a receiving antenna, a mixer, a bandpass filter, an analog-to-digital converter, and a baseband signal processing module connected in sequence; the voltage-controlled oscillator is connected to the mixer; the method specifically includes the following sub-steps:

[0115] Step 1.1: The voltage-controlled oscillator receives the RF signal from the waveform generator and modulates it. Part of the modulated signal is transmitted by the transmitting antenna, and the other part enters the mixer and serves as the mixer's local oscillator signal.

[0116] Step 1.2: The signal transmitted by the transmitting antenna is reflected by the target part, and the receiving antenna receives the reflected signal and sends it to the mixer;

[0117] Step 1.3: The mixer mixes the signal sent by the receiving antenna and the local oscillator signal, and sends the mixed signal to the bandpass filter for processing to obtain an intermediate frequency signal;

[0118] Step 1.4: The intermediate frequency signal enters the analog-to-digital converter for A / D sampling to obtain the data matrix of the echo signal, which is sent to the baseband signal processing module;

[0119] Step 1.5: The baseband signal processing module performs distance-dimensional Fourier transform on the data matrix of the echo signal to obtain the distance information matrix D[P k ,M];

[0120] M represents the number of RF signals emitted by the waveform generator;

[0121] P k Indicates the sequence number of the distance unit, 1≤k≤256.

[0122] In this embodiment, a radar radio frequency signal is received and modulated by a voltage-controlled oscillator. A portion of the modulated signal is emitted by a transmitting antenna, and another portion of the signal enters a mixer and serves as the mixer's local oscillator signal. The signal emitted by the transmitting antenna is partially reflected when it encounters a target, and the receiving antenna receives the reflected signal and enters the mixer. The mixer performs mixing processing on the received signal and the local oscillator signal, and uses a bandpass filter to filter out unusable harmonic components in the mixer output to obtain an intermediate frequency signal. The intermediate frequency signal enters an analog-to-digital converter for A / D sampling to obtain a data matrix S[N,M] of the echo signal. The echo data matrix S[N,M] is subjected to distance-dimensional Fourier processing to obtain distance information D[P] containing the target human respiratory and heartbeat signals. k ,M], as attached Figure 4 As shown in the figure, it can be seen that for objects at a distance of 4-6m, the thermal map color is uniform within the time range of 0 to 50s and hardly changes with time, indicating that the distance between the objects and the radar within these distances remains unchanged and they can be regarded as stationary objects; while the thermal map color within the distance of 1-4m changes with time, and there is target human life information and some noise within this range.

[0123] Specifically, the preprocessing in step 2 specifically includes the following steps:

[0124] Step 2.1: The distance information D[P k ,M] is processed by the moving target display technology shown in formula 2-1 to obtain the distance information matrix D′[P k ,M];

[0125]

[0126] in:

[0127] M represents the number of RF signals emitted by the waveform generator;

[0128] P k Indicates the sequence number of the distance unit, 1≤k≤256;

[0129] D[P k ,M] represents the distance information matrix containing the target human breathing and heartbeat signals;

[0130] ∑D[P k ,:] represents the matrix D[P k ,M] k The sum of all data within a distance unit;

[0131] Step 2.2, the distance information matrix D′[P k,M] perform the autocorrelation processing shown in formula 2-2 to obtain the data matrix S′[P k ,N];

[0132] S′[P k ,N]=xcorr(D′[P k ,M]) (2-2)

[0133] in:

[0134] xcorr(·) is the autocorrelation function;

[0135] N = 2M-1, which represents the number of RF signals after autocorrelation processing;

[0136] Step 2.3: According to formulas (2-3) and (2-4), the vital sign phase information y is extracted. r (t);

[0137] [a,P a ]=max(max(S′[P k ,N] T )) (2-3)

[0138] y r (t) = phase (S′ [P a ,:]) (2-4)

[0139] in:

[0140] S′[P k ,N] T Denote the matrix S′[P k ,N];

[0141] max(S′[P k ,N] T ) represents the matrix S′[P k ,N] T The maximum value of each column forms a row vector;

[0142] max(max(S′[P k ,N])) means getting the maximum value a in this row vector and the distance unit P where the maximum value is located a ;

[0143] S′[P a ,:] represents the matrix S′[P k ,N] a All data within a distance unit;

[0144] phase(·) represents the phase function.

[0145] In this embodiment, the distance information D[P k ,M] is processed by the moving target indication (MTI) technology to obtain the distance information matrix D′[P k ,M], the results are as follows Figure 5 As shown, for the distance information matrix D′[P k ,M] to perform autocorrelation processing and obtain the data matrix S′[P k ,N], such as Figure 6 As shown in the figure, after the moving target indication (MTI) technology processing and autocorrelation algorithm preprocessing, it can be clearly seen that the target human body is located 1m away from the radar. Due to the presence of the target human body's breathing and heartbeat, the chest surface produces micro-movements, so the chest cavity has different reflection capabilities for the radar signal, resulting in an amplitude value that changes with time; according to formula (2-3), the vital sign phase information y is extracted r (t), as shown in Figure 7(a), and its spectrum is shown in Figure 7(b). It can be seen from the spectrum that the frequency of the extracted human life breathing signal is about 0.2301Hz, and the frequency of the heartbeat signal is about 1.411Hz. During the experiment, the vital signs of the subject were measured by a heart rate monitor, and the breathing frequency was 0.26Hz and the heart rate was 1.40Hz. This data is basically consistent with the extracted vital sign signal data.

[0146] Specifically, step three includes the following steps:

[0147] Step 3.1: Use the variational mode decomposition algorithm to transform the vital sign phase information y r (t) decomposed into K1 modal components IMF;

[0148] Step 3.2: Perform fast Fourier transform on each modal component IMF to obtain the spectrum corresponding to each modal component IMF. Calculate the total energy E() of each modal component IMF and the energy E of each modal component IMF in the heartbeat frequency band on the spectrum. H (i) and the energy E in the respiratory frequency band R (i);

[0149] Step 3.3: Determine whether the following equation is true. If equation (3-1) is true, it is considered that this modal component IMF contains the heartbeat signal component and can be used to reconstruct the heartbeat signal. If not, it is considered that this modal component IMF does not contain the spectral characteristics of the human heartbeat signal.

[0150]

[0151] in:

[0152] i represents the serial number of the modal component, i=1, 2, ... K1;

[0153] δ represents the threshold value, and the value range is δ not less than 0.5;

[0154] Step 3.4, determine whether the following formula is true: Formula (3-2), then it is considered that this modal component IMF contains respiratory signal components and can be used to reconstruct the respiratory signal. If it is not satisfied, it is considered that this modal component IMF does not contain the spectral characteristics of the human respiratory signal;

[0155]

[0156] in:

[0157] i represents the decomposed modal component, i = 1, 2, ... K1;

[0158] δ represents the threshold value, and the value range is δ not less than 0.5;

[0159] Step 3.5: Add all the modal components IMF that satisfy formula (3-1) according to formula (3-3) to obtain the reconstructed heartbeat signal y H (t); add all the modal components IMF that satisfy formula (3-2) according to formula (3-4) to obtain the reconstructed respiratory signal y R (t);

[0160] y H (t)=∑IMF(t) (3-3)

[0161] y R (t)=∑IMF(t) (3-4)

[0162] t represents time, ranging from 0 to 50s;

[0163] In this embodiment, the variational mode decomposition algorithm is used to transform the input signal y r (t) is decomposed into 15 modal components IMF, as shown in Figure 8(a), and its spectrum is shown in Figure 8(b); each modal component IMF is subjected to fast Fourier transform, and the total energy E(j) of each modal component IMF is calculated in the frequency domain, and the energy of each modal component IMF in the heartbeat frequency band is E H (j) and the energy in the respiratory frequency band is E R (j), such as Figure 9 As shown, it is determined whether the modal component IMF contains the heartbeat signal component, and the modal component IMF containing the heartbeat signal component is added to obtain the reconstructed heartbeat signal y H(t), as shown in Figure 10(a), and its spectrum is shown in Figure 10(b); determine whether the modal component IMF contains a respiratory signal component, add the modal component IMF containing the respiratory signal component, and obtain the reconstructed respiratory signal y R (t), as shown in Figure 11(a), and its spectrum is shown in Figure 11(b). Figure 10(b) and 11(b) It can be seen that the decomposed and reconstructed respiratory signal has almost no clutter interference, while the decomposed and reconstructed heartbeat signal still has some respiratory signal harmonic components.

[0164] Specifically, step four includes the following steps:

[0165] Step 4.1, reconstruct the heartbeat signal y H (t) Perform variational mode decomposition algorithm to obtain K2 modal components IMF1;

[0166] Step 4.2: Perform fast Fourier transform on each modal component IMF1 to obtain the spectrum corresponding to each modal component IMF1. Calculate the total energy E1(j) of each modal component IMF1 on the spectrum. The energy of each modal component IMF1 in the heartbeat frequency band is

[0167] Step 4.3: Determine whether the following equation is true. If so, it is considered that the modal component IMF1 contains a heartbeat signal component and can be used to reconstruct the heartbeat signal. If not, it is considered that the modal component IMF1 does not contain the spectral characteristics of a human heartbeat signal.

[0168]

[0169] in:

[0170] j represents the serial number of the modal component, j = 1, 2, ... K2;

[0171] δ represents the threshold value, and the value range is δ not less than 0.5;

[0172] Step 4.4, add the modal components IMF1 that satisfy formula (4-1) according to formula (4-2) to obtain the reconstructed heartbeat signal y′ H (t):

[0173] y′ H (t) = ∑IMF1(t) (4-2).

[0174] In this embodiment, the reconstructed heartbeat signal y H(t) is processed by variational modal decomposition algorithm to obtain four modal components IMF1, as shown in Figure 12(a). Its spectrum is shown in Figure 12(b). Each modal component IMF1 is subjected to fast Fourier transform, and the total energy E1(j) of each modal component IMF1 is calculated in the frequency domain. The energy of each modal component IMF1 in the heartbeat frequency band is like Figure 13 As shown; it is determined that this modal component IMF1 contains a heartbeat signal component, and the modal component IMF1 containing the heartbeat signal component is added to obtain the reconstructed heartbeat signal y′ H (t), as shown in Figure 14(a), and its spectrum is shown in Figure 14(b). It can be seen from its spectrum that the harmonic components are successfully filtered out after further decomposition, and a pure heartbeat signal is obtained.

[0175] Verification example:

[0176] In order to verify the technical solution of the present invention, the above technical solution was adopted and the following experiment was set up:

[0177] In this experiment, two groups of experiments were set up under the condition of normal breathing of the subjects. Two processing methods were used to process the extracted life signals respectively to study the processing effect of the VMD algorithm and EEMD algorithm on the respiratory harmonic components and the extraction effect of the heartbeat signal. Other experimental equipment parameters and environment were consistent.

[0178] The first group performs VMD algorithm processing on the vital sign phase information extracted after preprocessing, and then performs a second VMD processing on the decomposed heartbeat signal, which is the processing algorithm used in this invention.

[0179] The second group performs VMD algorithm on the vital sign phase information extracted after preprocessing, and then performs EEMD processing on the decomposed heartbeat signal.

[0180] Experiment 1 (VMD+VMD Processing): The device acquires radar echo data and extracts vital sign phase information from the echo data, as shown in Figures 7(a) and 7(b). This data is then decomposed twice using the VMD algorithm, as shown in Figures 14(a) and 14(b). This demonstrates that the two VMD algorithm decompositions can substantially filter out the harmonic components of the respiratory signal, resulting in a purer heartbeat signal.

[0181] Experiment 2 (VMD + EEMD Processing): The VMD processing is identical to that in Experiment 1. The device acquires radar echo data, extracts vital sign phase information from the echo data, and decomposes it using the VMD algorithm to obtain a heartbeat signal containing harmonic components, as shown in Figures 11(a) and 11(b). This heartbeat signal is then decomposed using the EEMD algorithm and reconstructed using threshold conditions, as shown in Figures 15(a) and 15(b). The figures show that the EEMD algorithm does not completely separate the harmonic components of the respiratory signal.

[0182] Calculate the signal-to-noise ratio and error rate of the data obtained from the above two experiments as evaluation indicators, as shown in the attached figure. Figure 16 The figure shows the signal-to-noise ratio comparison of the heartbeat signals separated by the two algorithms. The horizontal axis represents the ensemble mean, and the vertical axis represents the signal-to-noise ratio. The bars in the figure represent the changes in the signal-to-noise ratio under different ensemble mean values. Figure 16 As can be seen from the table, as the ensemble mean increases, the signal-to-noise ratio of the method of the present application is significantly higher than that of the VMD+EEMD method. The higher the signal-to-noise ratio, the more accurate the isolated heartbeat signal. Furthermore, as shown in Table 2, the error rates of the heartbeat signal frequencies isolated by the two algorithms are 0.79% in Experiment 1 and 3.5% in Experiment 2. This indicates that Experiment 1 outperforms Experiment 2 in terms of heartbeat signal extraction accuracy. Therefore, it can be seen that the present application improves the accuracy of heartbeat signal extraction.

[0183] Table 2 Error rates of heartbeat signal frequencies separated by two algorithms

[0184]

[0185] Calculate the algorithm running time of the above two experiments as evaluation indicators, such as Figure 17 The following is a comparison of the running time of the two algorithms. The horizontal axis represents the ensemble average, and the vertical axis represents the running time. The bars in the figure represent the changes in the running time under different ensemble averages. Figure 17 As can be seen from the figure, as the ensemble mean increases, the runtime of the algorithm in Experiment 2 also increases. However, since the algorithm in Experiment 1 is not affected by the ensemble mean, its runtime remains basically unchanged and is generally much shorter than that of the algorithm in Experiment 2. Therefore, it can be seen that this application improves the accuracy of heartbeat signal extraction while also improving efficiency.

Claims

1. A method for extracting life signals from millimeter-wave radar based on VMD algorithm, characterized in that: The method specifically comprises the following steps: Step 1: Get the distance information D[P k ,M]; Step 2: Preprocess the distance information containing the target human breathing and heartbeat signals to obtain the vital sign phase information y r (t); Step 3: The phase information y of the vital signs is decomposed by the variational mode decomposition algorithm. r (t) is decomposed to obtain the reconstructed life breathing signal y R (t) and heartbeat signal y H (t); Step 4: The reconstructed heartbeat signal y is decomposed by variational mode decomposition algorithm. H (t) is decomposed to obtain the heartbeat signal y′ H (t); Step 4.1, reconstruct the heartbeat signal y H (t) Perform variational modal decomposition algorithm to obtain K2 modal components IMF1; Step 4.2: Perform fast Fourier transform on each modal component IMF1 to obtain the spectrum corresponding to each modal component IMF1. Calculate the total energy E1(j) of each modal component IMF1 on the spectrum. The energy of each modal component IMF1 in the heartbeat frequency band is Step 4.3: Determine whether the following equation is true. If so, it is considered that the modal component IMF1 contains a heartbeat signal component and can be used to reconstruct the heartbeat signal. If not, it is considered that the modal component IMF1 does not contain the spectral characteristics of a human heartbeat signal. in: j represents the serial number of the modal component, j = 1, 2, ... K2; δ represents the threshold value, and the value range is δ not less than 0.5; Step 4.4, add the modal components IMF1 that satisfy formula (4-1) according to formula (4-2) to obtain the reconstructed heartbeat signal y′ H (t): y′ H (t)=∑IMF1(t) (4-2)。 2. The method for extracting life signals from millimeter-wave radar based on VMD algorithm according to claim 1, wherein: The first step is to use a millimeter-wave radar life detection system to obtain distance information containing the target human body's breathing and heartbeat signals. The millimeter-wave radar life detection system includes a transmitting end and a receiving end; the transmitting end includes a voltage-controlled oscillator, a waveform generator, and a transmitting antenna connected together; the receiving end includes a receiving antenna, a mixer, a bandpass filter, an analog-to-digital converter, and a baseband signal processing module connected in sequence; the voltage-controlled oscillator is connected to the mixer; the method specifically includes the following sub-steps: Step 1.1: The voltage-controlled oscillator receives the RF signal from the waveform generator and modulates it. Part of the modulated signal is transmitted by the transmitting antenna, and the other part enters the mixer and serves as the mixer's local oscillator signal. Step 1.2: The signal transmitted by the transmitting antenna is reflected by the target part, and the receiving antenna receives the reflected signal and sends it to the mixer; Step 1.3: The mixer mixes the signal sent by the receiving antenna and the local oscillator signal, and sends the mixed signal to the bandpass filter for processing to obtain an intermediate frequency signal; Step 1.4: The intermediate frequency signal enters the analog-to-digital converter for A / D sampling to obtain the data matrix of the echo signal, which is sent to the baseband signal processing module; Step 1.5: The baseband signal processing module performs distance-dimensional Fourier transform on the data matrix of the echo signal to obtain the distance information matrix D[P k ,M]; in: M represents the number of RF signals emitted by the waveform generator; P k Indicates the sequence number of the distance unit, 1≤k≤256.

3. The method for extracting life signals from millimeter-wave radar based on VMD algorithm according to claim 1, wherein: The pretreatment in step 2 specifically includes the following steps: Step 2.1: The distance information D[P k ,M] is processed by the moving target display technology shown in formula 2-1 to obtain the distance information matrix D′[P k ,M]; in: M represents the number of RF signals emitted by the waveform generator; P k Indicates the sequence number of the distance unit, 1≤k≤256; D[P k ,M] represents the distance information matrix containing the target human breathing and heartbeat signals; ∑D[P k ,:] represents the matrix D[P k ,M] k The sum of all data within a distance unit; Step 2.2, the distance information matrix D′[P k ,M] perform the autocorrelation processing shown in formula 2-2 to obtain the data matrix S′[P k ,N]; S′[P k ,N]=xcorr(D′[P k ,M]) (2-2) in: xcorr(·) is the autocorrelation function; N = 2M-1, which represents the number of RF signals after autocorrelation processing; Step 2.3: According to formulas (2-3) and (2-4), the vital sign phase information y is extracted. r (t); [a,P a ]=max(max(S′[P k ,N] T )) (2-3) y r (t)=phase(S′[P a ,:]) (2-4) in: S′[P k ,N] T Denote the matrix S′[P k ,N]; max(S′[P k ,N] T ) represents the matrix S′[P k ,N] T The maximum value of each column forms a row vector; max(max(S′[P k ,N])) means getting the maximum value a in this row vector and the distance unit P where the maximum value is located a ; S′[P a ,:] represents the matrix S′[P k ,N] a All data within a distance unit; phase(·) represents the phase function.

4. The method for extracting life signals from millimeter-wave radar based on VMD algorithm according to claim 1, wherein: The step three specifically includes the following steps: Step 3.1: Use the variational mode decomposition algorithm to transform the vital sign phase information y r (t) decomposed into K1 modal components IMF; Step 3.2: Perform fast Fourier transform on each modal component IMF to obtain the spectrum corresponding to each modal component IMF. Calculate the total energy E(i) of each modal component IMF and the energy E(i) of each modal component IMF in the heartbeat frequency band on the spectrum. H (i) and the energy E in the respiratory frequency band R (i); Step 3.3: Determine whether the following equation is true. If equation (3-1) is true, it is considered that this modal component IMF contains the heartbeat signal component and can be used to reconstruct the heartbeat signal. If not, it is considered that this modal component IMF does not contain the spectral characteristics of the human heartbeat signal. in: i represents the serial number of the modal component, i=1, 2, ... K1; δ represents the threshold value, and the value range is δ not less than 0.5; Step 3.4, determine whether the following formula is true: Formula (3-2), then it is considered that this modal component IMF contains respiratory signal components and can be used to reconstruct the respiratory signal. If it is not satisfied, it is considered that this modal component IMF does not contain the spectral characteristics of the human respiratory signal; in: i represents the decomposed modal component, i = 1, 2, ... K1; δ represents the threshold value, and the value range is δ not less than 0.5; Step 3.5: Add all the modal components IMF that satisfy formula (3-1) according to formula (3-3) to obtain the reconstructed heartbeat signal y H (t); add all the modal components IMF that satisfy formula (3-2) according to formula (3-4) to obtain the reconstructed respiratory signal y R (t); y H (t)=∑IMF(t) (3-3) y R (t)=∑IMF(t) (3-4) t represents time, ranging from 0 to 50s.

Images