Millimeter wave radar breathing monitoring method and system based on periodic constraints

Abstract

The present invention discloses a millimeter wave radar respiratory monitoring method and system based on periodic constraints, including: obtaining a signal matrix collected by millimeter wave radar respiratory monitoring; decomposing the signal matrix into a low-rank matrix and a sparse matrix, applying Fourier spectrum constraints on the low-rank matrix, and establishing an optimization objective function; iteratively optimizing the objective function, introducing the periodic characteristics of the respiratory signal into the optimization process, obtaining a low-rank matrix and a sparse matrix, and further obtaining a respiratory signal. By introducing a periodic constraint model, the fitting accuracy of the millimeter wave radar to the normal respiratory signal is improved, so that sudden events such as apnea can be more effectively detected.

Description

Technical Field

[0001] The present invention belongs to the technical field of respiratory monitoring, and relates to a millimeter wave radar respiratory monitoring method and system based on periodic constraints. Background Art

[0002] With the continuous development of health monitoring technology, millimeter-wave radar has been widely used in human physiological signal monitoring due to its non-contact, strong penetration, and strong anti-interference ability. Existing millimeter-wave radar respiratory monitoring systems mainly rely on low-rank sparse decomposition models to separate normal respiratory signals from sudden events (such as apnea). However, due to the individual differences in respiratory signals and the influence of environmental noise, existing methods often have misjudgment problems when detecting apnea. In the existing technology, the periodic characteristics of normal respiratory signals have not been fully utilized, resulting in the model's weak ability to distinguish sudden abnormal events.

[0003] Therefore, there is an urgent need for an improved method that introduces periodic constraints on respiratory signals based on the existing model to improve the accuracy of respiratory signals and the ability to detect abnormal events. Summary of the invention

[0004] The purpose of the present invention is to provide a millimeter-wave radar respiratory monitoring method and system based on periodic constraints, by introducing a periodic constraint model, to improve the fitting accuracy of the millimeter-wave radar to the normal respiratory signal, so as to more effectively detect sudden events such as apnea.

[0005] The technical solution to achieve the purpose of the present invention is:

[0006] A millimeter wave radar breathing monitoring method based on periodic constraints comprises the following steps:

[0007] S01: Obtain the signal matrix collected by millimeter wave radar breathing monitoring;

[0008] S02: Decompose the signal matrix into a low-rank matrix and a sparse matrix, impose Fourier spectrum constraints on the low-rank matrix, and establish an optimization objective function;

[0009] S03: Iteratively optimize the objective function, introduce the periodic characteristics of the respiratory signal into the optimization process, obtain a low-rank matrix and a sparse matrix, and further obtain the respiratory signal.

[0010] In the preferred technical solution, the method for obtaining the signal matrix collected by the millimeter wave radar breathing monitoring in step S01 includes:

[0011] Millimeter wave radar The transmitted signal of a pulse is:

[0012]

[0013] in, is the amplitude of the transmitted signal, is the initial frequency, is the frequency modulation bandwidth, is the modulation period;

[0014] After being reflected by the human body surface, the echo pulse signal received by the millimeter wave radar is expressed as:

[0015]

[0016] in, is the time delay of the transmitted signal reflected by the human body, and , is the radial distance between the radar system and the human body, is the speed of light;

[0017] When the millimeter-wave radar is used for respiratory monitoring, the chest movement caused by breathing will cause the phase of the radar receiving signal to change. The relationship between the displacement of the human chest cavity is:

[0018]

[0019] in, is the displacement of the chest cavity over time, is the wavelength of the millimeter-wave radar, and the millimeter-wave radar receiving signal model is expanded to:

[0020]

[0021] Receiving Signals The sampling signal is used Indicates that its sampling length is , then the millimeter wave radar system receives a total of The received pulse sampling signals can form a sampling echo signal matrix as follows:

[0022] .

[0023] In the preferred technical solution, the method for establishing the optimization objective function in step S02 includes:

[0024] Decompose the sampled echo signal matrix into:

[0025]

[0026] in, is a low-rank matrix, is a sparse matrix, is the noise matrix;

[0027] In a low-rank matrix The Fourier spectrum constraint is imposed on the improved optimization objective function:

[0028]

[0029] in, is the kernel function of the matrix, It is a matrix norm, is the Frobenius norm of the matrix, and is a constraint parameter, Represents a low-rank matrix The Fourier transform of It is a predefined reference spectrum of the respiratory frequency, which is used to describe the periodic characteristics of normal respiratory signals.

[0030] In a preferred technical solution, the method for obtaining a reference spectrum includes:

[0031] S11: Long-term acquisition of respiratory signals of multiple healthy individuals in a quiet state using millimeter-wave radar;

[0032] S12: filtering the collected respiratory signal to remove high-frequency noise and other interfering signals and retain components related to the respiratory frequency;

[0033] S13: Perform Fourier transform on the preprocessed signal to convert the time domain signal into a frequency domain signal;

[0034] S14: Statistically analyzing the respiratory signal spectra of all individuals to calculate the mean value, standard deviation and frequency distribution range of the respiratory frequency;

[0035] S15: Model the spectral characteristics of the normal breathing signal obtained by statistics, use probability distribution to describe the concentrated area of ​​the breathing frequency, and generate a reference spectrum model .

[0036] In the preferred technical solution, the method for iteratively optimizing the objective function in step S03 includes:

[0037] S21: Introducing auxiliary variables , rewrite the optimization objective function as:

[0038]

[0039] S22: Construct Lagrangian function:

[0040]

[0041] in, , , is the Lagrange multiplier, is the penalty factor;

[0042] S23: Update :

[0043]

[0044] Where k is the number of iterations;

[0045] S24: Update :

[0046]

[0047] S25: Update ;

[0048]

[0049] This step uses Fourier transform to introduce periodic signal information into the optimization process so that It is consistent with the periodic characteristics of the respiratory signal;

[0050] S26: Update :

[0051]

[0052] S27: Update :

[0053]

[0054] S28: Update Lagrange multipliers:

[0055]

[0056] S29: When the change amplitude of all variables is less than the set threshold or reaches the maximum number of iterations, stop the iteration and output the results.

[0057] In the preferred technical solution, step S03 further includes:

[0058] The presence of apnea events is determined by analyzing the non-zero elements in the sparse matrix.

[0059] In the preferred technical solution, the method for determining apnea events includes:

[0060] S31: Setting a sliding time window length;

[0061] S32: Calculate the sparse matrix within the sliding time window The average energy of the signal or the number of non-zero elements in the . If the value is lower than the set threshold, it is determined as an apnea event. The formula is as follows:

[0062]

[0063] in, is a sparse matrix No. Line Column elements, , are the number of rows and columns respectively;

[0064] S33: If multiple consecutive time windows meet the apnea condition, it is further confirmed to be an apnea event.

[0065] The present invention also discloses a millimeter wave radar breathing monitoring system based on periodic constraints, comprising:

[0066] The acquisition module obtains the signal matrix collected by the millimeter wave radar breathing monitoring;

[0067] Optimize the objective function building module, decompose the signal matrix into a low-rank matrix and a sparse matrix, impose Fourier spectrum constraints on the low-rank matrix, and establish the optimization objective function;

[0068] The iterative optimization module iteratively optimizes the objective function, introduces the periodic characteristics of the respiratory signal into the optimization process, obtains a low-rank matrix and a sparse matrix, and further obtains the respiratory signal.

[0069] In the preferred technical solution, the method for establishing the optimization objective function in the optimization objective function construction module includes:

[0070] Decompose the sampled echo signal matrix into:

[0071]

[0072] in, is a low-rank matrix, is a sparse matrix, is the noise matrix;

[0073] In a low-rank matrix The Fourier spectrum constraint is imposed on the improved optimization objective function:

[0074]

[0075] in, is the kernel function of the matrix, It is a matrix norm, is the Frobenius norm of the matrix, and is a constraint parameter, Represents a low-rank matrix The Fourier transform of It is a predefined reference spectrum of the respiratory frequency, which is used to describe the periodic characteristics of normal respiratory signals.

[0076] The present invention further discloses a computer storage medium on which a computer program is stored. When the computer program is executed, the above-mentioned millimeter wave radar breathing monitoring method based on periodic constraints is implemented.

[0077] Compared with the prior art, the present invention has the following significant advantages:

[0078] 1. Improve the accuracy of respiratory signal recognition: By introducing periodic constraints on respiratory signals, normal respiratory signals are made more stable in the model, thereby improving the accuracy of recognition of abnormal events such as apnea, which is suitable for non-contact physiological monitoring in various scenarios.

[0079] 2. Enhance model robustness: In the presence of environmental noise, periodic constraints can enhance the signal model's ability to fit normal breathing signals and reduce false positives and negatives.

[0080] 3. Adapt to individual differences: Periodic constraints can be adaptively adjusted according to individual breathing characteristics to improve the universality of the model among different populations. BRIEF DESCRIPTION OF THE DRAWINGS

[0081] Figure 1 is a flow chart of a millimeter wave radar breathing monitoring method based on periodic constraints of the present invention;

[0082] Figure 2 It is a principle block diagram of the millimeter wave radar breathing monitoring system based on periodic constraints of the present invention. DETAILED DESCRIPTION

[0083] The principle of the present invention is to improve the fitting accuracy of millimeter wave radar to normal breathing signals by introducing a periodic constraint model, so as to more effectively detect sudden events such as apnea.

[0084] Embodiment 1:

[0085] like Figure 1 As shown, a millimeter wave radar breathing monitoring method based on periodic constraints includes the following steps:

[0086] S01: Obtain the signal matrix collected by millimeter wave radar breathing monitoring;

[0087] S02: Decompose the signal matrix into a low-rank matrix and a sparse matrix, impose Fourier spectrum constraints on the low-rank matrix, and establish an optimization objective function;

[0088] S03: Iteratively optimize the objective function, introduce the periodic characteristics of the respiratory signal into the optimization process, obtain a low-rank matrix and a sparse matrix, and further obtain the respiratory signal.

[0089] In a preferred embodiment, the method of obtaining the signal matrix collected by millimeter wave radar breathing monitoring in step S01 includes:

[0090] Millimeter wave radar The transmitted signal of a pulse is:

[0091]

[0092] in, is the amplitude of the transmitted signal, is the initial frequency, is the frequency modulation bandwidth, is the modulation period;

[0093] After being reflected by the human body surface, the echo pulse signal received by the millimeter wave radar is expressed as:

[0094]

[0095] in, is the time delay of the transmitted signal reflected by the human body, and , is the radial distance between the radar system and the human body, is the speed of light;

[0096] When the millimeter-wave radar is used for respiratory monitoring, the chest movement caused by breathing will cause the phase of the radar receiving signal to change. The relationship between the displacement of the human chest cavity is:

[0097]

[0098] in, is the displacement of the chest cavity over time, is the wavelength of the millimeter-wave radar, and the millimeter-wave radar receiving signal model is expanded to:

[0099]

[0100] Receiving Signals The sampling signal is used Indicates that its sampling length is , then the millimeter wave radar system receives a total of The received pulse sampling signals can form a sampling echo signal matrix as follows:

[0101] .

[0102] In a preferred embodiment, the method of establishing the optimization objective function in step S02 includes:

[0103] Decompose the sampled echo signal matrix into:

[0104]

[0105] in, is a low-rank matrix, is a sparse matrix, is the noise matrix;

[0106] In a low-rank matrix The Fourier spectrum constraint is imposed on the improved optimization objective function:

[0107]

[0108] in, is the kernel function of the matrix, It is a matrix norm, is the Frobenius norm of the matrix, and is a constraint parameter, Represents a low-rank matrix The Fourier transform of It is a predefined reference spectrum of the respiratory frequency, which is used to describe the periodic characteristics of normal respiratory signals.

[0109] In a preferred embodiment, the method for obtaining the reference spectrum includes:

[0110] S11: Long-term acquisition of respiratory signals of multiple healthy individuals in a quiet state using millimeter-wave radar;

[0111] S12: filtering the collected respiratory signal to remove high-frequency noise and other interfering signals and retain components related to the respiratory frequency;

[0112] S13: Perform Fourier transform on the preprocessed signal to convert the time domain signal into a frequency domain signal;

[0113] S14: Statistically analyzing the respiratory signal spectra of all individuals to calculate the mean value, standard deviation and frequency distribution range of the respiratory frequency;

[0114] S15: Model the spectral characteristics of the normal breathing signal obtained by statistics, use probability distribution to describe the concentrated area of ​​the breathing frequency, and generate a reference spectrum model .

[0115] In a preferred embodiment, the method for iteratively optimizing the objective function in step S03 includes:

[0116] S21: Introducing auxiliary variables , rewrite the optimization objective function as:

[0117]

[0118] S22: Construct Lagrangian function:

[0119]

[0120] in, , , is the Lagrange multiplier, is the penalty factor;

[0121] S23: Update :

[0122]

[0123] Where k is the number of iterations;

[0124] S24: Update :

[0125]

[0126] S25: Update ;

[0127]

[0128] This step uses Fourier transform to introduce periodic signal information into the optimization process so that It is consistent with the periodic characteristics of the respiratory signal;

[0129] S26: Update :

[0130]

[0131] S27: Update :

[0132]

[0133] S28: Update Lagrange multipliers:

[0134]

[0135] S29: When the change amplitude of all variables is less than the set threshold or reaches the maximum number of iterations, stop the iteration and output the results.

[0136] In a preferred embodiment, step S03 further includes:

[0137] The presence of apnea events is determined by analyzing the non-zero elements in the sparse matrix.

[0138] In a preferred embodiment, the method for determining a apnea event includes:

[0139] S31: Setting a sliding time window length;

[0140] S32: Calculate the sparse matrix within the sliding time window The average energy of the signal or the number of non-zero elements in the . If the value is lower than the set threshold, it is determined as an apnea event. The formula is as follows:

[0141]

[0142] in, is a sparse matrix No. Line Column elements, , are the number of rows and columns respectively;

[0143] S33: If multiple consecutive time windows meet the apnea condition, it is further confirmed to be an apnea event.

[0144] In another embodiment, a computer storage medium stores a computer program, which implements the above-mentioned millimeter wave radar breathing monitoring method based on periodic constraints when executed.

[0145] The millimeter wave radar breathing monitoring method based on periodic constraints can adopt any of the above-mentioned millimeter wave radar breathing monitoring methods based on periodic constraints, and the specific implementation will not be repeated here.

[0146] In another embodiment, if Figure 2 As shown, a millimeter wave radar breathing monitoring system based on periodic constraints includes:

[0147] The acquisition module 10 acquires the signal matrix collected by the millimeter wave radar breathing monitoring;

[0148] An optimization objective function building module 20 decomposes the signal matrix into a low-rank matrix and a sparse matrix, imposes Fourier spectrum constraints on the low-rank matrix, and establishes an optimization objective function;

[0149] The iterative optimization module 30 iteratively optimizes the objective function, introduces the periodic characteristics of the respiratory signal into the optimization process, obtains a low-rank matrix and a sparse matrix, and further obtains the respiratory signal.

[0150] Specifically, the workflow of the millimeter wave radar breathing monitoring system based on periodic constraints is described below by taking a preferred embodiment as an example:

[0151] In continuous pulse mode, millimeter wave radar detects the movement of the target by continuously transmitting and receiving a series of modulated continuous wave (CW) signals or frequency modulated continuous wave (FMCW) signals. The core idea is to identify the tiny displacements caused by human breathing by analyzing the phase or frequency changes of the return signal. FMCW radar transmits a linear frequency modulation signal during the detection process, and its frequency increases linearly with time. The frequency offset (i.e., Doppler shift) of the return signal is related to the distance and speed of the target and can be calculated by spectrum analysis. For respiratory monitoring, millimeter wave radar uses the phase changes of the reflected signal to sense tiny periodic movements of the human surface, such as the expansion and contraction of the chest cavity. This tiny movement causes a slight periodic change in the phase of the radar signal.

[0152] Millimeter wave radar The transmitted signal of a pulse is:

[0153] (1)

[0154] in is the amplitude of the transmitted signal, is the initial frequency, is the frequency modulation bandwidth, is the modulation period.

[0155] After being reflected by the human body surface, the echo pulse signal received by the radar can be expressed as:

[0156] (2)

[0157] in is the time delay of the radio signal reflected by the human body, and , is the radial distance between the radar system and the human body, The speed of light.

[0158] When the millimeter-wave radar system performs respiratory monitoring, the chest movement caused by breathing will cause the phase of the radar receiving signal to change. Phase change The relationship between the displacement of the human chest cavity is:

[0159] (3)

[0160] in is the displacement of the chest cavity over time, is the wavelength of the millimeter wave radar. Therefore, the radar receiving signal model of equation (2) will be expanded to:

[0161] (4)

[0162] make Indicates receiving signal The sampling length of the signal is , then the millimeter wave radar system receives a total of The received pulse sampling signals can form a sampling echo signal matrix

[0163] (5)

[0164] The present invention represents the signal matrix collected by the millimeter wave radar as two parts:

[0165] 1. Low-rank part: represents normal, periodic breathing signals.

[0166] 2. Sparse part: Indicates abnormal breathing behavior (such as apnea). These sudden events usually appear as sparse and irregular signals.

[0167] The above model can effectively separate normal behavior from abnormal behavior in the signal and detect apnea events more accurately. Specifically, the sampled echo signal matrix can be decomposed into the following model:

[0168] (6)

[0169] in is a low-rank matrix, representing a normal breathing signal. is a sparse matrix representing sudden apnea events. is the noise matrix, representing the modeled measurement errors and environmental noise.

[0170] In order to more accurately model the periodic characteristics of the respiratory signal, we use the low-rank matrix Fourier spectrum constraints are imposed on the normal breathing signal, so that the normal breathing signal is closer to the known breathing frequency characteristics in the frequency domain. The improved optimization problem is:

[0171] (7)

[0172] in It is the kernel function of the matrix, that is, the sum of the singular values ​​of the matrix, indicating the low rank of the matrix. It is a matrix norm, indicating the sparse nature of the matrix. is the Frobenius norm of the matrix, representing the energy of the noise. and is a constraint parameter. Represents a low-rank matrix The Fourier transform of . It is a predefined reference spectrum of the respiratory frequency, which is used to describe the periodic characteristics of normal respiratory signals.

[0173] The present invention obtains a reference spectrum The process can be summarized into the following steps:

[0174] Step 1: Data collection: Use millimeter wave radar to collect long-term respiratory signals of multiple healthy individuals in a quiet state. These signal data should include normal respiratory characteristics of different individuals under different conditions.

[0175] Step 2: Signal preprocessing: filter the collected respiratory signal to remove high-frequency noise and other interfering signals, and retain the main components related to the respiratory frequency.

[0176] Step 3: Fourier transform, perform Fourier transform on the preprocessed signal to convert the time domain signal into a frequency domain signal. In this way, the main frequency components of the respiratory signal in the frequency domain can be clearly observed.

[0177] Step 4: Statistical modeling, statistically analyze the respiratory signal spectra of all individuals, and calculate the mean, standard deviation and frequency distribution range of the respiratory frequency. Generally, the respiratory frequency range of healthy adults at rest is 0.1 to 0.5 Hz (i.e. 6 to 30 breaths / minute). By summarizing these spectrum data, a more universal normal respiratory reference spectrum can be obtained.

[0178] Step 5: Construct a reference spectrum model. Model the spectrum characteristics of the normal breathing signal obtained by statistics. Gaussian distribution or other suitable probability distributions can be used to describe the concentrated area of ​​the breathing frequency, and a reference spectrum model is generated as a reference in the Fourier constraint.

[0179] The present invention adopts the following steps to solve the joint optimization problem shown in formula (7):

[0180] Step 1: Introduce auxiliary variables , rewrite the optimization problem as:

[0181] (8)

[0182] Step 2: Construct the Lagrangian function

[0183] (9)

[0184] in , , is the Lagrange multiplier, is the penalty factor.

[0185] Step 3: Update

[0186] (10)

[0187] This update can be done by solving a linear equation, and combining periodic constraints helps to make More closely matches normal breathing signal patterns.

[0188] Step 4: Update

[0189] (11)

[0190] This step can be solved by the Singular Value Thresholding (SVT) algorithm to ensure that the matrix The low-rank property of .

[0191] Step 5: Update

[0192] (12)

[0193] This step uses Fourier transform to introduce periodic signal information into the optimization process so that It is consistent with the periodic characteristics of the respiratory signal and can be efficiently solved by fast Fourier transform (FFT) and inverse Fourier transform (IFFT).

[0194] Step 6: Update

[0195] (13)

[0196] This step can be solved by the soft thresholding algorithm to ensure that the matrix The sparsity of .

[0197] Step 7: Update

[0198] (14)

[0199] This step is used for denoising to ensure that the model is robust to environmental noise.

[0200] Step 8: Update the Lagrange multipliers

[0201] (15)

[0202] Step 9: When the change of all variables is less than the set threshold or the maximum number of iterations is reached, stop the iteration and output the results.

[0203] The improved algorithm adopted by the present invention combines the periodic characteristics of the respiratory signal to make the millimeter wave radar respiratory monitoring system more in line with the actual respiratory pattern in low-rank sparse decomposition, effectively improving the robustness and detection accuracy of the model. This method is particularly suitable for non-contact health monitoring applications that require high-precision respiratory detection.

[0204] Finally, the present invention determines whether there is apnea by analyzing the non-zero elements in the sparse matrix. The signal energy of the sparse matrix at each time point is calculated. Apnea events usually show lower signal energy or continuous sparse abnormalities at multiple time points. In order to accurately determine apnea events, the present invention uses the following judgment criteria:

[0205] 1. Time window setting: Set a sliding time window (such as 10 seconds to 30 seconds) to analyze the changes in the signal during this period. According to the definition of apnea (usually a breathing interruption of more than 10 seconds), you can set an appropriate window length.

[0206] 2. Energy threshold determination: Calculate the sparse matrix within the sliding time window The average energy of the signal or the number of non-zero elements in the . If the value is lower than the set threshold, it is determined as an apnea event. The formula is as follows:

[0207] (16)

[0208] in for Matrix Line Then set the apnea threshold ,like , it is determined as a respiratory arrest event.

[0209] 3. Duration determination: In the sparse matrix, if multiple consecutive time windows meet the apnea condition (such as energy below the threshold), it is further confirmed as an apnea event. A minimum duration is set as the criterion for judging apnea.

[0210] The present invention provides a method and system for introducing periodic constraints in millimeter wave radar respiratory monitoring, which can significantly improve the recognition accuracy of respiratory signals and the detection capability of emergencies, and is suitable for non-contact physiological monitoring in various scenarios.

[0211] The above embodiments are preferred implementation modes of the present invention, but the implementation modes of the present invention are not limited to the above embodiments. Any other changes, modifications, substitutions, combinations, and simplifications made without departing from the spirit and principles of the present invention shall be equivalent replacement modes and shall be included in the protection scope of the present invention.

Claims

1. A millimeter wave radar breathing monitoring method based on periodic constraints, characterized in that: The following steps are involved: S01: Obtain the signal matrix collected by millimeter wave radar breathing monitoring; S02: Decompose the signal matrix into a low-rank matrix and a sparse matrix, impose Fourier spectrum constraints on the low-rank matrix, and establish an optimization objective function; Methods for establishing optimization objective functions include: Decompose the sampled echo signal matrix into: ， in, is a low-rank matrix, is a sparse matrix, is the noise matrix; In a low-rank matrix The Fourier spectrum constraint is imposed on the improved optimization objective function: ， in, is the kernel function of the matrix, It is a matrix norm, is the Frobenius norm of the matrix, and is a constraint parameter, Represents a low-rank matrix The Fourier transform of It is a predefined reference spectrum of respiratory frequency, which is used to describe the periodic characteristics of normal respiratory signals; S03: Iteratively optimize the objective function, introduce the periodic characteristics of the respiratory signal into the optimization process, obtain a low-rank matrix and a sparse matrix, and further obtain a respiratory signal; the method for iteratively optimizing the objective function includes: S21: Introducing auxiliary variables , rewrite the optimization objective function as: ， S22: Construct Lagrangian function: ， in, , , is the Lagrange multiplier, is the penalty factor; S23: Update : ， Where k is the number of iterations; S24: Update : ， S25: Update ; ， This step uses Fourier transform to introduce periodic signal information into the optimization process so that It is consistent with the periodic characteristics of the respiratory signal; S26: Update : ， S27: Update : ， S28: Update Lagrange multipliers: ， S29: When the change amplitude of all variables is less than the set threshold or reaches the maximum number of iterations, stop the iteration and output the results.

2. The millimeter wave radar breathing monitoring method based on periodic constraints according to claim 1 is characterized in that: The method for obtaining the signal matrix collected by the millimeter wave radar breathing monitoring in step S01 includes: Millimeter wave radar The transmitted signal of a pulse is: ; in, is the amplitude of the transmitted signal, is the initial frequency, is the frequency modulation bandwidth, is the modulation period; After being reflected by the human body surface, the echo pulse signal received by the millimeter wave radar is expressed as: ， in, is the time delay of the transmitted signal reflected by the human body, and , is the radial distance between the radar system and the human body, is the speed of light; When the millimeter-wave radar is used for respiratory monitoring, the chest movement caused by breathing will cause the phase of the radar receiving signal to change. The relationship between the displacement of the human chest cavity is: ， in, is the displacement of the chest cavity over time, is the wavelength of the millimeter-wave radar, and the millimeter-wave radar receiving signal model is expanded to: ; Receiving Signals The sampling signal is used Indicates that its sampling length is , then the millimeter wave radar system receives a total of The received pulse sampling signals can form a sampling echo signal matrix as follows: 。 3. The millimeter wave radar breathing monitoring method based on periodic constraints according to claim 1 is characterized in that: Methods for obtaining reference spectrum include: S11: Long-term acquisition of respiratory signals of multiple healthy individuals in a quiet state using millimeter-wave radar; S12: filtering the collected respiratory signal to remove high-frequency noise and other interference signals and retain components related to the respiratory frequency; S13: Perform Fourier transform on the preprocessed signal to convert the time domain signal into a frequency domain signal; S14: Statistically analyzing the respiratory signal spectra of all individuals to calculate the mean value, standard deviation and frequency distribution range of the respiratory frequency; S15: Model the spectral characteristics of the normal breathing signal obtained by statistics, use probability distribution to describe the concentrated area of ​​the breathing frequency, and generate a reference spectrum model .

4. The millimeter wave radar breathing monitoring method based on periodic constraints according to claim 1 is characterized in that: Step S03 also includes: The presence of apnea events is determined by analyzing the non-zero elements in the sparse matrix.

5. The millimeter wave radar breathing monitoring method based on periodic constraints according to claim 4 is characterized in that: Methods for determining apnea events include: S31: Setting a sliding time window length; S32: Calculate the sparse matrix within the sliding time window The average energy of the signal or the number of non-zero elements in the . If the value is lower than the set threshold, it is determined as an apnea event. The formula is as follows: ， in, is a sparse matrix No. Line Column elements, , are the number of rows and columns respectively; S33: If multiple consecutive time windows meet the apnea condition, it is further confirmed to be an apnea event.

6. A millimeter wave radar breathing monitoring system based on periodic constraints, characterized in that: include: The acquisition module obtains the signal matrix collected by the millimeter wave radar breathing monitoring; Optimize the objective function building module, decompose the signal matrix into a low-rank matrix and a sparse matrix, impose Fourier spectrum constraints on the low-rank matrix, and establish the optimization objective function; Methods for establishing optimization objective functions include: Decompose the sampled echo signal matrix into: ， in, is a low-rank matrix, is a sparse matrix, is the noise matrix; In a low-rank matrix The Fourier spectrum constraint is imposed on the improved optimization objective function: ， in, is the kernel function of the matrix, It is a matrix norm, is the Frobenius norm of the matrix, and is a constraint parameter, Represents a low-rank matrix The Fourier transform of It is a predefined reference spectrum of respiratory frequency, which is used to describe the periodic characteristics of normal respiratory signals; The iterative optimization module iteratively optimizes the objective function, introduces the periodic characteristics of the respiratory signal into the optimization process, obtains a low-rank matrix and a sparse matrix, and further obtains the respiratory signal; the method for iteratively optimizing the objective function includes: S21: Introducing auxiliary variables , rewrite the optimization objective function as: ， S22: Construct Lagrangian function: ， in, , , is the Lagrange multiplier, is the penalty factor; S23: Update : ， Where k is the number of iterations; S24: Update : ， S25: Update ; ， This step uses Fourier transform to introduce periodic signal information into the optimization process so that It is consistent with the periodic characteristics of the respiratory signal; S26: Update : ， S27: Update : ， S28: Update Lagrange multipliers: ， S29: When the change amplitude of all variables is less than the set threshold or reaches the maximum number of iterations, stop the iteration and output the results.

7. A computer storage medium having a computer program stored thereon, characterized in that: When the computer program is executed, the millimeter wave radar breathing monitoring method based on periodic constraints described in any one of claims 1 to 5 is implemented.

Images