Vital sign signal enhancement method and system based on frequency modulated continuous wave radar

Abstract

The invention discloses a vital sign signal enhancement method and system based on a frequency modulated continuous wave radar, and the method comprises the steps: obtaining a radar echo signal, and recombining a fast time sampling point into a subarray in a slow time direction; screening the sub-arrays according to a signal quality evaluation mechanism, and rejecting low-quality sub-arrays to form sparse sub-arrays; digital beam forming of the distance dimension is executed on the sparse subarray, and asymmetric beam pattern constraint is introduced; and extracting vital sign phase information from the slow time signal output by the beam forming. According to the method, the signal-to-noise ratio of the vital sign signal can be improved, the signal detectability is enhanced, and the detection accuracy and robustness are guaranteed.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of signal enhancement, and relates to a vital sign signal enhancement method and system based on a frequency-modulated continuous wave radar. BACKGROUND

[0002] With the continuous progress of society and the improvement of living standards, people pay more and more attention to health problems, and the monitoring of human health status has become one of the focuses of social attention, and the heart rate is an important indicator for evaluating human health. Abnormal changes in heart rate may indicate potential health problems, such as pulmonary hypertension, cardiovascular disease, and arrhythmia. In addition, heart rate is closely related to stress level and emotional state. Through the monitoring of heart rate, individual emotional fluctuations can be effectively identified, thereby providing a scientific basis for emotional management and psychological intervention.

[0003] Millimeter wave detection of heart rate is mainly through detecting small amplitude displacement on the chest surface to extract information. This displacement signal is very weak, so it is easily disturbed by noise and chest displacement introduced by breathing. The existing frequency-modulated continuous wave radar for detecting vital signs usually extracts the phase change of the signal in the corresponding distance interval after performing fast Fourier transform on the fast-time data in the distance dimension. However, this scheme has its inherent shortcomings because the fast Fourier transform has spectrum leakage, which causes the interference signal introduced by similar abdominal movements to leak into the distance interval of interest, affecting the actual vital sign detection. SUMMARY

[0004] To solve the problems in the background art, the application provides a vital sign signal enhancement method and system based on a frequency-modulated continuous wave radar.

[0005] To achieve the above purpose, the technical scheme adopted by the application is as follows: a vital sign signal enhancement method based on a frequency-modulated continuous wave radar, comprising the following steps: Obtaining a radar echo signal, recombining fast-time sampling points along the slow-time direction into a subarray; Screening the subarray according to a signal quality evaluation mechanism to form a sparse subarray; Performing digital beamforming in the distance dimension on the sparse subarray based on an asymmetric beam pattern constraint; Extracting phase information of vital signs from the slow-time signal output by the digital beamforming.

[0006] Specifically, the expression of the i-th subarray in the subarray is: Wherein, A is a signal amplitude, ​​sampling time, distance of target and radar, time stamp of slow time dimension, ratio of modulation frequency and time, light speed, center frequency of radar, residual phase noise, and

[0007] Specifically, the sparse subarray corresponds to an array weight vector: ; When the subarray sequence number belongs to the selected subarray sequence , ; When does not belong to , = ; wherein and are the amplitude and phase of the first subarray weight, respectively.

[0008] Specifically, the signal quality evaluation mechanism comprises: In the heart rate frequency band, the energy concentration ratio and the heart peak bottom ratio HPFR of each subarray are calculated, only the subarrays with the energy concentration ratio greater than a preset threshold γ1 and the heart peak bottom ratio HPFR greater than a preset threshold γ2 are retained; In the respiratory frequency band, the respiratory peak bottom ratio BPFR of each subarray is calculated, and the subarrays with the respiratory peak bottom ratio BPFR less than a preset threshold γ3 are removed.

[0009] Specifically, the calculation of the energy concentration ratio , the heart peak bottom ratio HPFR and the respiratory peak bottom ratio BPFR are all based on the spectrum amplitude P(f); The spectrum valley values and on both sides of the highest peak in the heart rate frequency band define the peak range, is the ratio of the total energy in the range to the total energy of the heart rate frequency band; and are the spectrum valley values on both sides of the highest peak in the heart rate frequency band; and are the spectrum valley values on both sides of the highest peak in the respiratory frequency band; The average power of the high-frequency noise band is the average energy value within the preset high-frequency band.

[0010] Specifically, the asymmetric beammap constraint is achieved by optimizing the weight vector, and the objective function includes: Maximize the gain of the main valve in the chest target area; Minimize the residual power in the ventral direction zero-depression; Multipath range sidelobe energy minimization; And adjust various constraints by weighing factors.

[0011] Specifically, the asymmetric beammap constraint is achieved by optimizing the weight vector, and the objective function is: ; in The remaining energy in the abdominal region, The remaining energy in the multipath interference region, This is a penalty term used to suppress the maximum level of the sidelobes. and This is a trade-off factor used to adjust the importance of various constraints.

[0012] Specifically, the synthesized signal for digital beamforming is: in For the first The weights of each subarray, For the first The signals of each subarray The timestamp is for the slow time dimension.

[0013] The present invention also provides a vital sign signal enhancement system based on frequency modulated continuous wave radar, comprising: The subarray processing module is used to reassemble radar echo signals into subarrays and select sparse subarrays based on the signal quality assessment mechanism. The beamforming module is used to perform distance-dimensional digital beamforming on sparse subarrays and introduces asymmetric beammap constraints in the weighted design. The information extraction module is used to extract phase information of vital signs from the slow-time signal output by digital beamforming.

[0014] Compared with the prior art, the present application has the following beneficial effects: by reorganizing the echo signal into subarrays and removing low-quality subarrays according to a signal quality evaluation mechanism to form sparse subarrays, the signal-to-noise ratio of vital sign signals can be improved; in combination with digital beamforming with asymmetric beam pattern constraints, the main lobe of the beam can be focused on the chest area to strengthen the effective signal, while the abdominal area is suppressed to reduce the signal level in the multipath interference area, significantly reducing the influence of interference on signal extraction, thereby enhancing the detectability of vital sign signals and improving the accuracy and robustness of vital sign monitoring, providing effective technical support for non-contact vital sign detection based on frequency-modulated continuous wave radar. BRIEF DESCRIPTION OF DRAWINGS

[0015] Figure 1 is a principle diagram of a vital sign signal enhancement method based on frequency-modulated continuous wave radar according to the present application; Figure 2 is a simulation diagram of the effect of removing low signal-to-noise ratio subarrays according to the present application; Figure 3 is a flowchart of asymmetric beam interference suppression according to the present application; Figure 4 is a simulation diagram of the effect of asymmetric beam and uniform Chebyshev array interference suppression according to the present application. DETAILED DESCRIPTION

[0016] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0017] As shown in Figures 1-4 , the technical solutions adopted by the present application are as follows: a vital sign signal enhancement method based on frequency-modulated continuous wave radar, comprising the following steps: acquiring radar echo signals and reorganizing fast time sampling points into subarrays along the slow time direction; screening the subarrays according to a signal quality evaluation mechanism to form sparse subarrays; performing digital beamforming in the distance dimension on the sparse subarrays based on asymmetric beam pattern constraints; extracting phase information of vital signs from the slow time signals output by digital beamforming.

[0018] Specifically, the radar sensor transmits electromagnetic waves to the static human body to be measured using a linear frequency-modulated continuous wave, and the receiving antenna performs down-conversion and sampling processing on the received signals after receiving the echo signals. For a single target, the intermediate frequency signal of the frequency-modulated continuous wave radar is: is the amplitude of the intermediate frequency signal, reflecting the signal strength; is the ratio of the modulation frequency to time, i.e. the frequency modulation slope; is the distance between the target and the radar, which changes dynamically with the time stamp of the slow time dimension due to the slight displacement caused by human life activities (such as heartbeat, breathing); c is the speed of light; is the center frequency of the radar, which is the reference frequency of the signal; corresponds to the time stamp of the slow time dimension; corresponds to the time stamp of the fast time dimension; is the residual phase noise, reflecting the residual phase that has not been completely cancelled in the signal propagation process.

[0019] By expanding the sampling points in the fast time along the slow time into multiple sub-arrays.

[0020] Specifically, the expression of the i-th sub-array in the sub-array is: where is the corresponding sampling time (i.e. the fast time sampling interval), which determines the density of fast time sampling. is the signal amplitude of the sub-array, which is related to the signal strength of the sub-array. is the distance between the target and the radar. is the sub-array number, corresponding to different fast time sampling points.

[0021] Further, after the radar sensor sends a linear frequency modulated continuous wave to the static human body to be measured, the receiving antenna acquires the echo signal, which is first down-converted to an intermediate frequency signal to facilitate subsequent analysis. Then the analog signal is converted into a digital signal through sampling processing.

[0022] By expanding the sampling points in the fast time along the slow time direction to form multiple sub-arrays. Where fast time refers to the sequence of sampling time within a single frequency modulation period, and slow time refers to the time sequence of different frequency modulation periods. After expansion, each sub-array corresponds to the signal change of a specific fast time sampling point in the slow time. The two exponential terms respectively represent the modulation characteristics of the fast time dimension and the phase change of the slow time dimension, which together constitute the complete characteristics of the sub-array signal.

[0023] Assuming that when the radar detects K targets, the relevant expression can be arranged as: where, is the manifold matrix of the array. ​​

[0024] And is the steering vector corresponding to the th target, the index of the complex exponential term in the steering vector is related to the subarray index , reflecting the phase difference when different subarrays receive the target signal, which is determined by the spatial position relationship between the target and each subarray, and embodies the response characteristics of the array to target signals in different directions.

[0025] is a vector containing target signal characteristics, where each element is the signal amplitude of the th target, reflecting the dynamic change of the distance between the target and the radar with the slow-time dimension timestamp (such as the slight displacement caused by life activities). is the residual phase noise, which together constitutes the amplitude and phase information of the target signal. is the corresponding noise component, composed of the noise of each subarray, and by giving each subarray different weights, the final phase signal is synthesized.

[0026] Specifically, the synthesized signal of the digital beamforming is: where is the weight of the th subarray, used to adjust the contribution of the subarray in the synthesized signal. is the signal of the th subarray, containing the corresponding target signal and noise. is the slow-time dimension timestamp.

[0027] From the above two formulas, it can be found that the noise of each subarray will participate in the synthesis together with the subarray signal. When some subarrays have weak signal-to-noise ratio (the effective signal is basically covered by noise), their noise will be brought into the synthesized signal by the weight, which not only cannot enhance the effective signal, but also dilutes the overall signal quality, and has a negative impact on the final synthesized signal.

[0028] In order to solve this problem, a signal quality evaluation mechanism is introduced to evaluate and filter the subarray signals. By evaluating the quality of the subarray signals, the selected subarray sequence is determined, and based on this sequence, the array weight vector is defined to realize the targeted use of high-quality subarrays.

[0029] Specifically, the array weight vector corresponding to the sparse subarray formed is: that is, when the subarray index Belonging to the selected subarray sequence hour, ; when Not belonging to hour = This means that low-quality subarrays do not participate in signal synthesis, thus avoiding their noise from interfering with the overall signal.

[0030] The magnitude of the subarray weights is used to adjust the contribution intensity of the subarray signal in the synthesis, thereby increasing the proportion of effective signal. For the first The phase of each subarray weight is used to ensure that the subarray signal is phase-coordinated with other selected subarray signals, reducing phase cancellation and improving the consistency of the synthesized signal. This ensures that the final output is mainly constructed from high-quality subarray signals, thereby improving the robustness and reliability of the overall spectrum estimation.

[0031] The core principle is to retain high-quality subarrays through screening. It assigns reasonable amplitude and phase weights to the subarrays and excludes low-quality subarrays, so that the final synthesized signal is mainly constructed from subarray signals with effective information, thereby reducing noise introduction and improving the robustness and reliability of spectrum estimation.

[0032] Figure 2 The simulation results based on this removal approach are shown. This simulation was performed using Monte Carlo simulation with the following settings: radar center frequency. =59GHz, bandwidth BW=2GHz, slow time frame rate 50Hz, fast time sample count 128, sampling rate 7.14MHz. Figure 2 The image shows a target distance profile obtained using the MUSIC algorithm. It assumes the target is moving sinusoidally at a distance of 2 meters.

[0033] In the simulation, each subarray was given the same fixed noise power while being given target signals of varying intensities, causing the signal-to-noise ratio (SNR) of the subarrays to vary randomly between -5dB and 20dB.

[0034] First, each subarray is weighted with the same amplitude, then different phase weights are applied to focus the beam onto the target position, thereby extracting the phase information of the target's motion. The generated beam pattern is as follows. Figure 2 As shown.

[0035] After obtaining the full array signal, subarrays with a signal-to-noise ratio (SNR) less than or equal to 0 dB were discarded, and then the relevant phase information was re-extracted. Finally, the SNR of the extracted signals from the two schemes was compared. In this simulation, a total of 100,000 Monte Carlo experiments were conducted, and the final simulation results are as follows:Figure 2 The figure shows the difference in signal-to-noise ratio between the final extracted signal of the removed subarray and the non-removed subarray at different low signal-to-noise ratio subarray proportions.

[0036] It can be observed that the synthesized signal after removing the subarray achieves a higher signal-to-noise ratio, and this improvement becomes more significant as the proportion of low signal-to-noise ratio subarrays increases. Theoretically, the signal-to-noise ratio can be improved by more than 6dB.

[0037] Based on the simulation results described above, a subarray screening process based on heart rate signal quality is incorporated into the method design to improve the detectability of the heart rate signal.

[0038] Specifically, the signal quality evaluation mechanism includes: In the heart rate frequency band, the energy concentration ratio of each subarray is calculated and the heart peak-to-bottom ratio HPFR, only the subarray with greater than the preset threshold γ1 and HPFR greater than the preset threshold γ2 is retained; In the respiratory frequency band, the respiratory peak-to-bottom ratio BPFR of each subarray is calculated, and the subarray with BPFR less than the preset threshold γ3 is removed.

[0039] Specifically, the energy concentration ratio , the heart peak-to-bottom ratio HPFR and the respiratory peak-to-bottom ratio BPFR are all based on the spectral amplitude P(f); The spectral valley value on both sides of the highest peak in the heart rate frequency band and defines the peak range, which is the ratio of the total energy in this range to the total energy in the heart rate frequency band; and is the spectral valley value on both sides of the highest peak in the heart rate frequency band; and is the spectral valley value on both sides of the highest peak in the respiratory frequency band; The average power of the high-frequency noise band is the average energy in the preset high-frequency frequency band.

[0040] Further, the screening process is based on spectral features obtained through spectral analysis and is divided into two stages.

[0041] In the first stage, the frequency band of the normal person's heart rate, i.e. 0.8Hz to 2.5Hz, is checked, and two indicators are mainly calculated in this frequency band, one of which is the energy concentration ratio, which measures the degree of energy concentration around the heart rate peak, and can be represented as: where is the amplitude on the corresponding spectrum, and corresponds to the two spectral valley values on both sides of the highest peak in the 0.8Hz-2.5Hz band. The larger the parameter is, the stronger and more concentrated the heartbeat signal is.

[0042] Another indicator is the heart peak-to-floor ratio (HPFR), which quantifies the strength of the dominant peak relative to the noise floor, and can be expressed as: where the denominator can be used as the average power of the high-frequency noise band. Only the subarray that satisfies both and can be included in the subsequent phase extraction, where γ1 and are two empirical parameters.

[0043] The second stage is based on the breathing intensity of ordinary people for screening, and the corresponding parameter is the breathing peak-to-floor ratio (BPFR): and corresponds to the two spectral valley values on both sides of the highest peak in the 0.1Hz-0.5Hz band. If is less than a certain threshold γ3, the subarray is deleted.

[0044] The second stage of screening is a safeguard to avoid the situation where even if the signal-to-noise ratio of the heart signal is good enough, it is affected by the breathing harmonic and the first stage does not work. This two-stage process ensures that only frames with reliable heartbeat signal content are used, thereby improving the robustness and accuracy of signal extraction.

[0045] On the basis of completing signal screening, the direction of the beam is adjusted for interference suppression, and the flow chart is shown in Figure 3 . First, the actual detection area is divided by the MUSIC algorithm, which can be divided into the chest, abdomen, and multipath interference area.

[0046] Specifically, the asymmetric beam pattern constraint is realized by optimizing the weight vector, and the objective function includes: maximizing the chest target main lobe gain.

[0047] minimizing the abdominal direction nulling residual power.

[0048] minimizing the multipath range sidelobe energy.

[0049] And each constraint is adjusted by a weighting factor.

[0050] By optimizing the weight vector, the beam main lobe is focused on the chest region (where the actual target is located), enhancing the reception gain of the chest vital sign signal and ensuring that the effective signal dominates in the synthesis.

[0051] The abdominal direction nulling residual power minimization can be expressed by the formula: Wherein and correspond to the range before and after the abdomen, is the weight vector of the array, is the distance where the chest is located corresponds to the steering vector. By optimizing the weight vector, the signal residual energy in the abdominal region (from to ) approaches zero, forming an energy null, thereby suppressing the interference signal caused by abdominal movement (such as displacement caused by breathing).

[0052] In order to suppress multipath interference, the multipath range sidelobe energy minimization can be expressed by the formula: Wherein indicates that the target is to minimize the sidelobe energy of the multipath interference region to suppress the multipath signal interference to the effective signal, is the distance where the actual target such as the chest is located. Wherein is the weight vector of the array, is the distance where the chest is located corresponds to the steering vector.

[0053] The constraint condition ensures that the weight vector cooperates with the steering vector of the chest position to maintain the chest main lobe gain; at the same time, by minimizing the target, the sidelobe energy of the multipath interference region is reduced, and the interference of the multipath signal such as environmental reflection to the effective signal is reduced.

[0054] By adjusting the importance of the above three constraints through the weighing factor, the weight vector can be dynamically optimized according to the priority of the demand for chest signal enhancement, abdominal suppression, and multipath suppression in the actual scene, to ensure that the overall beam performance meets the demand for vital sign signal extraction.

[0055] Designing such a beam pattern can be expressed as an optimization problem, in which the antenna weight vector (a vector used to adjust the contribution degree of each subarray signal in the synthesis process, i.e. ) will be adjusted according to these spatial constraints. By optimizing the weight vector through the asymmetric beam pattern constraint, the non-target function can be obtained.

[0056] Specifically, the asymmetric beam pattern constraint is realized by optimizing a weight vector, and a target function is: wherein is the residual energy of the abdominal region, which is incorporated into the target function to suppress the interference signal generated by the abdominal movement by minimizing this value, and to ensure that the energy of the abdominal region is as low as possible. is the residual energy of the multipath interference region, and by reducing this value, the interference of the multipath signal such as environmental reflection on the effective signal can be reduced. is a penalty term for suppressing the maximum level of the sidelobe, which is used to avoid the introduction of new interference caused by the excessive sidelobe level, and to ensure the focusing property of the main lobe of the beam.

[0057] and are trade-off factors for adjusting the importance of each constraint. The priorities of the abdominal interference, multipath interference and sidelobe suppression in the actual scene can be adjusted. If the abdominal interference needs to be suppressed, the weight of can be increased. If the multipath interference is more significant, the can be increased to strengthen the constraint on . If the sidelobe is too high, the can be increased to enhance the effect of , so as to dynamically optimize the beam performance and adapt to the signal extraction requirements in different detection environments.

[0058] To verify the proposed algorithm, simulation experiments are also conducted. Two targets are placed in front of the radar at 2.00 m and 2.07 m, respectively. The target located at 2.00 m is set to oscillate at a frequency of 1 Hz and a displacement amplitude of 0.2 mm. The target located at 2.07 m is regarded as an abdominal interference source, and its oscillation frequency is 0.2 Hz and the displacement amplitude is 2 mm. In addition, a number of interference sources are arranged in the environment to simulate the multipath effect and noise in real conditions.

[0059] First, the beamforming scheme of the uniform Chebyshev array is used for interference suppression, and then the scheme proposed in the present application is used, Figure 4 the beam patterns formed by the two methods and the positions of the targets are shown. It can be found that the overall sidelobe level of the pattern formed by the present application on the side of the main lobe is reduced by 5 dB compared with the sidelobe level of the Chebyshev array, which means that less environmental interference power enters during the signal extraction process. Figure 4 The results of displacement demodulation of the 2.00 m target by the two methods are shown respectively. The method proposed in the present application realizes a root mean square error of 0.05 mm, which is significantly lower than the 0.128 mm realized by the other method, indicating that it has better interference suppression performance.

[0060] ​The present invention also provides a vital sign signal enhancement system based on frequency modulated continuous wave radar, comprising: The subarray processing module is used to reassemble radar echo signals into subarrays and select sparse subarrays based on the signal quality assessment mechanism. The beamforming module is used to perform distance-dimensional digital beamforming on sparse subarrays and introduces asymmetric beammap constraints in the weighted design. The information extraction module is used to extract phase information of vital signs from the slow-time signal output by digital beamforming.

[0061] In one specific embodiment, frequency-modulated continuous wave radar is used to detect vital signs of a seated human body, with the goal of extracting a clear heartbeat signal while reducing interference from abdominal respiratory movements and environmental multipath reflections.

[0062] The first step is radar signal acquisition.

[0063] The radar operates in a linear frequency modulated continuous wave mode, with the following parameters set: center frequency. =59GHz (millimeter wave band, suitable for detecting minute displacements), modulation bandwidth 2GHz (improving range resolution), slow time frame rate 50Hz (collecting 50 frames per second, covering the time changes of heartbeat and respiration), fast time sampling number 128 (collecting 128 points in the fast time during each signal transmission for calculating target distance). The radar emits electromagnetic waves towards the human body. After the receiving antenna captures the reflected echo, it first converts the high-frequency signal to an intermediate-frequency signal through down-conversion (for easier subsequent processing), and then converts the analog signal into a digital signal through sampling to obtain the raw echo data.

[0064] The second step is subarray recombination.

[0065] The signal is broken down into smaller parts. The original signal contains two dimensions: fast time (sampling moments within a single transmission) and slow time (the time series of multiple transmissions). To analyze the signal more precisely, the 128 sampling points in the fast time dimension are expanded along the slow time dimension, forming 128 subarrays. Each subarray is equivalent to capturing the signal variation of a specific fast time sampling point across all slow time frames.

[0066] For example, the subarray corresponding to the 3rd fast time sampling point It records the signal changes (including heartbeat, respiration and noise) of the sampling point in a slow time series of 50 frames per second as the human chest cavity or abdomen moves slightly.

[0067] The third step is subarray filtering.

[0068] Only high-quality signals are kept, not all sub-arrays are useful, some sub-arrays can be overwhelmed by noise or interference. Therefore, a two-stage screening is needed to retain high-quality sub-arrays, that is, to form a sparse sub-array.

[0069] First stage: pick out sub-arrays with strong heartbeat signals. The normal human heart rate frequency band is usually 0.8Hz-2.5Hz (about 48-150 times / minute). Do spectral analysis on the signal of each sub-array, focusing on this frequency band. Calculate the energy concentration ratio . Look at whether the energy around the heartbeat peak is concentrated (for example, the energy around the peak accounts for more than 60% of the total energy in this frequency band, i.e. γ1=0.6), the higher the ratio, the more focused the heartbeat signal.

[0070] Calculate the heart peak bottom ratio (HPFR). Look at whether the heartbeat peak intensity is much higher than the noise (for example, the peak intensity is more than 3 times the noise bottom, i.e. γ2=3), the higher the ratio, the more prominent the heartbeat signal than the noise.

[0071] Only sub-arrays that meet both conditions (for example, 90 out of 128 sub-arrays meet the conditions) can enter the next stage.

[0072] Second stage: exclude sub-arrays with strong respiratory interference. The normal human breathing frequency is usually 0.1Hz-0.5Hz (about 6-30 times / minute), if the respiratory signal is too strong, it may mask the heartbeat. Therefore, calculate the breathing peak bottom ratio (BPFR) of each sub-array in the respiratory frequency band, if the value is lower than the threshold γ3=2 (indicating that the respiratory interference is too strong), the sub-array is removed. Finally, 80 clean sub-arrays are left to form a sparse sub-array.

[0073] Fourth step, asymmetric beamforming.

[0074] Align the signal to the chest and avoid interference. Perform digital beamforming on the 80 sparse sub-arrays after screening by adjusting the weight of each sub-array (i.e. in ), so that the beam can intelligently focus.

[0075] Main lobe aligns with the chest: the weight design ensures that the signal in the chest area (e.g. 2.00m away from the radar) is amplified to the maximum extent.

[0076] Null aligns with the abdomen: by optimizing the weight, the signal energy in the abdominal area (e.g. 2.05-2.10m away from the radar) is close to zero, i.e. is minimized, suppressing the interference of abdominal movement caused by breathing.

[0077] Lowering the side lobe reduces multipath: through the objective function , the side lobe energy in the area where the environmental reflection (such as the multipath signal reflected by the wall) is suppressed, avoiding the mixing of these signals into the effective signal.

[0078] For example, set the trade-off factor = 0.3 (emphasis on suppressing multipath), = 0.2 (emphasis on suppressing sidelobe), the final beam can accurately lock the chest while shielding the abdomen and environmental interference.

[0079] Step 5, extract the heartbeat signal.

[0080] After beamforming, the synthesized signal has been greatly weakened. Extracting the phase change (reflecting the small displacement of the chest cavity) from this signal can demodulate the clear heartbeat signal. For example, the detected heartbeat frequency is 1.1 Hz (66 times per minute), the chest displacement amplitude is about 0.1 mm, and the displacement measurement error is only 0.05 mm (much lower than the 0.128 mm of the unoptimized scheme), achieving high-quality vital sign extraction.

[0081] Although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art can modify the technical solutions described in the foregoing embodiments or make equivalent replacements to some of the technical features, and any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for vital sign signal enhancement based on frequency-modulated continuous wave radar, characterized in that, The method comprises the following steps: Obtaining radar echo signals, and reorganizing fast time sampling points into sub-arrays along the slow time direction; Screening the sub-arrays according to a signal quality evaluation mechanism to form sparse sub-arrays; Performing digital beamforming in the distance dimension on the sparse sub-arrays based on an asymmetric beam pattern constraint; Extracting phase information of vital signs from slow time signals output by the digital beamforming.

2. The method of claim 1, wherein the method is based on a frequency-modulated continuous wave radar. The subarray in the first The expression for the subarray is: wherein is the signal amplitude, is the sampling time, is the target to radar distance, is the slow time dimension time stamp, is the ratio of modulation frequency to time, is the speed of light, is the radar center frequency, is the residual phase noise and .

3. The method of claim 1, wherein the method is based on a frequency-modulated continuous wave radar. The array weight vector corresponding to the sparse sub-arrays: ； When the subarray sequence number belongs to the selected subarray sequence , ; When Not belonging to Time = ; wherein and are the magnitude and phase of the th subarray weight, respectively.

4. The method of claim 1, wherein, The signal quality evaluation mechanism comprises: In the heart rate frequency band, calculate the energy concentration ratio of each subarray and the heart peak to floor ratio HPFR, only keep the subarray whose energy concentration ratio is greater than a preset threshold γ1 and whose HPFR is greater than a preset threshold γ2 In the respiratory frequency band, calculating the respiratory peak bottom ratio (BPFR) of each sub-array, and removing the sub-array with a BPFR less than a preset threshold γ3.

5. The method of claim 4, wherein the method is based on a frequency-modulated continuous wave radar. the energy concentration ratio The calculation of the heart peak floor ratio HPFR and the breath peak floor ratio BPFR are both based on the spectral amplitude P(f). the spectral valley on both sides of the highest peak in the heart rate band and defining a peak range, the ratio of the sum of the energy in this range to the total energy of the heart rate band; and are the spectral valley values on both sides of the highest peak in the heart rate band; and are the spectral valley values on both sides of the highest peak in the respiratory band The average power of the high-frequency noise band is the average energy in a preset high-frequency band.

6. The method of claim 1, wherein, The asymmetric beam pattern constraint is realized by optimizing the weight vector, and the objective function comprises: Maximizing the main lobe gain of the chest target; Minimizing the residual power of the zero trap in the abdominal direction; Minimizing the sidelobe energy in the multipath range; And adjusting each constraint by a weighting factor.

7. The method of claim 6, wherein the method is based on a frequency-modulated continuous wave radar. The asymmetric beam pattern constraint is realized by optimizing the weight vector, and the objective function is: ； wherein is the remaining energy of the abdominal region, is the remaining energy of the multi-path interference region, is a penalty term for suppressing the sidelobe maximum level, and is a trade-off factor for adjusting the importance of each constraint.

8. The method of claim 1, wherein the method is based on a frequency-modulated continuous wave radar. The composite signal of the digital beamforming is: wherein is the weight of the th sub-array, is the signal of the th sub-array, is the time stamp of the slow time dimension.

9. A frequency modulated continuous wave radar based vital signs signal enhancement system, characterized in that, Comprise: A sub-array processing module configured to reorganize radar echo signals into sub-arrays, and screen sparse sub-arrays according to a signal quality evaluation mechanism; A beamforming module configured to perform digital beamforming in the distance dimension on the sparse sub-arrays, and introduce an asymmetric beam pattern constraint in the weight design; An information extraction module configured to extract phase information of vital signs from slow time signals output by the digital beamforming.

Images