Non-contact blood pressure measurement method based on pulse transit time

Abstract

The application discloses a non-contact blood pressure measurement method based on pulse transmission time, and relates to the technical field of medical health monitoring.The chest and wrist of a target are simultaneously monitored by a radar device, and the pulse transmission distance and time are calculated, thereby significantly improving the blood pressure monitoring precision, and overcoming the limitation that the traditional method only relies on a single pulse signal to cause insufficient information amount; meanwhile, by utilizing micro kinetic energy accumulation and point cloud modeling technology, the key parts can be accurately positioned and the signal quality can be enhanced under complex environment, the signal-to-noise ratio is effectively improved, and the sensing problem under the low signal-to-noise ratio environment is solved.In addition, the pulse wave signal is processed through a series of enhancement algorithms, reliable extraction of weak physiological signals is realized, and the non-contact blood pressure monitoring is more accurate and reliable.Based on the calculation method of the human physiological model, the blood pressure measurement result has a clear physiological basis, and the problem that the neural network mapping is not interpretable is avoided.

Description

Technical Field

[0001] This invention relates to the field of medical and health monitoring technology, and in particular to a non-contact blood pressure measurement method based on pulse transit time. Background Technology

[0002] Blood pressure is a core vital sign, and its continuous monitoring is crucial for the prevention of cardiovascular diseases. Traditional blood pressure monitoring technologies mainly include cuff blood pressure monitors and wearable photoplethysmography (PPG) devices. While these methods each have their advantages, they also have significant limitations. Cuff blood pressure monitors measure blood pressure by inflating to block blood flow. Although they offer high accuracy, the measurement process is uncomfortable for patients and continuous monitoring is not possible. Wearable devices provide continuous monitoring through skin contact, but they are not suitable for people with sensitive skin and offer poor comfort during long-term wear.

[0003] In recent years, non-contact blood pressure measurement methods based on wireless sensing technology have gradually developed. These technologies utilize complex neural network algorithms to directly analyze the acquired wireless signals, avoiding the problems associated with skin contact. However, existing technologies still face many challenges in practical applications. First, the working mechanism of neural network models lacks transparency, making it difficult to explain the physiological mechanisms involved in blood pressure calculation. Second, these methods have extremely high requirements for signal quality, while pulse signals in real-world environments are often subject to various interferences. Furthermore, well-trained models have limited adaptability to new environments, requiring extensive labeled data for model optimization.

[0004] Current non-contact blood pressure measurement technologies suffer from problems such as complex system implementation and high cost, limiting their widespread application in home environments. In particular, improvements are needed in areas such as measurement accuracy and stability, device portability, and user adaptability. These issues make current non-contact blood pressure measurement technologies insufficient to meet the growing demand for home health monitoring. Summary of the Invention

[0005] In view of this, in order to achieve the goal of robust and interpretable non-contact blood pressure estimation, this invention provides a non-contact blood pressure measurement method based on pulse transit time. It solves the problem of low signal-to-noise ratio through a multi-domain diversity strategy; solves the problem of insufficient pulse signal information through a multi-point measurement mechanism; and solves the problem of uninterpretable neural network models through a blood pressure estimation model based on pulse wave transit time.

[0006] Therefore, the present invention provides the following technical solution:

[0007] This invention proposes a non-contact blood pressure measurement method based on pulse transit time, the method comprising:

[0008] By using a single radar device to simultaneously monitor the target's chest and wrist, and accumulating the micro-motion energy in each region, the distance and angle information of the target's chest and wrist can be obtained sequentially.

[0009] Based on the distance and angle information of the target chest and target wrist, the position information of the chest and wrist is obtained, and the pulse transmission distance is calculated;

[0010] The received signal is sequentially subjected to frequency domain diversity, time domain diversity, and spatial domain diversity to enhance the quality of the pulse signals from the chest and wrist. The pulse transmission time is calculated based on the delay between the enhanced chest and wrist pulse signals.

[0011] The target blood pressure is calculated based on the pulse transmission distance and the pulse transmission time using a human physiological model.

[0012] Furthermore, by accumulating the microkinetic energy in each region, distance and angle information of the target's chest and wrist are obtained sequentially, including:

[0013] Perform a fast Fourier transform on the energy data along the time dimension;

[0014] The energy within the human body micro-motion zone under each distance chamber is accumulated, and the accumulated energy is detected by constant false alarm rate to obtain the location of the target;

[0015] The energy at various angles is subjected to a fast Fourier transform along the time dimension, the energy within the micro-motion zone of the human body is accumulated, and the accumulated energy is subjected to constant false alarm rate detection to obtain the angle information of the target.

[0016] Furthermore, based on the distance and angle information of the target chest and target wrist, the position information of the chest and wrist is obtained, including:

[0017] The K-means clustering algorithm was used to cluster the results of constant false alarm rate detection to obtain the specific locations of the chest and wrist.

[0018] Further, calculate the distance the pulse travels: ,in Indicates the distance the pulse travels. and These represent the distances between the radar and the chest and wrist, respectively. and These represent the horizontal angles of the radar relative to the chest and wrist, respectively.

[0019] Furthermore, frequency domain diversity, time domain diversity, and spatial domain diversity are sequentially performed on the received signal to enhance the quality of the pulse signal from the chest and wrist, including:

[0020] Frequency domain diversity of the received signal includes: dividing the received signal into several sub-signals of the same length but different starting frequencies, merging the phase-aligned sub-signals, and obtaining the enhanced frequency domain signal;

[0021] Perform time-domain diversity on the frequency-domain diversity signal, including: merging multiple signals in each frame using an equal-gain method to improve signal quality and obtain an enhanced time-domain signal;

[0022] Spatial diversity is performed on the signal after time-domain diversity, including dividing the target into several spatial grids in physical space, filtering and merging the signals from different spatial grids to obtain the pulse signals of the chest and wrist after spatial diversity.

[0023] Furthermore, the received signal is divided into several sub-signals of the same length but different starting frequencies. The phase-aligned sub-signals are then combined to obtain the enhanced frequency domain signal, including:

[0024] The received signal is constructed as a four-dimensional matrix. The four dimensions are antenna, fast time, slow time, and frame time.

[0025] Using a window with a length of Step size is The sliding window edge will be in the fast time dimension The complete signal in the middle is divided into A length of The difference in initial frequency is Sub-signals;

[0026] Calculate the phase shift between each sub-signal : ;

[0027] in, Indicates the distance between the target and the radar. c Represents the speed of light. Indicates frequency resolution. ;

[0028] Phase compensation is performed sequentially based on the phase difference between the sub-signals;

[0029] The phase-compensated sub-signals are combined to obtain the enhanced signal. : .

[0030] Furthermore, multiple signals in each frame are combined using an equal-gain method to improve signal quality, resulting in an enhanced time-domain signal, including:

[0031] Use equal gain to apply the following to each frame: The signals are combined to improve signal quality, resulting in an enhanced signal. : .

[0032] Furthermore, the target is divided into several spatial grids in physical space, and the signals from different spatial grids are filtered and merged to obtain the pulse signals from the chest and wrist after spatial diversity, including:

[0033] Point cloud models were constructed centered on the target's wrist and chest cavity; the point cloud coordinates are respectively... and ,in , , and The value can be:

[0034] ;

[0035] ;

[0036] ;

[0037] ;

[0038] in, , It is distance resolution. It is the interval between adjacent point clouds at the same distance;

[0039] Beamforming technology is used to focus on each point in the point cloud model, and the phase compensation coefficient of each receiver is calculated based on the point cloud coordinates. and The signals obtained after beamforming and focusing of the point cloud data from the chest and wrist were obtained. and :

[0040] ;

[0041] ;

[0042] Where: represents all indices for that dimension. It refers to the number of antennas. and It is calculated by the following formula:

[0043] ;

[0044] ;

[0045] in, It is the signal wavelength. It is the spacing of the receiving antenna array. The number of antennas is the azimuth angle;

[0046] For chest pulse signals, an equal-gain method is used to merge the signals of all spatial points in the chest point cloud, and the merged signal is then... As a template; calculate the signal of each spatial point and the template. The correlation coefficient is used to determine the weighting coefficient. The signals from all spatial points are fused to obtain the spatially diversified chest pulse signal. : ;

[0047] in,

[0048] ;

[0049] ;

[0050] in, express The mean, This represents the number of spatial points in the point cloud. and These respectively represent the first point cloud in the chest area. n The and the first m Signals at each point;

[0051] For wrist pulse signals, the autocorrelation signal of the pulse signal at each spatial point in the wrist point cloud is cross-correlated with the diversity-enhanced chest pulse signal. The maximum cross-correlation value is recorded as the matching score for that point. The scores are then compared... The enhanced wrist pulse signal is obtained by coherently accumulating the signals from each spatial point. .

[0052] Furthermore, the pulse transit time is calculated based on the delay between the enhanced pulse signals from the chest and wrist, including:

[0053] The first h The peak value at the transition point is subtracted from the trough value of its preceding neighbor, and this difference is taken as the peak-to-trough value of that transition point. , No. h Peak and valley values ​​at the transition point If the following conditions are met:

[0054] ;

[0055] Then retain the transition point, where, This represents the first-level dynamic threshold coefficient;

[0056] For the retained transition points, their peak and valley values If the following conditions are met:

[0057] ;

[0058] Then retain the transition point, where This represents the second-level dynamic threshold coefficient. express The median value;

[0059] For the retained transition points, calculate the first... h With the h -1 Spacing between transition points If it satisfies:

[0060] ;

[0061] Then retain the transition point, where This represents the third-level dynamic threshold coefficient;

[0062] After screening using a three-level dynamic threshold, a reliable transition point sequence between wrist pulse signals and chest pulse signals was obtained. ;

[0063] Based on the reliable transition point sequence of the wrist pulse signal and chest pulse signal, the transit time of all pulses was calculated: ;

[0064] in , H This indicates the total number of transition points.

[0065] Furthermore, the target blood pressure is calculated using a human physiological model, including:

[0066] For human arteries exhibiting characteristics of the Von Hypoelasticity model, a formula is established within the range of human blood pressure to determine pulse delivery time. and pulse transmission distance L and blood pressure P Mapping between:

[0067] ;

[0068] in, and It is a constant.

[0069] The advantages and positive effects of this invention are as follows: This invention simultaneously monitors the target's chest and wrist using radar equipment and calculates the pulse transmission distance and time, significantly improving blood pressure monitoring accuracy and overcoming the limitations of traditional methods that rely solely on a single pulse signal, resulting in insufficient information. Simultaneously, by utilizing micro-motion energy accumulation and point cloud modeling techniques, it can accurately locate key areas and enhance signal quality even in complex environments, effectively improving the signal-to-noise ratio and solving the perception problem in low signal-to-noise ratio environments. Furthermore, through a series of enhancement algorithms processing the pulse wave signal, reliable extraction of weak physiological signals is achieved, making non-contact blood pressure monitoring more accurate and reliable. The calculation method based on a human physiological model provides clear physiological evidence for blood pressure measurement results, avoiding the problem of uninterpretable neural network mapping. Ultimately, this technology achieves non-contact measurement without the need for wearing a device, greatly improving ease of use and user experience. Attached Figure Description

[0070] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0071] Figure 1 This is a schematic diagram illustrating the working process of a non-contact blood pressure measurement method based on pulse transit time in an embodiment of the present invention;

[0072] Figure 2 This is a flowchart of a non-contact blood pressure measurement method based on pulse transit time in an embodiment of the present invention;

[0073] Figure 3 This is a schematic diagram of a non-contact blood pressure measurement method based on pulse transit time in an embodiment of the present invention. Detailed Implementation

[0074] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0075] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0076] The core idea of ​​this invention is to utilize a radar device to simultaneously monitor the target's chest and wrist. By analyzing the spatial distance between these two locations and the delay between pulse wave signals, the pulse transmission distance and time are obtained. The target's blood pressure is then calculated using a human physiological model. Specifically, this invention first accumulates the micro-motion energy in the scene to obtain the positional information of the chest and wrist, and then calculates the pulse transmission distance. Second, this invention establishes point cloud models centered on the chest and wrist coordinates, and enhances the quality of the wrist and chest pulse signals through a series of enhancement algorithms. Simultaneously, the pulse transmission time is calculated based on the delay between the chest and wrist pulse wave signals. Finally, the target's blood pressure is calculated using a human physiological model. This invention solves the problem of low signal-to-noise ratio in complex scenes, which makes blood pressure sensing difficult, overcomes the problem of insufficient information from relying solely on pulse signals, and avoids the problem of uninterpretable neural network mapping processes.

[0077] When the user is stationary in front of the radar device with their arm extended, such as Figure 1 As shown, pulse transmission time and distance are obtained by monitoring pulse wave signals from the chest and wrist, which can then be used to calculate the user's blood pressure based on a human physiological model. However, both the wrist pulse signal (pulse) and the chest pulse signal (heartbeat) are extremely weak. The chest cavity movement caused by the heartbeat is only 0.5 mm, and the wrist movement caused by the pulse is only 0.03 mm, which is only 6% of the heartbeat. In addition, the echo signal received by the radar contains not only wrist and chest pulse signals, but also static interference caused by static objects such as human tissue, as well as dynamic interference caused by breathing and micro-movements of the body. Successfully separating and obtaining pure wrist and chest pulse signals is extremely challenging. Therefore, it is necessary to accurately locate the chest cavity and wrist, and at the same time, use a multi-domain diversity strategy to improve signal quality.

[0078] This embodiment is based on the following system configuration: the system operates on a Texas Instruments AWR1443FMCW radar transceiver; the transceiver operates at 77 GHz with a bandwidth of 3.98 GHz, providing a range resolution of 4.37 cm; the transceiver uses a 3-transmit, 4-receive antenna, and with the aid of TDM-MIMO mode, it can form a 12-element virtual array, providing an AoA resolution of 0.25 radians. Task: Estimate the user's blood pressure based on the wireless signal received by the radar.

[0079] System workflow as follows Figure 2 As shown, the process is generally divided into four stages: target localization, sensing signal enhancement, signal segmentation, and blood pressure estimation, as detailed below:

[0080] 1) Target positioning stage:

[0081] This embodiment employs a single-device method to measure signals from multiple parts of the human body. Software weights multi-antenna data with different values, sequentially aligning the beam with the wrist and heart to achieve simultaneous measurement of pulse and heartbeat. Specifically, this embodiment utilizes a single millimeter-wave radar device to accumulate the micro-motion energy of each region, sequentially obtaining distance and angle information for the target chest and wrist. Static objects such as bones generate static interference, making it impossible to directly locate the wrist and chest cavity from the range-angle map (RAM). Therefore, this embodiment utilizes long-time information to amplify signals such as the human pulse and heartbeat. Specifically, there are slight differences in the rate of energy change between the target chest, wrist, and static objects. To obtain the rate of energy change, this invention performs FFT on the energy data along the time dimension. The micro-movements of the human chest and wrist are mainly caused by the heartbeat and pulse. The frequencies of the heartbeat and pulse are between 0.67Hz and 3.67Hz. According to ∆f = 1 / T, 10s of long-time information can provide a frequency resolution of 0.1Hz. Therefore, this invention accumulates the energy within the human body's micro-motion bands under each bin, and then performs CFAR detection on the accumulated energy to obtain the target's location. Determining the target angle is similar to determining the distance. First, an FFT is performed on the energy at each angle along the time dimension. Second, the energy within the human body's micro-motion bands is accumulated. Finally, CFAR detection is performed on the accumulated energy to obtain the target's angle information and generate a Dynamic-Range-Angle Map (DRAM). The positions of the wrist and chest are clearly visible in the DRAM. Then, the results obtained from the CFAR detection algorithm are clustered using a K-means clustering algorithm to obtain the specific positions of the chest and wrist, and the pulse transmission distance is calculated. ,in and These represent the distances between the radar and the chest and wrist, respectively. and These represent the horizontal angles of the radar relative to the chest and wrist, respectively.

[0082] 2) Sensing signal enhancement stage:

[0083] Both wrist and chest pulse signals are very weak and easily affected by dynamic interference such as breathing and subtle body movements, making direct use of data from the point of strongest energy unreliable. Inspired by diversity mechanisms in the field of communications, the inventors proposed a sensing signal enhancement method based on multi-domain diversity. This method fully utilizes limited hardware resources by fusing multi-dimensional projections of the target signal in the frequency, time, and spatial domains to improve the quality of the target signal. The principle of enhanced sensing is as follows: Figure 3 As shown, the received signal is subjected to frequency domain diversity, time domain diversity, and spatial domain diversity in sequence. The wrist pulse signal and the chest pulse signal are subjected to the same frequency domain and time domain diversity measures, but different spatial domain diversity measures.

[0084] Specifically, this invention first performs frequency domain diversity on the received signal. In the frequency domain, common millimeter-wave radars can provide a sweep bandwidth of 4 GHz. Therefore, this invention divides the complete received signal into several sub-signals of the same length but different starting frequencies. Since noise at different times is independent of each other, the phase-aligned sub-signals can be combined to improve the quality of the received signal. Specifically, the received signal is first constructed into a four-dimensional matrix. The four dimensions are antenna, fast time, slow time, and frame time. A window of length is used... Step size is The sliding window edge will be in the fast time dimension The complete signal in the middle is divided into A length of The difference in starting frequency is The sub-signals are then calculated. Next, the phase shift between each sub-signal is calculated. As shown below:

[0085] ;

[0086] in, The distance between the target and the radar is represented by c, where c represents the speed of light. Indicates frequency resolution. Phase compensation is performed sequentially based on the phase difference between the sub-signals. Finally, the phase-compensated sub-signals are combined to improve signal quality, resulting in an enhanced frequency domain signal. As shown below:

[0087] ;

[0088] in, This represents all indices under any one of the four dimensions: antenna, fast time, slow time, and frame time. express The transpose of .

[0089] Secondly, in this embodiment, time-domain diversity is performed on the signal after frequency-domain diversity. Generally, the frame period is on the order of... Since pulse and heartbeat are extremely slow movements, the impact of such slow motion on the phase of the intermediate frequency (IF) signal within the same frame is negligible. Therefore, the phase of the IF signal within a frame can be considered the same. In the time domain, in vital sign monitoring, the frame period of millimeter-wave radar is generally less than 50 milliseconds, and vital sign signals are slowly varying signals, remaining almost unchanged within a frame. Therefore, an equal gain approach can be used to measure the phase of the IF signal within each frame. The signals are combined to improve signal quality, resulting in an enhanced time-domain signal. In this embodiment, the number of signals per frame As shown below:

[0090] ;

[0091] Finally, this invention performs spatial diversity on the signal after time-domain diversity to further improve the quality of the received signal. In the spatial domain, targets in common sensing tasks do not exist in a single spatial grid. For example, in vital sign monitoring and blood pressure monitoring, human movement can have similar effects on wireless signals in multiple spatial grids. Similarly, the pulse and heartbeat of a person can have the same effect on surrounding body tissues. Therefore, the wrist, chest, and surrounding tissues all contain pulse or heartbeat information. Based on the characteristic that the mechanical movement of the heart affects the torso, this invention establishes a weighted spatial diversity method that divides the target into several spatial grids in physical space. Noise in different spatial grids is independent of each other, thus signals from different spatial grids can be merged to improve the quality of the received signal. Specifically, this invention constructs point cloud models centered on the wrist and chest of the target. The point cloud coordinates are as follows: and ,in , , and The possible values ​​are as follows:

[0092] ;

[0093] ;

[0094] ;

[0095] ;

[0096] in, , This refers to distance resolution, in this embodiment... , h This refers to the interval between adjacent point clouds at the same distance. In this embodiment, the interval between adjacent point clouds is defined as... h The value is 0.1m, in different embodiments. h Different values ​​can be taken. This invention utilizes beamforming technology to focus on each point cloud in a point cloud model. Based on the point cloud coordinates, the phase compensation coefficient of each receiver is calculated. and This allows us to obtain the beamformed and focused signals from the point clouds at the chest and wrist. and :

[0097] ;

[0098] ;

[0099] in, It refers to the number of antennas. and It can be calculated using the following formula:

[0100] ;

[0101] ;

[0102] in, It is the signal wavelength. It is the spacing of the receiving antenna array. This represents the number of antennas at the azimuth angle. In this embodiment... , .

[0103] Because the area covered by a point cloud is larger than the actual area occupied by the human wrist, some point cloud information cannot reflect human movement. Filtering out invalid spatial points in the point cloud and selecting the spatial points with the best signal quality is a challenge. This invention first filters and fuses chest pulse signals. Specifically, this invention first uses an equal-gain method to merge the signals of all spatial points in the chest point cloud, and then merges the resulting signal... This serves as a template. Next, the signal at each spatial point is compared to the template. The correlation coefficient is used to determine the weighting coefficient. Finally, the signals from all spatial points are fused to obtain the spatially diversityd signal. :

[0104] ;

[0105] in,

[0106] ;

[0107] ;

[0108] express The mean, and These respectively represent the first point cloud in the chest area. n The and the first m Signals at each point; This represents the number of spatial points in the point cloud, as shown in this embodiment. .

[0109] Secondly, this invention filters and fuses wrist pulse wave signals, such as... Figure 3 As shown, specifically, firstly, the autocorrelation signal of the pulse signal at each spatial point in the wrist point cloud is cross-correlated with the diversity-enhanced chest pulse signal, and the maximum cross-correlation value is recorded as the matching score for that point. Finally, the signals from the five spatial points with the highest scores are coherently accumulated to obtain the enhanced signal. By employing frequency, time, and spatial diversity strategies, the quality of wrist and chest pulse signals was significantly improved, and noise was substantially suppressed.

[0110] 3) Signal segmentation stage:

[0111] Neither wrist pulse signals nor chest pulse signals are simple sinusoidal signals, and directly finding the transition point during cardiac systole and diastole by identifying local peaks is unreliable. This invention sets a three-level dynamic threshold based on local peaks, effectively achieving the calibration of transition points and signal segmentation. Specifically, firstly, the first... h The peak value at the transition point is subtracted from the trough value of its preceding neighbor, and this difference is taken as the peak-to-trough value of that transition point. . No. h Peak and valley values ​​at the transition point If the following conditions are met:

[0112] ;

[0113] Then retain the transition point, where, This represents the first-level dynamic threshold coefficient, in this embodiment. Secondly, for the retained transition points, their peak and valley values... If the following conditions are met:

[0114] ;

[0115] Then retain the transition point, where This represents the second-level dynamic threshold coefficient, in this embodiment. , express The median value. Next, for the retained transition points, calculate the... h With the h -1 Spacing between transition points If it satisfies:

[0116] ;

[0117] Then retain the transition point, where This represents the third-level dynamic threshold coefficient, in this embodiment. Finally, after screening using a three-level dynamic threshold, a reliable transition point sequence between wrist pulse signals and chest pulse signals was obtained. .

[0118] 4) Blood pressure estimation stage:

[0119] Calculate the transit time of all pulses using the transition point between the wrist pulse signal and the chest pulse signal: ;

[0120] in , H This represents the total number of transition points. For human arteries exhibiting characteristics of the Von Hypoelasticity model, this invention establishes a formula within the range of human blood pressure to realize pulse delivery time. and pulse transmission distance L and blood pressure P The mapping between them is as follows: ;

[0121] in, and It is a constant and can be obtained through linear optimization.

[0122] In this embodiment, radar equipment simultaneously monitors the target's chest and wrist, calculating pulse transmission distance and time, significantly improving blood pressure monitoring accuracy and overcoming the limitations of traditional methods that rely solely on a single pulse signal, resulting in insufficient information. Simultaneously, by utilizing micro-motion energy accumulation and point cloud modeling techniques, key areas can be accurately located and signal quality enhanced even in complex environments, effectively improving the signal-to-noise ratio and solving the perception challenge in low signal-to-noise ratio environments. Furthermore, a series of enhancement algorithms process the pulse wave signal, enabling reliable extraction of weak physiological signals, making non-contact blood pressure monitoring more accurate and reliable. The calculation method based on a human physiological model provides clear physiological evidence for blood pressure measurement results, avoiding the uninterpretable problems of neural network mapping. Ultimately, this technology achieves non-contact measurement without the need for wearing devices, greatly improving ease of use and user experience.

[0123] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A non-contact blood pressure measurement method based on pulse transit time, characterized by, The method comprises: Simultaneously monitoring the target chest and the target wrist by using a radar device, and sequentially obtaining distance information and angle information of the target chest and the target wrist by accumulating micro-motion energy of each region; Based on the distance information and the angle information of the target chest and the target wrist, position information of the chest and the wrist is obtained, and a pulse transmission distance is calculated; The received signals are sequentially subjected to frequency domain diversity, time domain diversity and space domain diversity to enhance the quality of the pulse signals of the chest and the wrist, and a pulse transmission time is calculated according to the delay between the pulse signals of the chest and the wrist after signal enhancement; Based on the pulse transmission distance and the pulse transmission time, the target blood pressure is calculated through a human physiological model; The quality of the pulse signals of the chest and the wrist is enhanced by sequentially subjecting the received signals to frequency domain diversity, time domain diversity and space domain diversity, which comprises: The received signals are subjected to frequency domain diversity, which comprises: dividing the received signals into a plurality of sub-signals with the same length and different starting frequencies, and merging the phase-aligned sub-signals to obtain enhanced frequency domain signals; The signals subjected to frequency domain diversity are subjected to time domain diversity, which comprises: merging a plurality of signals in each frame by using an equal gain to improve the signal quality, and obtaining enhanced time domain signals; The signals subjected to time domain diversity are subjected to space domain diversity, which comprises: dividing the target in the physical space into a plurality of spatial grids, screening and merging the signals of different spatial grids to obtain the pulse signals of the chest and the wrist after space domain diversity; The received signals are divided into a plurality of sub-signals with the same length and different starting frequencies, and the phase-aligned sub-signals are merged to obtain enhanced frequency domain signals, which comprises: The received signal is constructed as a four-dimensional matrix with four dimensions for antenna, fast time, slow time and frame time, respectively; Using a window with a length of Step size is The sliding window edge will be in the fast time dimension The complete signal in the middle is divided into A length of The difference in initial frequency is Sub-signals; calculating a phase shift between each of the sub-signals : ; wherein, denotes the distance between the target and the radar, c denotes the speed of light, denotes the frequency resolution, ; The phase difference between the sub-signals is sequentially compensated; The phase compensated sub-signals are combined to obtain an enhanced signal : ; wherein, represents all indices in any of the four dimensions of antenna, fast time, slow time and frame time, represents the transpose of the phase shift of the n th sub-signal, n represents the loop variable taking values from 1 to .​ 2. The non-contact blood pressure measurement method based on pulse transit time according to claim 1, characterized in that, The distance information and the angle information of the target chest and the target wrist are sequentially obtained by accumulating the micro-motion energy of each region, which comprises: The energy data along the time dimension is subjected to fast Fourier transform; The energy in the human micro-motion band under each distance bin is accumulated, and the accumulated energy is subjected to constant false alarm rate detection to obtain the position of the target; The energy along the time dimension in each angle is subjected to fast Fourier transform, the energy in the human micro-motion band is accumulated, and the accumulated energy is subjected to constant false alarm rate detection to obtain the angle information of the target.

3. A non-contact blood pressure measurement method based on pulse transit time according to claim 2, characterized in that, The position information of the chest and the wrist is obtained based on the distance information and the angle information of the target chest and the target wrist, which comprises: The K-means clustering algorithm is used to cluster the results obtained by constant false alarm rate detection to obtain the specific positions of the chest and the wrist.

4. The non-contact blood pressure measurement method based on pulse transit time according to claim 3, characterized in that, The pulse transit distance is calculated as: where represents the pulse transit distance, and represent the distance of the radar from the chest and wrist, respectively, and represent the horizontal angle of the radar from the chest and wrist, respectively.

5. The non-contact blood pressure measurement method based on pulse transit time according to claim 1, wherein, The plurality of signals in each frame are merged by using an equal gain to improve the signal quality, and enhanced time domain signals are obtained, which comprises: The signals in each frame are combined in an equal-gain manner to improve the signal quality and obtain enhanced signals : : .

6. The non-contact blood pressure measurement method based on pulse transit time according to claim 5, characterized in that, The target in the physical space is divided into a plurality of spatial grids, the signals of different spatial grids are screened and merged to obtain the pulse signals of the chest and the wrist after space domain diversity, which comprises: The point cloud model is constructed respectively with the wrist and chest position of the target as the center; the point cloud coordinates are respectively and wherein , , and the values of the above are: ； ； ； ； wherein, , is the distance resolution, is the interval of adjacent point clouds at the same distance; Each point cloud in the point cloud model is focused using beamforming technology, and according to the point cloud coordinates, a phase compensation coefficient of each receiver is calculated and , to obtain the focused signals of the chest and wrist point clouds and : ； ； wherein represents all indices in any one of the four dimensions of antennas, fast time, slow time and frame time, is the number of antennas, and is calculated by the following equation: ； ； wherein is the signal wavelength, is the spacing of the receive antenna array, is the number of antennas in azimuth; For the chest pulse signal, all the signals of the spatial points in the chest point cloud are combined in an equal gain manner, and the combined signals are used as a template; the correlation coefficient of each spatial point signal and the template is calculated to determine the weighting coefficient ; and the signals of all spatial points are fused to obtain the spatial diversity chest pulse signal . : ; The pulse transmission time is calculated according to the delay between the pulse signals of the chest and the wrist after signal enhancement, which comprises: ； ； wherein, represents the mean value of represents the number of spatial points in the point cloud, and respectively represent the signal at the n th and the m th point in the chest point cloud. For wrist pulse signal, the cross-correlation between the autocorrelation signal of each spatial point pulse signal in wrist point cloud and the chest pulse signal after diversity enhancement is performed, and the maximum value of cross-correlation is recorded as the matching score of the point. The signals of the top space points are coherently accumulated to obtain the enhanced wrist pulse signal .

7. The non-contact blood pressure measurement method based on pulse transit time according to claim 1, wherein, The target blood pressure is calculated through a human physiological model, which comprises: the first h the peak value of the transition point is subtracted from the trough value of its forward adjacent wave, and the difference is taken as the peak-trough value of the transition point , the first h the peak-trough value of the transition point if the following is satisfied: ； then the transition point is retained, wherein denotes the first level dynamic threshold coefficient; For the remaining transition points, the peak-to-valley value If the following is satisfied: ； then the transition point is retained, wherein denotes a second level dynamic threshold coefficient, denotes the median value of For the remaining transition points, the interval between the (i-1)th and the ith transition points is calculated h and the ith transition point is calculated as h -1 transition point if it satisfies: ； then the transition point is retained, wherein denotes a third level dynamic threshold coefficient; After three levels of dynamic threshold screening, the reliable transition point sequence of the wrist pulse signal and the chest pulse signal is obtained ; Based on the reliable transition point sequences of the wrist pulse signal and the chest pulse signal, all pulse transit times are calculated: ; wherein , H denotes the total number of transition points.

8. The non-contact blood pressure measurement method based on pulse transit time according to claim 7, characterized in that, ​ For human arteries with characteristics of von super-elasticity model, a formula is established to realize the mapping between pulse transit time and pulse transit distance L and blood pressure P in the range of human blood pressure. ； wherein and is a constant, .

Images