Heart rate monitoring method, device, system and smart mattress based on BCG signal

Abstract

The present application provides a heart rate monitoring method, device, system and smart mattress based on BCG signals. The monitoring method includes: performing frequency domain transformation on the collected initial BCG signal and initial noise reference signal respectively to obtain a target BCG frequency domain signal and a target noise reference frequency domain signal; obtaining a first attenuation coefficient and a second attenuation coefficient based on the target BCG frequency domain signal and the target noise reference frequency domain signal; obtaining a target spectrum signal based on the first attenuation coefficient and the second attenuation coefficient; performing motion state detection on the initial noise reference signal to obtain a motion state detection result; and performing fusion processing on the target spectrum signal based on the motion state detection result to obtain a target heart rate monitoring value. The present application solves the problem of low accuracy of heart rate monitoring based on BCG signals in the prior art. The present application not only improves the accuracy of the heart rate monitoring value, but also avoids monitoring errors caused by static weights.

Description

Technical Field

[0001] The present invention relates to the technical field of heart rate monitoring, and in particular to a heart rate monitoring method, device, system and smart mattress based on BCG signals. Background Art

[0002] The ballistocardiogram (BCG) is a physical vibration signal generated by cardiac activity. Its periodic variations can be used to detect important physiological indicators such as heart rate and respiration. BCG signals do not require human contact during acquisition, offering advantages such as low cost, wide applicability, and high safety. They are widely used in smart healthcare and intelligent monitoring.

[0003] BCG signals collected from a resting person exhibit a strong correlation with standard ECG signals in terms of waveform characteristics and the physiological information they contain. Heart rate fluctuations can be reflected by counting the JJ intervals of the BCG signal. However, smart care devices based on BCG signals often cannot guarantee that the user remains restful while in use. To detect subtle BCG signal variations, these devices typically use pressure-sensitive and piezoelectric sensors as the front-end for BCG signal acquisition. Consequently, these BCG-based smart care devices face the following challenges in practical use: 1. Self-induced vibrations, such as rolling over or involuntarily shaking the legs, can severely interfere with the reliability of sensor data. When the BCG signal is mixed with strong non-periodic noise, extracting information such as heart rate and respiration becomes extremely difficult. 2. While pressure-sensitive and piezoelectric sensors can sense subtle vibrations, they cannot distinguish between static objects and the human body. Even non-living objects, if applied with sufficient pressure by the pressure-sensitive or piezoelectric sensors, can generate data, causing the algorithm to interpret non-existent heart rates based on the sensor data.

[0004] It can be seen that the heart rate monitoring method based on BCG signals in the related art has serious noise interference or misjudgment, and the accuracy of heart rate monitoring is poor. Summary of the Invention

[0005] The present application provides a heart rate monitoring method, device, system and smart mattress based on BCG signals, which solves the problem of low accuracy of heart rate monitoring based on BCG signals in the existing technology. The present application uses a noise reference signal to assist in noise reduction, which not only improves the accuracy of the heart rate monitoring value, but also avoids monitoring errors caused by static heavy objects.

[0006] In the first aspect, the present application provides a heart rate monitoring method based on a BCG signal, the monitoring method comprising: performing frequency domain transformation on an initial BCG signal and an initial noise reference signal collected within a current sliding time window, respectively, to obtain a target BCG frequency domain signal and a target noise reference frequency domain signal; obtaining a first attenuation coefficient and a second attenuation coefficient within the current sliding time window according to the target BCG frequency domain signal and the target noise reference frequency domain signal; obtaining a target spectrum signal according to the first attenuation coefficient and the second attenuation coefficient; performing motion state detection on the initial noise reference signal to obtain a motion state detection result; and performing fusion processing on the target spectrum signal according to the motion state detection result to obtain a target heart rate monitoring value.

[0007] Optionally, according to the target BCG frequency domain signal and the target noise reference frequency domain signal, obtaining the first attenuation coefficient within the current sliding time window, including: according to the target BCG frequency domain signal, obtaining the target BCG frequency domain signal output result within the current sliding time window; according to the target BCG frequency domain signal output result and the target noise reference frequency domain signal, obtaining the first attenuation coefficient within the current sliding time window.

[0008] Optionally, a mean filtering algorithm is performed on the target BCG frequency domain signal to obtain a mean value of the target BCG frequency domain signal within the current sliding time window; and a calculation formula for obtaining the first attenuation coefficient is as follows based on the target BCG frequency domain signal mean value and the target noise reference frequency domain signal:

[0009]

[0010] Among them, P noise (t2,f) represents the target noise reference frequency domain signal at time t2 in the current sliding time window, Indicates the mean value of the target BCG frequency domain signal from time t1 to time t2 within the current sliding time window, where time t1 is the start time of the current sliding time window and time t2 is the end time of the current sliding time window.

[0011] Optionally, according to the target BCG frequency domain signal and the target noise reference frequency domain signal, a second attenuation coefficient within the current sliding time window is obtained, including: according to the target BCG frequency domain signal, obtaining a clean signal output result within the current sliding time window; according to the clean signal output result and the target noise reference frequency domain signal, obtaining the second attenuation coefficient within the current sliding time window.

[0012] Optionally, a mean filtering algorithm is performed on the target BCG frequency domain signal to obtain a clean signal mean within the current sliding time window; and a calculation formula for obtaining the second attenuation coefficient is as follows based on the clean signal mean and the target noise reference frequency domain signal:

[0013]

[0014] in, represents the mean value of the clean signal from t1 to t2 in the current sliding time window, P noise (t2,f) represents the target noise reference frequency domain signal at time t2 in the current sliding time window.

[0015] Optionally, the initial BCG signal and the initial noise reference signal collected within the current sliding time window are respectively transformed in the frequency domain to obtain a target BCG frequency domain signal and a target noise reference frequency domain signal, including: performing scalar summation on the initial noise reference signal collected within the current sliding time window to obtain a target noise reference signal; performing frequency domain transform on the initial BCG signal and the target noise reference signal to obtain a BCG frequency domain signal and a noise reference frequency domain signal; and performing normalization on the BCG frequency domain signal and the noise reference frequency domain signal to obtain a target BCG frequency domain signal and a target noise reference frequency domain signal.

[0016] Optionally, when the initial noise reference signal includes an initial acceleration signal, motion state detection is performed on the initial noise reference signal to obtain a motion state detection result, including: obtaining an acceleration change according to the initial acceleration signal; and obtaining the motion state detection result based on the comparison results of the acceleration change with a first threshold and a second threshold respectively.

[0017] In the second aspect, the present application provides a heart rate monitoring device based on BCG signals, and the monitoring device includes: a frequency domain transformation module, which is used to perform frequency domain transformation on the initial BCG signal and the initial noise reference signal collected in the current sliding time window, respectively, to obtain a target BCG frequency domain signal and a target noise reference frequency domain signal; an attenuation coefficient acquisition module, which is used to obtain a first attenuation coefficient and a second attenuation coefficient in the current sliding time window according to the target BCG frequency domain signal and the target noise reference frequency domain signal; a spectrum signal acquisition module, which is used to obtain a target spectrum signal according to the first attenuation coefficient and the second attenuation coefficient; a motion state detection module, which is used to perform motion state detection on the initial noise reference signal to obtain a motion state detection result; and a fusion processing module, which is used to perform fusion processing on the target spectrum signal according to the motion state detection result to obtain a target heart rate monitoring value.

[0018] In a third aspect, the present application provides a heart rate monitoring system based on BCG signals, the monitoring system comprising a sensor for collecting BCG signals, a collection device for collecting noise reference signals, and the heart rate monitoring device based on BCG signals.

[0019] In a fourth aspect, the present application provides a smart mattress, which includes the heart rate monitoring system based on BCG signals.

[0020] Compared with the existing technology, this application has the following beneficial effects:

[0021] 1. This application uses a noise reference signal to assist in noise reduction of the BCG signal, which can accurately calculate the changes in human heart rate in some complex usage environments and improve the accuracy of heart rate calculation.

[0022] 2. This application obtains two attenuation coefficients of the clean signal in different ways, and performs a fusion operation on the two attenuation coefficients to obtain the target spectrum signal, thereby reducing the sampling error of the BCG signal and further improving the heart rate calculation accuracy.

[0023] 3. The present application uses a noise reference signal to detect the motion state of the human body, which is used to determine whether the source of the initial BCG signal is a static weight or the user. It can determine whether the user is in a quiet state or in motion, and can also determine the intensity of the user's motion. The different motion state detection results are fused with the heart rate value currently extracted from the target spectrum signal to finally obtain the target heart rate monitoring value of the detection target within the current sliding time window. This not only improves the accuracy of the heart rate monitoring value, but also avoids monitoring misjudgments caused by static weights. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 FIG2 is a flow chart of a heart rate monitoring method based on BCG signals provided in an embodiment of the present application;

[0025] Figure 2 Shown Figure 1 Specific flow diagram of step S102;

[0026] Figure 3 FIG2 is a flow chart of another heart rate monitoring method based on BCG signals provided in an embodiment of the present application;

[0027] Figure 4 FIG2 is a schematic structural diagram of a heart rate monitoring device based on BCG signals provided in an embodiment of the present application;

[0028] Figure 5 FIG2 is a schematic diagram showing an application scenario of a heart rate monitoring system based on BCG signals on a smart mattress provided by an embodiment of the present application;

[0029] Figure 6 The figure shows a schematic diagram of the accuracy of a heart rate monitoring method based on BCG signals provided in an embodiment of the present application. DETAILED DESCRIPTION

[0030] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0031] In a first aspect, the present invention provides a heart rate monitoring method based on BCG signals, which specifically includes the following embodiments:

[0032] Figure 1 FIG. 1 is a flow chart of a heart rate monitoring method based on BCG signals provided in an embodiment of the present application. Figure 1 The monitoring method specifically comprises the following steps:

[0033] Step S101 : performing frequency domain transformation on the initial BCG signal and the initial noise reference signal collected in the current sliding time window to obtain a target BCG frequency domain signal and a target noise reference frequency domain signal.

[0034] In this embodiment, the initial BCG signal and the initial noise reference signal collected within the current sliding time window are respectively frequency-domain transformed to obtain a target BCG frequency-domain signal and a target noise reference frequency-domain signal, including: performing scalar summation on the initial noise reference signal collected within the current sliding time window to obtain a target noise reference signal; performing frequency-domain transform on the initial BCG signal and the target noise reference signal to obtain a BCG frequency-domain signal and a noise reference frequency-domain signal; and performing normalization on the BCG frequency-domain signal and the noise reference frequency-domain signal to obtain a target BCG frequency-domain signal and a target noise reference frequency-domain signal.

[0035] It should be noted that the initial noise reference signal includes a displacement signal, a velocity signal and an initial acceleration signal. The initial acceleration signal is an initial n-axis acceleration signal, where n is a positive integer not less than 1; that is, the initial n-axis acceleration signal includes an initial uniaxial acceleration signal, an initial biaxial acceleration signal, an initial triaxial acceleration signal, an initial six-axis acceleration signal, etc.

[0036] In another embodiment, the signal acquisition front end collects the initial BCG signal and the initial n-axis acceleration signal, the sampling frequency is not less than 25 Hz, the current sliding time window length is 8s, and the step length is 2s, wherein the 8s data processed in each run includes 6s historical data and 2s current data.

[0037] The initial BCG signal in the current time window and the data of each axis in the initial n-axis acceleration signal are bandpass filtered within a range of 0.6 to 2.4 Hz, corresponding to a human heart rate of 36 to 144 beats per minute. To simplify the data processing process and avoid the problem of losing some motion information by using only data from a certain axis, the target acceleration signal is obtained by applying specific filtering methods to the data of each axis, including mean filtering and median filtering.

[0038] Specifically, when the initial noise reference signal is the initial three-axis acceleration signal, the filtered x-axis, y-axis, and z-axis data are scalar summed and averaged according to Formula 1 to obtain the target acceleration signal:

[0039]

[0040] Among them, A CC Represents the target acceleration signal, A CCx Represents the x-axis acceleration signal, A CCy Represents the y-axis acceleration signal, A CCz represents the z-axis acceleration signal; in this embodiment, the target acceleration signal is used as the target noise reference signal.

[0041] In this embodiment, the filtered initial BCG signal and the target noise reference signal are subjected to frequency domain transformation respectively. The time domain signal can be converted into a frequency domain signal by using Fourier transform, discrete cosine transform, Laplace transform, etc., and then the BCG frequency domain signal BCG obtained after the transformation is converted into a frequency domain signal. fft Sum noise reference frequency domain signal ACC fft Normalize the target BCG frequency domain signal and the target noise reference frequency domain signal according to Formula 2:

[0042]

[0043] Among them, N is the normalized data, f is the BCG to be processed fft or ACC fft data, abs(f) represents the BCG to be processed fft or ACC fft The absolute value of each data in the data, argmax(abs(f)) represents the BCG to be processed fft or ACC fft The maximum absolute value of all data in the data.

[0044] Step S102: Acquire a first attenuation coefficient and a second attenuation coefficient within a current sliding time window according to the target BCG frequency domain signal and the target noise reference frequency domain signal.

[0045] like Figure 2 As shown, in this embodiment, obtaining the first attenuation coefficient and the second attenuation coefficient in the current sliding time window according to the target BCG frequency domain signal and the target noise reference frequency domain signal specifically includes the following steps:

[0046] Step S201 : acquiring a target BCG frequency domain signal output result and a clean signal output result within a current sliding time window according to the target BCG frequency domain signal.

[0047] Step S202 : obtaining a first attenuation coefficient within a current sliding time window according to the target BCG frequency domain signal output result and the target noise reference frequency domain signal.

[0048] Step S203 : obtaining a second attenuation coefficient within the current sliding time window according to the clean signal output result and the target noise reference frequency domain signal.

[0049] It should be noted that the target BCG frequency domain signal can be processed by a mean filtering algorithm, a median filtering algorithm or an arbitrary weight matrix weighted filtering algorithm to obtain the corresponding target BCG frequency domain signal output result and the clean signal output result.

[0050] In this embodiment, the initial noise reference signal is used as the initial acceleration signal, and the target noise reference signal is used as the target acceleration frequency domain signal. noise (f) As a noise reference, filter out P signal (f) motion noise, we get P clean (f). For non-periodic random noise, the action of the Wiener filter is shown in Formula 3:

[0051]

[0052] Assuming that the signal frequency received by the sensor is composed of the target BCG frequency domain signal frequency and the noise signal frequency, an attenuation coefficient W(f) for the clean signal is obtained by formula 3, where P noise (f) is the noise reference, which can be estimated based on the target acceleration frequency domain signal. signal (f) is the target BCG frequency domain signal, the clean signal P clear (f) can be iterated from the historical output of the Wiener filter, or it can be obtained from the P in the current filter. signal (f) Subtract P noise (f)Get.

[0053] In order to enhance the noise filtering effect, this embodiment uses two acquisition methods to obtain P clear(f), thus obtaining two attenuation coefficients of the clean signal. clear (f) = P signal (f)-P noise (f) and Formula 3 show that, taking the target BCG frequency domain signal output result as the target BCG frequency domain signal mean as an example, the calculation formula of the first attenuation coefficient W1 is shown in Formula 4:

[0054]

[0055] Among them, P noise (t2,f) represents the target noise reference frequency domain signal at time t2 in the current sliding time window, represents the mean value of the target BCG frequency domain signal from time t1 to t2 in the current sliding time window (that is, obtained by performing mean filtering algorithm on the target BCG frequency domain signal), where time t1 is the starting time of the current sliding time window and time t2 is the ending time of the current sliding time window. Formula (4) calculates the process of the first attenuation coefficient W1 changing with the signal frequency at time t2.

[0056] In another embodiment, a set of weight coefficients [α1, α2, α3, ..., α n ], where the middle weight coefficient ratio is high and the weight coefficient ratio on both sides is low, α1+α2+α3+...α n =1, a weighted average filtering process is performed on the target BCG frequency domain signal from time t1 to t2 in the current sliding time window of the corresponding moment through this group of weight coefficients to obtain the weighted mean of the target BCG frequency domain signal instead of the target BCG frequency domain signal mean in formula (4) for calculation, thereby obtaining the first attenuation coefficient W1.

[0057] In another embodiment, the target BCG frequency domain signal from time t1 to t2 in the current sliding time window is subjected to median filtering to obtain the median of the target BCG frequency domain signal. The median filtering process can effectively eliminate the sudden change interference, and the target BCG frequency domain signal mean in formula (4) is replaced by the median of the target BCG frequency domain signal to perform the operation, thereby obtaining the first attenuation coefficient W1.

[0058] It should be noted that since the attenuation coefficient W1 calculated by formula (4) may be a negative value, it is necessary to judge the calculated W1 value in the actual algorithm processing process. If it is a positive value, it is used normally; if it is a negative value, the W1 calculation result is discarded, and only the second attenuation coefficient W2 is used in subsequent processing.

[0059] In another embodiment, taking the clean signal output result as the clean signal mean as an example, the calculation formula of the second attenuation coefficient W2 can be obtained according to the historical output iteration process of the filter and Formula 3 as shown in Formula (5):

[0060]

[0061] According to formula 3, P clear (f)=W(f)((P clear (f)+P noise (f)), deduce P clear (f)=W(f)P signal (f), so the numerator in Eq. 5 is In fact, it is the mean value of the clean signal output by Wiener filter 2 from time t1 to t2 in the current sliding time window. This formula reflects the characteristic that the second attenuation coefficient W2 is iteratively solved according to the historical output. Since no historical data is generated when the algorithm is initialized, the historical attenuation coefficient W2 cannot be obtained. Therefore, when the algorithm is initialized, the attenuation coefficient W2 must be preset first. The initial attenuation coefficient W2 can be set to 2 to facilitate the operation of the algorithm.

[0062] In other embodiments, the clean signal mean in the above formula 5 can be replaced by the clean signal weighted mean or the clean signal median, etc., and the specific processing and acquisition method is similar to the above-mentioned target BCG frequency domain signal weighted mean or the target BCG frequency domain signal median.

[0063] In another embodiment, the time t1 to t2 is not limited to the current sliding time window, and t1 only needs to be a historical time and t2 needs to be the current time.

[0064] Step S103: Obtain a target spectrum signal according to the first attenuation coefficient and the second attenuation coefficient.

[0065] In this embodiment, according to the first attenuation coefficient and the second attenuation coefficient, a calculation formula for obtaining the target spectrum signal is:

[0066]

[0067] Step S104: performing motion state detection on the initial noise reference signal to obtain a motion state detection result.

[0068] In this embodiment, when the initial noise reference signal includes an initial acceleration signal, motion state detection is performed on the initial noise reference signal to obtain a motion state detection result, including: obtaining the acceleration change in each axis according to the initial acceleration signal; and obtaining the motion state detection result according to the comparison results of the acceleration change in each axis with the first threshold and the second threshold.

[0069] It should be noted that the acceleration of each axis is calculated based on the initial n-axis acceleration signal and the change in acceleration is used to assist in motion state detection. The first threshold is set to α1 and the second threshold is set to α2 (α2>α1). The motion state detection rule is:

[0070] (1) If the change in a certain axial acceleration of the acceleration signal processed within the current sliding time window is greater than or equal to the second threshold value α2, it is judged that the initial BCG signal collected at this time is subject to large interference noise, and the motion state detection result is that the human body is in a dynamic state; (2) If the change in the three axial accelerations of the acceleration signal processed within the current sliding time window is less than the second threshold value α2 but greater than or equal to the first threshold value α1, it is judged that the initial BCG signal collected at this time is less subject to noise interference, and the motion state detection result is that the human body is in a static state; (3) If the change in the three axial accelerations of the acceleration signal processed within the current sliding time window is less than the first threshold value α1, it is judged that the BCG signal value collected at this time may be caused by the weight of a static heavy object and is not generated by a living body, and the motion state detection result is that there is no person.

[0071] Step S105: performing fusion processing on the target spectrum signal according to the motion state detection result to obtain a target heart rate monitoring value.

[0072] It should be noted that if Figure 3 As shown, peak tracking is performed on the clean target spectrum signal, and the frequency corresponding to the spectrum peak of the frequency domain data within the current sliding time window is used as the heart rate frequency. During the peak tracking process, since this embodiment is based on a 2-second time window step, the heart rate can be calculated every 2 seconds. During this 2-second processing, the heart rate will not fluctuate significantly. Therefore, during peak tracking, the heart rate frequency corresponding to the spectrum peak within the current time window should be based on the historical heart rate frequency and cannot deviate from the historical heart rate by more than 10 bpm.

[0073] In this embodiment, the fusion processing process includes: optimizing the accuracy of the target heart rate detection value based on the motion state detection result determined by the acceleration signal change. When the motion state detection result is that the human body is in motion, the BCG signal at this time is strongly interfered by motion noise, and the current heart rate monitoring value obtained by the BCG signal at this time cannot be considered a reliable heart rate value. M historical heart rates and the current heart rate monitoring value are used for smoothing filtering to reduce the variability of the current unreliable heart rate, and the heart rate value obtained after processing is used as the current target heart rate monitoring value; when the motion state detection result is that there is no one (meaning there is a static heavy object pressing), even if the BCG sensor has a signal, the heart rate is no longer calculated based on this signal, but the judgment value of the unmanned state is output as the current target heart rate monitoring value; further, M is dynamically adjusted with the mean square error σ of the three-axis (i.e., x-axis, y-axis, z-axis) scalar sum of the current three-axis acceleration signal and the offset of the second threshold α2. The specific calculation formula is as follows:

[0074]

[0075] Wherein, M is a positive integer, ranging from 0≤M≤5; is the ceiling function, when When M is 1, the human body is in a static state. At this time, the BCG signal is less affected by motion noise. The heart rate calculated based on the BCG signal at this moment is more accurate. The historical heart rate and the current heart rate do not need to be used for smoothing filtering. That is, M takes a value of 0, and the current heart rate value can be directly used as the current target heart rate monitoring value. When the human body is in a dynamic state but the movement is relatively mild, the value of M increases with the increase of the human body's movement intensity, and its value range is 0<M<5; when When the human body is in a dynamic state and the movement is relatively intense, the BCG signal is too strongly disturbed by the movement noise and the heart rate variability is too large. If M takes too many values, it may cause excessive noise and the calculated heart rate accuracy is low.

[0076] In this embodiment, the motion state detection includes in-position detection and motion intensity detection. The in-position detection can determine whether the source of the initial BCG signal is a static weight or the user, while the motion intensity detection can determine whether the user is in a quiet state or a motion state while in position, and can also determine the intensity of the user's in-position motion.

[0077] In another embodiment, when the initial noise reference signal is an initial three-axis acceleration signal, the first threshold α1 and the second threshold α2 can be obtained by the following calibration method:

[0078] The initial three-axis acceleration signals are collected within 5 minutes of the horizontal static state. According to the rule of 8-second time window and 2-second step length, the mean square error of the acceleration signals of the three axes (x-axis, y-axis, and z-axis) in each time window is calculated in sequence. The maximum mean square error and the minimum mean square error in the three axes are obtained based on the obtained mean square error. The first threshold α1 is further obtained by formula 8:

[0079] α1=max(x max_std -x min_std ,y max_std -y min_std ,z max_std -z min_std ) (8)

[0081] Among them, x max_std 、y max_std 、z max_std They are the maximum mean square error in the x-axis, y-axis, and z-axis directions, respectively. min_std 、y min_std 、z min_std are the minimum mean square errors in the x-axis, y-axis, and z-axis directions respectively.

[0082] When calibrating the second threshold α2, it is necessary to collect the initial three-axis acceleration signals within 5 minutes when the human body turns over, shakes legs, etc. The specific calibration method is the same as the calibration method of the first threshold α1 mentioned above.

[0083] It should be noted that when the initial noise reference signal is the initial n-axis acceleration signal, when calibrating the first threshold α1 and the second threshold α2, it is only necessary to calculate the mean square error of the acceleration signals in the n axes in each time window in turn, and then obtain the maximum mean square error and the minimum mean square error in the n axes according to the obtained mean square error. The specific processing method is similar to the above processing method and will not be elaborated here.

[0084] In another embodiment, the first threshold α1 and the second threshold α2 may also be obtained by the following calibration method:

[0085] A comparison of the mean square error values ​​within the sliding time window of the accelerometer in the state where a static weight presses the accelerometer and the state where the accelerometer is left alone shows no significant difference. Therefore, the acceleration signal of the accelerometer when it is left alone can be directly calibrated as the first threshold α1; when the user is in a state of strenuous exercise and the accelerometer acquisition signal has a large offset, the second threshold α2 is calibrated.

[0086] In another embodiment, if Figure 6As shown, the real heart rate used for accuracy comparison is collected by the Polar heart rate belt worn by the tested person during the test, and the qualification standard for each set of test heart rate is whether the test heart rate is within the range of ±10% of the real heart rate. Based on this, it can be seen that the overall accuracy of the test heart rate monitoring values ​​obtained by this application can reach more than 98%.

[0087] Compared with the existing technology, this application has the following beneficial effects:

[0088] 1. This application uses the acceleration signal as a noise reference to assist in noise reduction of the BCG signal, which can accurately calculate the changes in human heart rate in some complex usage environments and improve the accuracy of heart rate calculation.

[0089] 2. This application obtains two attenuation coefficients of the clean signal in different ways, and performs a fusion operation on the two attenuation coefficients to obtain the target spectrum signal, thereby reducing the sampling error of the BCG signal and further improving the heart rate calculation accuracy.

[0090] 3. This application uses acceleration signals to detect the motion state of the human body, which is used to determine whether the source of the initial BCG signal is a static weight or the user. It can also determine whether the user is in a quiet state or in motion. The different motion state detection results are fused with the heart rate value currently extracted from the target spectrum signal to finally obtain the target heart rate monitoring value of the detection target within the current sliding time window. This not only improves the accuracy of the heart rate monitoring value, but also avoids monitoring errors caused by static weights.

[0091] In a second aspect, the present invention provides a heart rate monitoring device based on BCG signals, specifically including the following embodiments:

[0092] Figure 4 FIG. 1 is a structural diagram of a heart rate monitoring device based on BCG signals provided in an embodiment of the present application. Figure 4 As shown, the monitoring device includes:

[0093] The frequency domain transformation module 110 is used to perform frequency domain transformation on the initial BCG signal and the initial noise reference signal collected in the current sliding time window to obtain a target BCG frequency domain signal and a target noise reference frequency domain signal;

[0094] An attenuation coefficient acquisition module 120 is configured to acquire a first attenuation coefficient and a second attenuation coefficient within a current sliding time window according to the target BCG frequency domain signal and the target noise reference frequency domain signal;

[0095] A spectrum signal acquisition module 130, configured to obtain a target spectrum signal 140 according to the first attenuation coefficient and the second attenuation coefficient;

[0096] A motion state detection module 150 is configured to perform motion state detection on the initial noise reference signal to obtain a motion state detection result;

[0097] The fusion processing module 160 is configured to perform fusion processing on the target spectrum signal according to the motion state detection result to obtain a target heart rate monitoring value.

[0098] In a third aspect, the present invention provides a heart rate monitoring system based on BCG signals, the monitoring system comprising a sensor for collecting BCG signals, a collection device for collecting noise reference signals, and the above-mentioned heart rate monitoring device based on BCG signals.

[0099] It should be noted that the sensor of the BCG signal includes two piezoelectric sensors and a pressure-sensitive sensor arranged between the two piezoelectric sensors; when the noise reference signal is an acceleration signal, its corresponding acquisition device includes an accelerometer and / or a gyroscope.

[0100] In a fourth aspect, the present invention provides a smart mattress, which includes the above-mentioned BCG signal-based heart rate monitoring system.

[0101] It should be noted that if Figure 5 As shown, the smart mattress includes a mattress body 10, a sponge pad 20, and a heart rate monitoring system arranged between the mattress body 10 and the sponge pad 20, wherein the heart rate monitoring system includes a sensor 31, an acquisition device 32, a data connection line 33 and a main control box 34. In this embodiment, the acquisition device adopts an accelerometer. When the user uses the smart mattress normally, in order to ensure that the BCG signal detection area covers the user's chest area and more accurately receive the human body's BCG signal, the sensor and accelerometer are arranged at one-third of the mattress length; the main control box contains a micro control unit, a battery, a Bluetooth module, etc. The micro control unit is embedded with the functional module of the monitoring device, which can adjust the output heart rate in real time according to the BCG signal and acceleration signal transmitted by the data connection line, and the output heart rate is sent out by the Bluetooth module.

[0102] Among them, the above-mentioned heart rate monitoring system based on BCG signals is not limited to smart mattresses, but can also be applied to traditional smart home environments, such as embedded layout in smart round stools, smart chairs, sofas, etc. When the user shakes his legs, twists his legs, etc. on the round stools and chairs, it can effectively remove noise and ensure the accuracy of the monitored heart rate value. The present invention realizes non-perceptible heart rate monitoring in smart home devices and can improve the accuracy of heart rate monitoring values.

[0103] Finally, it should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus comprising the element.

[0104] The above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the purpose and scope of the technical solutions of the present invention, and all of these should be included in the scope of the claims of the present invention.

Claims

1. A heart rate monitoring method based on BCG signal, characterized in that: The monitoring method comprises: Perform frequency domain transformation on the initial BCG signal and initial noise reference signal collected in the current sliding time window to obtain the target BCG frequency domain signal and target noise reference frequency domain signal; Acquire a first attenuation coefficient and a second attenuation coefficient within a current sliding time window according to the target BCG frequency domain signal and the target noise reference frequency domain signal; Obtaining a target spectrum signal according to the first attenuation coefficient and the second attenuation coefficient; Performing motion state detection on the initial noise reference signal to obtain a motion state detection result; According to the motion state detection result, the target spectrum signal is fused to obtain a target heart rate monitoring value; Obtaining the first attenuation coefficient within the current sliding time window includes: According to the target BCG frequency domain signal, an output result of the target BCG frequency domain signal within the current sliding time window is obtained; Performing mean filtering on the target BCG frequency domain signal to obtain the mean of the target BCG frequency domain signal within the current sliding time window; The first attenuation coefficient in the current sliding time window is obtained according to the BCG frequency domain signal mean and the target noise reference frequency domain signal. The calculation formula of the first attenuation coefficient is: Among them, P noise (t2,f) represents the target noise reference frequency domain signal at time t2 in the current sliding time window, represents the mean value of the target BCG frequency domain signal from time t1 to time t2 within the current sliding time window, where time t1 is the start time of the current sliding time window and time t2 is the end time of the current sliding time window; Obtaining the second attenuation coefficient within the current sliding time window includes: Obtaining a clean signal output result within the current sliding time window according to the target BCG frequency domain signal; Performing mean filtering on the target BCG frequency domain signal to obtain the clean signal mean within the current sliding time window; The second attenuation coefficient in the current sliding time window is obtained according to the clean signal mean and the target noise reference frequency domain signal. The calculation formula of the second attenuation coefficient is: in, represents the mean value of the clean signal from time t1 to time t2 in the current sliding time window, P noise (t2, f) represents the target noise reference frequency domain signal at time t2 in the current sliding time window.

2. The heart rate monitoring method based on BCG signal according to claim 1, characterized in that: Perform frequency domain transformation on the initial BCG signal and the initial noise reference signal collected in the current sliding time window to obtain the target BCG frequency domain signal and the target noise reference frequency domain signal, including: Performing scalar summation on the initial noise reference signals collected within the current sliding time window to obtain a target noise reference signal; Performing frequency domain transformation on the initial BCG signal and the target noise reference signal respectively to obtain a BCG frequency domain signal and a noise reference frequency domain signal; Normalization processing is performed on the BCG frequency domain signal and the noise reference frequency domain signal respectively to obtain the target BCG frequency domain signal and the target noise reference frequency domain signal.

3. The heart rate monitoring method based on BCG signal according to claim 1, characterized in that: When the initial noise reference signal includes an initial acceleration signal, performing motion state detection on the initial noise reference signal to obtain a motion state detection result includes: Acquiring an acceleration change according to the initial acceleration signal; The motion state detection result is obtained according to the comparison results of the acceleration change with the first threshold and the second threshold respectively.

4. A heart rate monitoring device based on BCG signals, characterized in that: The heart rate monitoring method based on BCG signals according to any one of claims 1 to 3 is used, wherein the monitoring device comprises: The frequency domain transformation module is used to perform frequency domain transformation on the initial BCG signal and the initial noise reference signal collected in the current sliding time window to obtain the target BCG frequency domain signal and the target noise reference frequency domain signal; an attenuation coefficient acquisition module, configured to acquire a first attenuation coefficient and a second attenuation coefficient within a current sliding time window according to the target BCG frequency domain signal and the target noise reference frequency domain signal; a spectrum signal acquisition module, configured to obtain a target spectrum signal according to the first attenuation coefficient and the second attenuation coefficient; A motion state detection module, configured to perform motion state detection on the initial noise reference signal to obtain a motion state detection result; The fusion processing module is used to perform fusion processing on the target spectrum signal according to the motion state detection result to obtain a target heart rate monitoring value.

5. A heart rate monitoring system based on BCG signals, characterized in that: The monitoring system includes a sensor for collecting BCG signals, a collection device for collecting noise reference signals, and the BCG signal-based heart rate monitoring device according to claim 4.

6. A smart mattress, characterized in that: The smart mattress includes the BCG signal-based heart rate monitoring system according to claim 5.

Images