CN111329462A - Real-time unbound heart rate extraction method - Google Patents
Real-time unbound heart rate extraction method Download PDFInfo
- Publication number
- CN111329462A CN111329462A CN202010145370.XA CN202010145370A CN111329462A CN 111329462 A CN111329462 A CN 111329462A CN 202010145370 A CN202010145370 A CN 202010145370A CN 111329462 A CN111329462 A CN 111329462A
- Authority
- CN
- China
- Prior art keywords
- data
- heart
- heartbeat
- signal
- piezoelectric film
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/02—Detecting, measuring or recording pulse, heart rate, blood pressure or blood flow; Combined pulse/heart-rate/blood pressure determination; Evaluating a cardiovascular condition not otherwise provided for, e.g. using combinations of techniques provided for in this group with electrocardiography or electroauscultation; Heart catheters for measuring blood pressure
- A61B5/024—Detecting, measuring or recording pulse rate or heart rate
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/103—Detecting, measuring or recording devices for testing the shape, pattern, colour, size or movement of the body or parts thereof, for diagnostic purposes
- A61B5/11—Measuring movement of the entire body or parts thereof, e.g. head or hand tremor, mobility of a limb
- A61B5/1102—Ballistocardiography
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7203—Signal processing specially adapted for physiological signals or for diagnostic purposes for noise prevention, reduction or removal
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7225—Details of analog processing, e.g. isolation amplifier, gain or sensitivity adjustment, filtering, baseline or drift compensation
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Biomedical Technology (AREA)
- Medical Informatics (AREA)
- Veterinary Medicine (AREA)
- Physiology (AREA)
- Public Health (AREA)
- Physics & Mathematics (AREA)
- Signal Processing (AREA)
- Biophysics (AREA)
- Pathology (AREA)
- General Health & Medical Sciences (AREA)
- Heart & Thoracic Surgery (AREA)
- Animal Behavior & Ethology (AREA)
- Molecular Biology (AREA)
- Surgery (AREA)
- Cardiology (AREA)
- Artificial Intelligence (AREA)
- Psychiatry (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Power Engineering (AREA)
- Dentistry (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Measuring Pulse, Heart Rate, Blood Pressure Or Blood Flow (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
Abstract
The invention relates to a real-time unbound heart rate extraction method, which utilizes a flexible piezoelectric film sensor to collect a heart impact signal of a human body, and utilizes the flexible piezoelectric film sensor fixed on the surface of a seat to collect a piezoelectric film vibration signal caused by the heart impact phenomenon, then the piezoelectric film vibration signal is processed by a charge amplifier and is transmitted to an upper computer for processing and analysis through data collection equipment, and finally the heart rate and the heart beat period are extracted. The method can automatically detect whether the human body leaves the chair or not, and can continuously monitor the physiological information of the human body such as the heart rate and the heartbeat cycle under the human body sitting posture. The method does not need a large amount of priori knowledge and a large amount of sample training, can obtain the heartbeat template by combining period estimation and interval division, has relatively simple calculation and low requirement on the measurement environment, can obtain successive heartbeat intervals, and has higher detection precision and stronger robustness.
Description
Technical Field
The invention relates to the field of unbound physiological information monitoring, in particular to a real-time unbound heart rate extraction method.
Background
The sedentariness is the body posture which needs to be kept for a long time by office people, student groups and drivers at present, the sedentariness is accompanied with the static state of the body, so the demand of the body on the heart is reduced, the heart function can be declined by the sedentariness for the sedentariness, heart diseases such as arrhythmia and coronary heart disease are caused, and the sedentariness heart rate monitoring has important significance for sitting posture. The method for extracting the heart rate by adopting the heart attack signal in the existing unrestrained heart rate extraction research is wide, the heart attack means that the body generates vibration in the three-dimensional direction due to the flow of blood in the heart blood pumping process, according to the Newton's third law, the body can generate reaction force with the same size and the opposite direction, and the force contains information such as heart rate respiration.
In the existing study on heart rate extraction based on the cardiac shock signal in a sitting posture, a commonly used method is to perform spectrum analysis on the cardiac shock signal, change the signal from a time domain to a frequency domain through Fourier transform and the like, and extract heart rate features from the frequency domain, but the frequency distribution of the cardiac shock signal is easily overlapped by the frequency of a noise signal, and a successive heartbeat interval cannot be obtained at the same time. In addition, for example, a method for automatically detecting abnormal heart rate based on the cardiac shock signal in Jiang Fang, Wang Xu, Yanbo, et al (J. the school news of northeast university, Nature science edition), 2010,31(12): 1685-. Similar to the method, a heart rate feature extraction method based on machine learning is adopted, and the heart attack signals are subjected to feature training through a support vector machine, a convolutional neural network and the like, but the method needs a large number of samples for learning and training, and the calculation process is complicated. Therefore, a new heart rate extraction method for the sitting posture ballistocardiogram signal is urgently needed, and the invention is a system and a method for solving the need.
Disclosure of Invention
Aiming at the defects and shortcomings of the technologies, the invention provides a real-time unbound heart rate extraction method, particularly a sitting posture-oriented method, and aims to improve the heart rate detection precision of the sitting posture-oriented method. The method can automatically detect whether the human body leaves the chair or not, and can continuously monitor the physiological information of the human body such as the heart rate and the heartbeat cycle under the human body sitting posture. The method does not need a large amount of priori knowledge and a large amount of sample training, can obtain the heartbeat template by combining period estimation and interval division, has relatively simple calculation and low requirement on the measurement environment, can obtain successive heartbeat intervals, and has higher detection precision and stronger robustness. The invention provides important technical support for the monitoring of the unbounded physiological information facing to the sitting posture, and has important significance for preventing sedentariness, cardiovascular diseases and the like.
The technical scheme adopted by the invention for solving the problems is as follows:
a real-time unrestrained heart rate extraction method utilizes a flexible piezoelectric film sensor to collect a heart impact signal of a human body, and comprises the following steps:
the method comprises the following steps of firstly, determining whether a human body is located at a position to be detected within a continuous period of a seconds and is in a static state, if the condition is met, preprocessing data of the continuous period of a seconds, removing baseline drift and high-frequency noise, and obtaining a preprocessed heart impact signal:
second, estimating the heart cycle
Extracting all positive peaks in a-second data of the preprocessed heart shock signals by using a peak detection algorithm, recording the time axis position, respectively calculating an autocorrelation function R and a short-time average amplitude difference function D of each time axis position in 1.5-a-1.5-second data according to a formula (1) and a formula (2), and obtaining a core function Q according to a formula (3) by using the autocorrelation function and the short-time average amplitude difference function, wherein the core function Q is expressed as a peak value at the heart beat cycle position;
detecting a positive wave peak of a core function, setting a wave peak threshold value delta, taking the position of the positive wave peak with amplitude exceeding delta as a heartbeat cycle position, and calculating the average interval of adjacent heartbeat cycle positions as an estimated heartbeat cycle at a certain positive wave peak;
calculating estimated heartbeat cycles of all positive wave crests in the data of 1.5-a-1.5 seconds to form a heartbeat cycle sequence, and calculating an arithmetic mean value to serve as a heartbeat cycle estimated value T in the data of a seconds;
wherein x is the preprocessed ballistocardiogram signal; m is the time axis position of the positive wave crest in 1.5-a-1.5 seconds of data, and the value range of k is 0-N; n represents the total length of time intercepted at two sides of the time axis position of the positive wave peak; the value range of N is [0.8f,3f ], and f is the signal sampling frequency;
thirdly, dividing the partitions:
dividing the preprocessed heart impact signal into continuous intervals, wherein the interval length is the heartbeat period estimated value T obtained in the second step; judging the validity of the interval, if the maximum amplitude of the interval exceeds the set threshold range [2e-4,5e-4]If the interval is invalid, the heart attack signal is interfered by the outside; if the maximum amplitude is within the threshold range, the interval is an effective interval; positioning the time axis position of the maximum amplitude from the effective interval as the time axis position of the main characteristic maximum J peak of the ballistocardiographic signal in the a second data;
the fourth step: heart beat template
Taking the time axis position of the maximum J peak of the main characteristics of the obtained heart impact signal in the data of a second as the center, acquiring data b with the data length of T.f/2 forwards and backwards at the time axis position of the maximum J peak, forming data c by the center and the data b, and calculating the arithmetic mean value of the data c as the heart beat template of the data of the current a second;
every time continuous a second preprocessed heart impact signal data is obtained, recalculating a heartbeat template corresponding to the next a second data;
and a sixth step: location of heart beat
Calculating a correlation coefficient r between each a-second heart attack signal and a heart beat template thereof according to a formula (5), and positioning the heart beat position of the current heart attack signal according to the correlation coefficient; dividing the number of the heartbeat positions by a to obtain the heart rate; calculating the difference of the adjacent heartbeat positions to obtain the successive heartbeat interval
Wherein x isi，yiThe preprocessed heart impact signal is obtained;
is a heartbeat template signal; and M is the length of the preprocessed ballistocardiogram signal.
The method is particularly suitable for extracting the heart rate in a sitting posture, the flexible piezoelectric film sensor is placed on the surface of the seat in the sitting posture, and the specific process of the first step is as follows:
the upper computer collects the charge quantity generated by the flexible piezoelectric film sensor in real time through data collection equipment, and the signal sampling frequency is set to be 1000 Hz; calculating the standard deviation of data per second in real time in the acquisition process, judging the threshold value and setting the threshold value range as [2e ]-4,5e-4]Judging whether a person exists in the seat according to the standard deviation of the collected data and monitoring; if the standard deviation of the acquired data is not within the threshold range, the monitoring system does not work, because the human body is out of seat or the human body motion amplitude is large at the moment, the heart rate monitoring cannot be carried out, and the state is recorded as state 1; if the standard deviation of the collected data is within the threshold range, the situation that no person is out of the seat and the human body is in a static state is indicated, and at the moment, a-second data are continuously collected and recorded as a state 2; if the standard deviation of the data in the a second time is in the threshold range, preprocessing is carried out, otherwise, state 1 judgment is carried out, and state 2 judgment is carried out when the data is in the state 1, and continuous a second data are formed again.
A monitoring system for extracting a real-time unbound heart rate comprises a charge amplifier, data acquisition equipment and an upper computer, and is characterized by further comprising a flexible piezoelectric film sensor and a seat; the flexible piezoelectric film sensor is arranged on the surface of the seat and is connected with data acquisition equipment through a charge amplifier, the data acquisition equipment transmits data to an upper computer, and the upper computer is loaded with the real-time unbound heart rate extraction method.
Compared with the prior art, the invention has the advantages that:
(1) the invention provides a novel method capable of monitoring the heart rate in real time without binding, which does not need training and learning a large number of samples, utilizes the main characteristics of signals to establish a template and update the template, reduces the calculated amount, and ensures the accuracy of the template and the real-time performance of calculation.
(2) The method of the invention utilizes the periodic variation of the main characteristics of the cardiac shock signal, and can carry out the heartbeat period estimation on the cardiac shock signal only by adopting two types of data, namely an improved autocorrelation function and a short-time average amplitude difference function.
(3) The invention develops a novel sitting posture-oriented heart rate monitoring system, which measures the weak vibration of a piezoelectric film caused by the heartbeat through a flexible piezoelectric film sensor fixed on the surface of a seat, avoids the direct contact of the sensor and the surface of the skin, and removes the uncomfortable feeling. Under the position of sitting, wherein the amplitude of the heart impulse force parallel to the direction of the spine is the largest and is influenced by gravity to be small, the vibration of the object contacted with a person can be caused, the vibration signal of the object contacted with the person can be collected through the flexible piezoelectric film sensor, and the heart rate information under the position of sitting is extracted.
(4) In the process of establishing the heartbeat template, in the process of carrying out interval division on the preprocessed heart attack signals, the heartbeat period estimated value is adopted as the division length of the interval, the step solves the problem that the division intervals of the heart attack signals are different in length, and multiple complete heartbeats can be contained when the interval is too long; too short an interval may not contain a complete heartbeat signature. The obtained estimated heartbeat period can be adopted to ensure that each interval contains a complete heartbeat to the maximum extent, and meanwhile, the position of the J peak in the current interval is convenient to extract according to the maximum amplitude characteristic of the J peak of the main characteristic, so that a heartbeat template is established. Experiments show that compared with the electrocardiogram detection of a standard medical clinical equipment Anru N7000 polysomnography monitor, the relative error of the heart rate is not more than 2.41%, the average error of the heartbeat period is not more than 30ms, and the electrocardiogram detection device has higher accuracy and practicability and stronger interference resistance.
Drawings
FIG. 1 is a schematic diagram of a structure of an embodiment of a real-time unbound heart rate extraction monitoring system of the present invention:
FIG. 2 is a flow chart of the method of the present invention:
FIG. 3 is a schematic cross-sectional structure diagram of a flexible piezoelectric film sensor:
FIG. 4 is a waveform of the preprocessed ballistocardiogram signal:
fig. 5 is a waveform diagram of a heart cycle sequence at a certain positive peak of the kernel function Q:
FIG. 6 is a schematic diagram of division of the preprocessed cardiac shock signal into intervals:
fig. 7 is a waveform diagram of the heartbeat template:
fig. 8 is a waveform diagram of the correlation coefficient:
FIG. 9 is a diagram of the marker of all heartbeat positions of the method of the present invention:
fig. 10 is a diagram of a heartbeat location marker of the electrocardiogram module.
Detailed Description
In order to make the present invention achieve the above functions, the present invention is further described below with reference to the drawings and examples, and the embodiments of the present invention include the following examples, which should not be construed as limiting the scope of the present invention.
The invention discloses a real-time unbound heart rate extraction method, which is used for acquiring a heart impact signal of a human body in a longitudinal direction (parallel to the spine direction) under a sitting posture by utilizing a flexible piezoelectric film sensor, and mainly comprises the following steps:
step one, seat separation judgment:
the flexible piezoelectric film sensor generates weak electric charge under the action of cardiac shock, the electric charge is amplified by the charge amplifier and then collected by the data collecting equipment, and the signal sampling frequency is set to be 1000 Hz. Calculating standard deviation of real-time data per second in the acquisition process, judging threshold value and setting threshold value range as [2e ]-4,5e-4]And judging whether the seat is occupied or not according to the standard deviation of the collected data and monitoring. If the standard deviation of the acquired data is not within the threshold range, the monitoring system does not work, because the human body is out of seat or the human body motion amplitude is large at the moment, the heart rate monitoring cannot be carried out, and the state is recorded as state 1; if the standard deviation of the collected data is within the threshold range, the situation that no person is out of the seat and the human body is in a static state is indicated, and at the moment, a-second data are continuously collected and recorded as a state 2; if the standard deviation of the data in the a second time is within the range of the threshold value, entering the next step, otherwise, judging the state 1, and judging the state 2 when the data is in the state 1, and recombining the continuous a second data; and a takes a value of 10-15.
Secondly, preprocessing signals:
and preprocessing the a second data in the state 2 acquired in the first step, wherein the specific processing process comprises the steps of removing baseline drift and removing high-frequency noise. And fitting and removing the trend term by using a polynomial, then decomposing and reconstructing the signal without the trend term by adopting wavelet transform, and finally removing high-frequency noise by using a low-pass filter.
Thirdly, estimating a heartbeat cycle:
and (3) extracting all positive peaks (peaks larger than zero) in the data of a seconds from the heart shock signal preprocessed in the second step by using a peak detection algorithm, recording the time axis position, calculating an autocorrelation function R and a short-time average amplitude difference function D of each time axis position in the data of 1.5-a-1.5 seconds according to a formula (1) and a formula (2), and obtaining a core function Q according to a formula (3) by using the autocorrelation function and the short-time average amplitude difference function, wherein the core function Q is the ratio of the autocorrelation function to the short-time average amplitude difference function, and the core function Q is expressed as a peak value at the heartbeat cycle position. Detecting a positive wave crest of a core function, setting a wave crest threshold value delta, taking the position of the positive wave crest with amplitude exceeding delta as a heartbeat cycle position, and calculating the average interval of adjacent heartbeat cycle positions as an estimated heartbeat cycle at a certain positive wave crest. Calculating estimated heartbeat cycles of all positive wave crests in the data of 1.5-a-1.5 seconds to form a heartbeat cycle sequence, and calculating an arithmetic mean value to serve as a heartbeat cycle estimated value T in the data of a seconds;
wherein x is the preprocessed impact signal; m is the time axis position of the positive wave crest in 1.5-a-1.5 seconds of data, and the value range of k is 0-N; n represents the total length of time intercepted at two sides of the time axis position where the positive wave peak of the preprocessed heart impact signal is located; the value range of N is [0.8f,3f ], and f is the signal sampling frequency. For example, the sampling frequency is 1000Hz, the data acquisition device acquires 1000 data points per second, and the value range of N is 800-3000, corresponding to 0.8-3 seconds. N reflects the sum of two heart beat intervals, so the corresponding heart rate range is 40-150 times/min.
Fourthly, dividing the partitions:
and dividing the heart attack signal preprocessed in the second step into continuous intervals, wherein the interval length is the heart cycle estimated value T calculated in the third step. Judging the validity of the interval, if the maximum amplitude of the interval exceeds the set threshold range [2e ]-4,5e-4]If the interval is invalid, the interval is judged to be invalid because body movement is inevitable in the measurement process, so that the heart impact signal is interfered; if the maximum amplitude is in the threshold range, the interval is valid, and then the time axis position of the maximum amplitude is positioned from the valid interval to be used as the time axis position of the maximum J peak of the main characteristics of the heart impact signal in the a second data.
Fifth step, heartbeat mould
And taking the time axis position of the maximum J peak of the main characteristics of the heart impact signal in the data of a second obtained in the fourth step as the center, intercepting data b of which the forward data length and the backward data length are both T.f/2 from the corresponding time axis position, forming data c of which the length is T.f by the center and the data b, and calculating the arithmetic average value of the data c to be used as the heartbeat template of the current data a. Every time continuous a second preprocessed heart impact signal data is obtained, recalculating a heartbeat template corresponding to the next a second data;
sixth step, heart beat location
And (5) calculating the correlation coefficient of each a-second preprocessed heart impact signal and the heart beat template thereof according to the formula (5). And (3) carrying out correlation threshold judgment on the correlation coefficient, setting the correlation threshold to be 0.85, and if the correlation coefficient exceeds the correlation threshold, ensuring that the similarity between the heart attack signal and the heart beat template is higher and one heart beat is obtained. And detecting all the cardiac shock signals with the similarity degree exceeding the correlation threshold value, and marking the heartbeat positions of the cardiac shock signals. And dividing the number of all the heartbeat positions by a to obtain the heart rate, and calculating the average value of the intervals of all the adjacent heartbeat positions to obtain the heartbeat period.
Wherein x isi，yiThe preprocessed heart impact signal is obtained;
is a heartbeat template signal; and M is the length of the preprocessed ballistocardiogram signal.
The monitoring system (see fig. 1) of the invention applying the method comprises: the device comprises a seat 1, a flexible piezoelectric film sensor 2, a charge amplifier 3, data acquisition equipment 4 and an upper computer 5; the flexible piezoelectric film sensor is placed on the surface of a seat, a human body 6 is in close contact with the surface of the flexible piezoelectric film sensor, vibration signals of the piezoelectric film, caused by cardiac shock of the human body 6, are collected, the signals are processed by a charge amplifier and then collected by data collecting equipment, and finally transmitted to an upper computer to process data, the upper computer is communicated with a mobile terminal, and the method (see figure 2) provided by the invention is adopted to extract heart rate and heartbeat cycle and send the heart rate and heartbeat cycle to the mobile terminal equipment for checking. The mobile terminal device can be a portable intelligent device such as a smart phone, an iPad and a smart watch.
The flexible piezoelectric film sensor used in the invention is a PVDF piezoelectric film sensor with wide frequency response range and high sensitivity, mainly comprises a three-layer structure (see figure 3), an upper protective layer, an upper electrode layer, a PVDF piezoelectric film, a lower electrode layer and a lower protective layer are sequentially arranged from top to bottom, the overall thickness is about 8-10mm, the size is set to 400mmx400mm by referring to the size of a common seat cushion, and the protective layer is made of polyethylene material.
Examples
The embodiment is a sitting posture-oriented unbound heart rate monitoring method, and the size of the flexible piezoelectric film sensor is 9mm in overall thickness and 400mmx400mm in length and width. The method comprises the following steps:
step one, seat separation judgment:
the upper computer amplifies signals through the charge amplifier and collects the trend of charge quantity change of the flexible piezoelectric film sensor in real time through the data collection equipment, and the signal sampling frequency is set to be 1000 Hz. Calculating the standard deviation of data per second in real time in the acquisition process, judging the threshold value and setting the threshold value range as [2e ]-4,5e-4]And judging whether the seat is occupied or not according to the standard deviation of the collected data and monitoring. If the standard deviation of the acquired data is not within the threshold range, the monitoring system does not work, because the human body is out of seat or the human body motion amplitude is large at the moment, the heart rate monitoring cannot be carried out, and the state is recorded as state 1; if the standard deviation of the collected data is within the threshold range, it indicates that no person is out of the seat and the human body is in a static state, and at the moment, continuously collecting data for 10 seconds, and recording the data as a state 2; and if the standard deviation of the data within the time of 10 seconds is within the range of the threshold value, entering the next step, otherwise, judging the state 1, and judging the state 2 when the data is in the state 1, and recombining the continuous data of 10 seconds.
Secondly, preprocessing signals:
and preprocessing the 10-second data in the state 2 acquired in the first step, wherein the specific processing processes include baseline drift removal and high-frequency noise removal. And fitting and removing a trend term by using a polynomial, then decomposing and reconstructing the signal without the trend term by adopting wavelet transform, and finally removing high-frequency noise by using a low-pass filter to obtain a preprocessed heart impact signal (see fig. 4).
The objects of a second data (10 second data) mentioned in the following process are all referred to as preprocessed ballistocardiographic signals.
Thirdly, estimating a heartbeat cycle:
and (4) extracting all positive peaks (peaks larger than zero) in the 10-second data after the pretreatment in the second step by using a peak detection algorithm and recording the position of the time axis. Taking a certain positive peak as an example, respectively calculating an autocorrelation function R, a short-time average amplitude difference function D and a kernel function Q of each time axis position in 1.5-8.5 seconds of data according to a formula (1), a formula (2) and a formula (3), and obtaining the heart cycle of the current positive peak through the kernel function Q. Detecting the positive peak of the function Q, and setting a peak threshold value 5e-3The position of the positive peak exceeding the peak threshold is taken as the current cycle position (see fig. 5), and the average interval of the adjacent cycle positions is calculated as the estimated heartbeat cycle at a certain positive peak. Calculating estimated heartbeat cycles of all positive wave crests in the data of 1.5-8.5 seconds to form a heartbeat cycle sequence, and calculating an arithmetic mean value of the heartbeat cycle sequence as a heartbeat cycle estimated value T in the data of 10 seconds;
wherein x is the preprocessed ballistocardiogram signal; m is the time axis position of the positive wave crest in 1.5-8.5 seconds of data, and the value range of k is 0-N; n represents the total length of time intercepted at two sides of the time axis position of the positive wave peak; the value range of N is [0.8f,3f ], and f is the signal sampling frequency. For example, the sampling frequency is 1000Hz, the data acquisition device acquires 1000 data points per second, and the value range of N is 800-3000, corresponding to 0.8-3 seconds. N is the sum of two heart rate intervals, so the corresponding heart rate ranges from 40 to 150 beats/min, 3000 is selected for N in this example.
Fourthly, dividing the partitions:
dividing the impact cardiac signal preprocessed in the second step into continuous intervals (see fig. 6), wherein the interval length is the estimated value T of the heartbeat cycle calculated in the third step. Judging the validity of the interval, if the maximum amplitude of the interval exceeds the set validity threshold range [2e ]-4,5e-4]If the interval is invalid, the interval is judged to be invalid because body movement is inevitable in the measurement process, so that the heart impact signal is interfered; if the maximum amplitude is within the range of the validity threshold, the interval is valid, and then the time axis position of the maximum amplitude is positioned from the valid interval to be used as the time axis position of the maximum J peak of the main characteristic of the heart impact signal in 10 seconds.
Fifth step, heartbeat mould
And taking the time axis position of the J peak in the data calculated in the fourth step within 10 seconds as the center, and intercepting data b with the forward data length and the backward data length of the corresponding time axis position being 500T to form data c with the length being 1000T + 1. The arithmetic mean of the data c is calculated as the heartbeat template for the current data a (see fig. 7). And monitoring the cardiac shock signal of the next 10 seconds of data, repeating the first step to the fifth step every time the heartbeat template of continuous 10 seconds of data is obtained, and updating the heartbeat template.
Sixth step, heart beat location
The correlation coefficient between the heartbeat signal and its heartbeat template every 10 seconds is calculated according to equation (5) (see fig. 8). And (3) carrying out correlation threshold judgment on the correlation coefficient, setting the correlation threshold to be 0.85, and if the correlation coefficient exceeds the threshold, ensuring that the similarity degree of the heart attack signal and the heart beat template is higher and one heart beat is obtained. All the ballistocardiogram signals with the similarity degree exceeding the correlation threshold are detected, and the heartbeat positions are marked accordingly (see fig. 9). And dividing the number of all the heartbeat positions by a to obtain the heart rate, and calculating the average value of the intervals of all the adjacent heartbeat positions to obtain the heartbeat period.
Wherein x isi，yiThe preprocessed heart impact signal is obtained;
is a heartbeat template signal; and M is the length of the preprocessed ballistocardiogram signal.
In the embodiment, a medical contrast experiment is added, and the electrocardiogram detection of the medical clinical equipment Anlanhua polysomnography N7000 is utilized to be compared with the heart rate detection result of the invention. The experimental environment is a quiet environment, no person walks around, people sit on the chair fixed with the flexible piezoelectric film sensor, the cardiac electrodes of the electrocardiogram module are attached to the left pectoral muscle and the right pectoral muscle of the person, the sitting posture is kept to the greatest extent, and the monitoring system and the electrocardiogram module are used for measuring simultaneously.
The system used in the application acquires data processed by the upper computer through the data acquisition equipment, obtains all heartbeat positions through the method of the embodiment, and calculates the heart rate and the heartbeat period.
Storing data acquired by electrocardiogram offline, obtaining all heartbeat positions in the data by using a threshold detection method, and calculating to obtain corresponding heart rate and heartbeat period.
Manually aligning the heartbeat positions obtained by the two systems, comparing the heart rates and the heartbeat cycles obtained by the two systems, wherein the initial time positions of the two data are the same, and the time lengths of the two data are the same. Compared with the detection result (see fig. 10) of the electrocardiogram module (the sampling frequency is 1000Hz), the relative error of the heart rate of the method of the embodiment (see fig. 9) is not more than 2.41%, the average difference of the heart cycles is not more than 30ms, and the method has high accuracy.
The method can also be used for extracting the heart rate in a lying state, and the flexible piezoelectric film sensor needs to be arranged in the back area of a human body in the lying state.
The invention is not described and is applicable to the prior art.
The above detailed description of the embodiments of the present invention is merely a preferred embodiment of the present invention, and should not be considered as limiting the scope of the present invention. All equivalent changes and modifications made within the scope of the claims of the present application shall fall within the protection scope of the present application.
Claims (7)
1. A real-time unrestrained heart rate extraction method utilizes a flexible piezoelectric film sensor to collect a heart impact signal of a human body, and comprises the following steps:
the method comprises the following steps of firstly, determining whether a human body is located at a position to be detected within a continuous period of a seconds and is in a static state, if the condition is met, preprocessing data of the continuous period of a seconds, removing baseline drift and high-frequency noise, and obtaining a preprocessed heart impact signal:
second, estimating the heart cycle
Extracting all positive peaks in a-second data of the preprocessed heart shock signals by using a peak detection algorithm, recording the time axis position, respectively calculating an autocorrelation function R and a short-time average amplitude difference function D of each time axis position in 1.5-a-1.5-second data according to a formula (1) and a formula (2), and obtaining a core function Q according to a formula (3) by using the autocorrelation function and the short-time average amplitude difference function, wherein the core function Q is expressed as a peak value at the heart beat cycle position;
detecting a positive wave peak of a core function, setting a wave peak threshold value delta, taking the position of the positive wave peak with amplitude exceeding delta as a heartbeat cycle position, and calculating the average interval of adjacent heartbeat cycle positions as an estimated heartbeat cycle at a certain positive wave peak;
calculating estimated heartbeat cycles of all positive wave crests in the data of 1.5-a-1.5 seconds to form a heartbeat cycle sequence, and calculating an arithmetic mean value to serve as a heartbeat cycle estimated value T in the data of a seconds;
wherein x is the preprocessed ballistocardiogram signal; m is the time axis position of the positive wave crest in 1.5-a-1.5 seconds of data, and the value range of k is 0-N; n represents the total length of time intercepted at two sides of the time axis position of the positive wave peak; the value range of N is [0.8f,3f ], and f is the signal sampling frequency;
thirdly, dividing the partitions:
dividing the preprocessed heart impact signal into continuous intervals, wherein the length of each interval is the obtained heart cycle estimation value T; judging the validity of the interval, if the maximum amplitude of the interval exceeds the set valid threshold range [2e-4,5e-4]If the interval is invalid, the heart attack signal is interfered by the outside; if the maximum amplitude is within the effective threshold range, the interval is an effective interval; positioning the time axis position of the maximum amplitude from the effective interval as the time axis position of the main characteristic maximum J peak of the ballistocardiographic signal in the a second data;
the fourth step: heart beat template
Taking the time axis position of the maximum J peak of the main characteristics of the obtained heart impact signal in the data of a second as the center, acquiring data b with the data length of T.f/2 forwards and backwards at the time axis position of the maximum J peak, forming data c by the center and the data b, and calculating the arithmetic mean value of the data c as the heart beat template of the data of the current a second;
every time continuous a second preprocessed heart impact signal data is obtained, recalculating a heartbeat template corresponding to the next a second data;
and a sixth step: location of heart beat
Calculating a correlation coefficient r between each a-second heart attack signal and a heart beat template thereof according to a formula (5), and positioning the heart beat position of the current heart attack signal according to the correlation coefficient; dividing the number of the heartbeat positions by a to obtain the heart rate; calculating a difference value of the intervals of adjacent heartbeat positions to obtain a successive heartbeat interval;
wherein x isi，yiIs (pre-processed cardioblast signal;
is a heartbeat template signal; m is the length of the ballistocardiogram signal.
2. The method according to claim 1, wherein the method is used for heart rate extraction in a sitting posture, a flexible piezoelectric film sensor is placed on a seat surface in the sitting posture, and the specific process of the first step is as follows:
the upper computer collects the charge quantity generated by the flexible piezoelectric film sensor in real time through data collection equipment, and the signal sampling frequency is set to be 1000 Hz; calculating the standard deviation of data per second in real time in the acquisition process, judging the threshold value and setting the threshold value range as [2e ]-4,5e-4]Judging whether a person exists in the seat according to the standard deviation of the collected data and monitoring; if the standard deviation of the acquired data is not within the threshold range, the monitoring system does not work, because the human body is out of seat or the human body motion amplitude is large at the moment, the heart rate monitoring cannot be carried out, and the state is recorded as state 1; if the standard deviation of the collected data is within the threshold range, the situation that no person is out of the seat and the human body is in a static state is indicated, and at the moment, a-second data are continuously collected and recorded as a state 2; if the standard deviation of the data in the a second time is in the threshold range, preprocessing is carried out, otherwise, state 1 judgment is carried out, and state 2 judgment is carried out when the data is in the state 1, and continuous a second data are formed again.
3. The method according to claim 1, wherein the correlation coefficient between the heartbeat template and the heartbeat signal every a seconds is calculated according to formula (5), the threshold value of the correlation coefficient is determined, the threshold value of the correlation is set to 0.85, and if the correlation coefficient exceeds the threshold value of the correlation, the degree of similarity between the heartbeat template and the heartbeat signal is high, and the heartbeat is determined.
4. The method of claim 1, wherein the sampling frequency is 1000Hz, a is 10-15, and the peak threshold δ is 5e-3。
5. A monitoring system for extracting a real-time unbound heart rate comprises a charge amplifier, data acquisition equipment and an upper computer, and is characterized by further comprising a flexible piezoelectric film sensor and a seat; the flexible piezoelectric film sensor is arranged on the surface of the seat and is connected with data acquisition equipment through a charge amplifier, the data acquisition equipment transmits data to an upper computer, and the upper computer is loaded with the method of any one of claims 1 to 4.
6. The system of claim 5, wherein the host computer is in communication with the mobile terminal.
7. The system according to claim 5, wherein the flexible piezoelectric film sensor is a PVDF piezoelectric film sensor, and comprises an upper protective layer, an upper electrode layer, a PVDF piezoelectric film, a lower electrode layer and a lower protective layer which are sequentially arranged from top to bottom, wherein the overall thickness is 8-10mm, and the size is 400mmx400 mm.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010145370.XA CN111329462B (en) | 2020-03-05 | 2020-03-05 | Real-time non-binding heart rate extraction method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010145370.XA CN111329462B (en) | 2020-03-05 | 2020-03-05 | Real-time non-binding heart rate extraction method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111329462A true CN111329462A (en) | 2020-06-26 |
| CN111329462B CN111329462B (en) | 2022-05-20 |
Family
ID=71174509
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010145370.XA Active CN111329462B (en) | 2020-03-05 | 2020-03-05 | Real-time non-binding heart rate extraction method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111329462B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112022134A (en) * | 2020-07-22 | 2020-12-04 | 无锡金童科技有限公司 | Heartbeat detection method based on ballistocardiogram |
| CN114847619A (en) * | 2022-04-29 | 2022-08-05 | 北京航空航天大学 | Mining sign detection emergency penetration transmitting backpack based on small low-frequency transmitting antenna |
| CN115153475A (en) * | 2022-06-01 | 2022-10-11 | 中国第一汽车股份有限公司 | Vehicle-mounted heartbeat interval detection method and device and vehicle |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103006223A (en) * | 2012-12-13 | 2013-04-03 | 中国人民解放军第四军医大学 | Household non-contact sleeping monitoring device and method |
| CN105919584A (en) * | 2016-06-23 | 2016-09-07 | 电子科技大学 | Heart rate estimation method and device for wearable heart rate monitoring equipment |
| CN107137071A (en) * | 2017-04-26 | 2017-09-08 | 可瑞尔科技(扬州)有限公司 | It is a kind of to analyze the method that heart impact signal is used for calculating short-term heart beat value |
| CN108992053A (en) * | 2018-06-21 | 2018-12-14 | 河北工业大学 | A method of real-time chainless detection heart rate and eartbeat interval |
-
2020
- 2020-03-05 CN CN202010145370.XA patent/CN111329462B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103006223A (en) * | 2012-12-13 | 2013-04-03 | 中国人民解放军第四军医大学 | Household non-contact sleeping monitoring device and method |
| CN105919584A (en) * | 2016-06-23 | 2016-09-07 | 电子科技大学 | Heart rate estimation method and device for wearable heart rate monitoring equipment |
| CN107137071A (en) * | 2017-04-26 | 2017-09-08 | 可瑞尔科技(扬州)有限公司 | It is a kind of to analyze the method that heart impact signal is used for calculating short-term heart beat value |
| CN108992053A (en) * | 2018-06-21 | 2018-12-14 | 河北工业大学 | A method of real-time chainless detection heart rate and eartbeat interval |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112022134A (en) * | 2020-07-22 | 2020-12-04 | 无锡金童科技有限公司 | Heartbeat detection method based on ballistocardiogram |
| CN114847619A (en) * | 2022-04-29 | 2022-08-05 | 北京航空航天大学 | Mining sign detection emergency penetration transmitting backpack based on small low-frequency transmitting antenna |
| CN114847619B (en) * | 2022-04-29 | 2023-06-16 | 北京航空航天大学 | Mining sign detection emergency penetration signaling knapsack based on small-sized low-frequency signaling antenna |
| CN115153475A (en) * | 2022-06-01 | 2022-10-11 | 中国第一汽车股份有限公司 | Vehicle-mounted heartbeat interval detection method and device and vehicle |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111329462B (en) | 2022-05-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111329462B (en) | Real-time non-binding heart rate extraction method | |
| US20240130627A1 (en) | Continuous non-invasive monitoring of a pregnant human subject | |
| CN107137071B (en) | Method for calculating short-term heart rate value by analyzing heart attack signal | |
| JP7037366B2 (en) | Wearable technology for joint health assessment | |
| EP3065628B1 (en) | Biomechanical activity monitoring | |
| Salahuddin et al. | Detection of acute stress by heart rate variability using a prototype mobile ECG sensor | |
| EP2375982B1 (en) | Method and apparatus for the analysis of ballistocardiogram signals | |
| US7409242B2 (en) | Active muscle display device | |
| JP6457117B2 (en) | Continuous non-invasive monitoring of pregnant subjects | |
| US20090018410A1 (en) | Body parameter sensing | |
| JP2018512243A5 (en) | ||
| Vehkaoja et al. | Correlation approach for the detection of the heartbeat intervals using force sensors placed under the bed posts | |
| CN108992053B (en) | Method for real-time non-binding detection of heart rate and heartbeat interval | |
| Liu et al. | Automatic detection of respiratory rate from electrocardiogram, respiration induced plethysmography and 3D acceleration signals | |
| WO2012136744A4 (en) | Ecg mat | |
| TWI629049B (en) | A method for analyzing a heart shock signal for calculating a short-term heart rate value | |
| Shen et al. | Ballistocardiogram-based heart rate variation monitoring using unsupervised learning | |
| CN116369888B (en) | Non-contact heart rate variability data acquisition method and device | |
| Akhbardeh et al. | Ballistocardiogram classification using a novel transform so-called AliMap and biorthogonal wavelets | |
| RU2741233C1 (en) | Method for differential diagnosis of essential tremor and early and first stages of parkinson's disease by analysing flare-like activity of muscles | |
| Patil et al. | Reconstruction of respiratory signal from ECG | |
| Ansari et al. | Extraction of respiratory rate from impedance signal measured on arm: A portable respiratory rate measurement device | |
| Ravariu et al. | Manufacturing of capacitive electrodes on Si-substrate for electrophysiological applications | |
| Barna et al. | Heart-rate detection using EMFi sensors on a normal looking chair | |
| Chaudhuri et al. | Review of ECG Analysis |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |