A method, apparatus and storage medium for motion heart rate measurement

Abstract

The application discloses a kind of method, device and storage medium of motion heart rate measurement, the method mainly includes S1, to the original signal collected to the wearable device is handled, to obtain the frequency spectrum signal after spectral reduction, in this step, by filtering processing to original heart rate signal, enhance the anti-interference ability;S2, to the frequency spectrum signal obtained is tracked to spectrum peak, to obtain heart rate spectrum peak value;In this step, by tracking to spectrum peak to the frequency spectrum signal obtained, reduce the heart rate deviation excessively large problem caused by single time spectrum anomaly (noise is not handled cleanly, or does not collect heart rate signal), to improve the accuracy of heart rate output value;S3, to the heart rate spectrum peak value is handled, obtains heart rate value.Through above method step, the accuracy and stability of heart rate output value are greatly improved in the present application.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of wearable devices, in particular to a method and device for measuring heart rate during exercise and a storage medium. BACKGROUND

[0002] In people's daily work and life, exercise is one of the important means to maintain physical and mental health, and the heart rate value in the process is the core indicator for monitoring the exercise state. For wearable devices, they have characteristics that other forms of devices cannot match in obtaining user physiological information parameters: convenient and fast, and more easily achieve real-time monitoring all day long. Common physiological signals include optical volume pulse wave (PPG), electrocardiogram (ECG), electrodermal activity (EDA), etc. Among them, PPG signal is relatively easy to obtain, has lower requirements for data measurement sites, and has fewer restrictions on users during data collection, etc. Therefore, PPG is often used as an important measurement signal in wearable devices. However, the PPG signal collected by the wearable device often mixes with motion artifacts, baseline drift, respiratory interference and other noises, which causes inaccurate heart rate calculation based on the PPG signal. Since the motion artifacts generated in different exercise scenarios are not the same, especially in intermittent high-intensity exercise, it is difficult to remove the motion artifacts in the PPG signal. At present, there are great challenges for the wearable PPG scheme to accurately and real-timely calculate the heart rate under different exercise states and exercise state switching. SUMMARY

[0003] The present application aims to overcome the shortcomings of the prior art and provide a method and device for measuring heart rate during exercise and a storage medium to improve the accuracy and stability of heart rate output.

[0004] To achieve the above-mentioned purpose, the technical solution of the present application is as follows:

[0005] In a first aspect, the present application provides a method for measuring heart rate during exercise, comprising:

[0006] S1, processing the original signal collected by the wearable device to obtain a spectrum-subtracted frequency spectrum signal;

[0007] S2, performing spectral peak tracking on the obtained frequency spectrum signal to obtain a heart rate spectral peak value;

[0008] S3, processing the heart rate spectral peak value to obtain a heart rate value.

[0009] In a second aspect, the present application provides a method for measuring heart rate during exercise, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-mentioned method.

[0010] In a third aspect, the present application provides a computer readable storage medium, which stores a computer program, and the computer program, when executed by a processor, implements the steps of the method described above.

[0011] Compared with the prior art, the present application has the following advantages:

[0012] Since the wearable PPG heart rate measurement scheme is affected by a lot of noise, if different types of noise are not well processed, the stability of the heart rate measurement scheme will be reduced, that is, the heart rate value will jump. In order to enhance the anti-interference ability of the present scheme, the present application adopts a specific filter set: second-order difference filtering, median filtering, mean filtering, band-pass filtering, adaptive filtering (RLS), CZT transformation, and spectral subtraction. In order to reduce the excessive heart rate deviation caused by single-time spectrum anomaly (noise not well processed or heart rate signal not collected), that is, to increase the accuracy of the scheme output, the present application adopts a hidden Markov model for spectral peak tracking. In order to reduce the excessive heart rate deviation caused by the previous time output heart rate and the current time calculation heart rate, and to reduce the direct output that causes the heart rate to jump significantly, the present application adopts an output post-processing strategy based on signal quality. Through the above methods, the accuracy and stability of the heart rate output of the present application are greatly improved. BRIEF DESCRIPTION OF DRAWINGS

[0013] Figure 1 A method flowchart for measuring exercise heart rate is provided for embodiment 1 of the present application.

[0014] Figure 2 A specific flowchart for step S1 is provided.

[0015] Figure 3 A specific flowchart for step S2 is provided.

[0016] Figure 4 A device composition schematic diagram for measuring exercise heart rate is provided for embodiment 2 of the present application. DETAILED DESCRIPTION

[0017] The technical solutions of the present application will be further described below in combination with the drawings and embodiments.

[0018] Embodiment 1

[0019] Since the existing wearable PPG heart rate measurement method is affected by a lot of noise, if different types of noise are not well processed, the stability of the heart rate measurement scheme will be reduced, that is, the heart rate value will jump. In order to improve the accuracy and stability of the heart rate output, the present embodiment provides a method for measuring exercise heart rate, mainly including the following steps:

[0020] S1, filtering the original motion heart rate signal collected by the wearable device to obtain a spectrum signal after spectral subtraction.

[0021] In this step, the anti-interference ability is enhanced by filtering the original heart rate signal. The storage of the original motion heart rate collected by the wearable device includes device-side storage, APP-side storage, and server-side storage. The device side stores heart rate information for a short period of time, and the number of stored information should not exceed 30. The APP side stores user information for a medium and short period of time, and stores user heart rate information for nearly a week. When the time limit is exceeded, the APP side obtains relevant data from the server side. The server side stores user information for a long period of time. When the device side or the APP side initiates a heart rate information synchronization requirement, the server side needs to respond and issue data.

[0022] S3, performing spectrum peak tracking on the obtained spectrum signal to obtain a heart rate spectrum peak value.

[0023] In this step, by performing spectrum peak tracking on the obtained spectrum signal, the problem of too large heart rate deviation caused by single time spectrum anomaly (noise not processed cleanly or heart rate signal not collected) is reduced, thereby improving the accuracy of heart rate output value.

[0024] S3, processing the heart rate spectrum peak value to obtain a heart rate value.

[0025] In this way, by the above method steps, the accuracy and stability of the heart rate output value are greatly improved

[0026] In a specific embodiment, the original signal involved in the motion heart rate measurement includes a green light PPG signal and a three-axis acceleration sensor signal (ACC signal), and the sampling rate is 25Hz. However, in some other embodiments, the sampling rate includes but is not limited to 25Hz.

[0027] The specific processing procedure of step S1 is shown in Figure 2 , and specifically includes the following steps:

[0028] S11, performing second-order difference processing on the fixed window length ACC original signal, and the difference formula is as follows:

[0029] y i = x i+2 + x i - 2x i+1

[0030] Wherein, i is the index of the signal, y i is the second-order difference signal, and x iThe input signal of the differential filter is a differential signal. The second-order differential filter can effectively suppress low-frequency signals and reduce the influence of low-frequency signals. In other embodiments, the order of the differential filter includes but is not limited to the second order.

[0031] In S12, the second-order differential filtered signal is subjected to a third-order median filter, so as to reduce the influence of signal step jumps on subsequent signal processing. In other embodiments, the order of the median filter includes but is not limited to the third order.

[0032] In S13, the median filtered signal is subjected to a seventh-order mean filter, so as to suppress signal components above 4 Hz. In other embodiments, the order of the mean filter includes but is not limited to the seventh order.

[0033] In S14, the mean filtered signal is subjected to a butterworth band-pass filter with a passband of [0.5, 4] Hz, so as to suppress low-frequency signals below 0.5 Hz and high-frequency signals above 4 Hz. In other embodiments, the method for implementing the band-pass filter includes but is not limited to the butterworth band-pass filter, and the passband includes but is not limited to [0.5, 4] Hz.

[0034] In S15, the sum of the processed ACC signals is subjected to a CZT (Chirp z-transform) transform. Under the same frequency spectrum calculation point number, the CZT transform has higher frequency spectrum resolution than the Fourier transform. After obtaining the frequency spectrum graph of the sum of the ACC signals, the ACC frequency spectrum signal concentration a is calculated, and the calculation method is as follows:

[0035]

[0036] Wherein, p i is the frequency point of the i-th peak of the ACC frequency spectrum. In this embodiment, three is selected considering that the strongest spectrum peak point may be the fundamental frequency, the second harmonic, or the third harmonic of the motion frequency. In other embodiments, the three peak points include but are not limited to the above-mentioned three peak points. Th1 is a spectrum peak width threshold, including but not limited to 10BPM. f(*) is the spectrum corresponding to the sum of the three-axis ACC signals. At the same time, the maximum value P of the spectrum in the passband [0.5, 4] Hz is calculated. If the spectrum concentration a is greater than the threshold value and the maximum value P of the spectrum is greater than the threshold requirement, the ACC frequency spectrum signal meets the requirement.

[0037] In other embodiments, the object of obtaining the three-axis acceleration frequency spectrum includes but is not limited to the sum of the three-axis acceleration, and can also be the resultant acceleration or other combinations of the three-axis acceleration. The method for obtaining the PPG or ACC frequency spectrum signal includes but is not limited to the CZT transform.

[0038] S16, if the ACC spectrum signal meets the requirements, the PPG signal is respectively filtered by the RLS adaptive filter of three-axis acceleration x, y, z, and if the ACC spectrum does not meet the requirements, the PPG signal is not filtered by the adaptive filter. The spectrum limitation of the ACC can effectively reduce the spectrum confusion caused by wrist rotation, action switching and the like, so as to cause the adaptive filter coefficient to be difficult to converge, and finally cause the filtering effect of the adaptive filter to be poor, and even possibly introduce noise; in other embodiments, the adaptive filter includes but is not limited to RLS, and can also be other adaptive filters or a combination thereof.

[0039] S17, the PPG signal is subjected to CZT transformation to obtain the spectrum information of [0.5, 4] Hz thereof;

[0040] S18, if the ACC spectrum signal meets the requirements of S15, spectral subtraction is performed, and the spectral subtraction method is as follows:

[0041]

[0042] wherein, f PPGC (i) is the spectrum amplitude of the frequency i after spectral subtraction, f PPG (i) is the spectrum before processing, f ACCN (i) is the normalized sum spectrum of the acceleration, f PPGN (i) is the spectrum of the PPG signal after normalization; the spectrum signal after spectral subtraction is the final output of signal processing, and is used for real-time heart rate calculation.

[0043] In other embodiments, the spectral subtraction method includes but is not limited to the above method, and the acceleration spectrum of each axis can be squared, normalized, and then averaged as the ACC spectrum signal, and finally the above spectral subtraction or other spectral subtraction is performed;

[0044] In a specific implementation, the spectral peak tracking process of S2 is as shown in Figure 2 , and mainly includes the following steps:

[0045] S21, the PPG spectrum signal and the ACC spectrum signal are obtained, wherein the ACC spectrum signal is obtained in S15, and the PPG spectrum signal is obtained in S18;

[0046] S22, the [0.5, 4] Hz strongest spectrum amplitude, the motion fundamental frequency, and the spectrum concentration degree characteristics in the ACC spectrum signal are calculated, and the frequency points of the three spectrum peaks with the largest amplitude, the spectrum concentration degree, and the maximum amplitude in the PPG spectrum signal are calculated;

[0047] S23, evaluate the state in which the current signal is located using a hidden Markov model, wherein the visible state of the hidden Markov model is composed of PPG features and ACC features, wherein the strongest amplitude of the ACC can divide the ACC state into strong and weak states, the ACC spectral concentration feature can divide the ACC state into clear and chaotic states, similarly, the PPG maximum amplitude can divide the PPG signal into strong and weak states, the PPG spectral concentration feature can divide the PPG state into clear and chaotic states, and the PPG signal can be divided into two states of existing harmonic and non-existing harmonic by whether the frequency points of the three spectral peaks with the largest amplitude of the PPG signal exist harmonic relationship, that is, the visible state of the model is divided into 42 kinds, and the hidden state of the model is set to 6 kinds, corresponding to the following 6 kinds of situations in which the heart rate value of the user can appear:

[0048] 1) The frequency point corresponding to the true heart rate value is the PPG main frequency, that is, the frequency point with the maximum spectral amplitude;

[0049] 2) The frequency point corresponding to the true heart rate value is the harmonic or half frequency of the main frequency, that is, the maximum frequency point of the spectrum can be the half frequency or the multiple frequency of the heart rate;

[0050] 3) The frequency point corresponding to the true heart rate value is not the half frequency or multiple frequency of the PPG main frequency, but also corresponds to a PPG spectral peak;

[0051] 4) The frequency point corresponding to the true heart rate value is the ACC main frequency;

[0052] 5) The frequency point corresponding to the true heart rate value is the half frequency or multiple frequency of the ACC main frequency;

[0053] 6) The frequency point corresponding to the true heart rate value does not exist in the PPG spectrum and the ACC spectrum;

[0054] The above 6 situations of the model include not only normal heart rate measurement scenarios, but also ACC and PPG co-frequency scenarios, and scenarios in which heart rate signals cannot be collected. The state transition matrix and the observation matrix in the above model are obtained by a large amount of data statistics;

[0055] In other embodiments, the model used by the spectrum tracking includes but is not limited to a hidden Markov model, and can also be other types of models; the observation state and the hidden state of the hidden Markov model used include but are not limited to the above types.

[0056] S24, obtain the hidden state of the current heart rate measurement in step S23, and the selection rules of the heart rate spectral peak in the above 6 states are as follows:

[0057] 1) The heart rate spectral peak is selected from the strongest point of the PPG spectrum;

[0058] 2) The heart rate spectral peak is selected from the half frequency or multiple frequency spectral peak of the PPG main frequency which has the smallest frequency point difference from the previous moment;

[0059] 3) Heart rate spectrum peak selection selects the frequency point of the spectrum peak with the smallest difference from the previous time;

[0060] 4) Heart rate spectrum peak selection selects the strongest point of ACC spectrum;

[0061] 5) Heart rate spectrum peak selection selects the frequency point with the smallest difference from the previous time among the half frequency or multiple frequency spectrum peaks of the main frequency of ACC;

[0062] 6) Heart rate spectrum peak selection keeps consistent with the heart rate spectrum peak of the previous time.

[0063] In other embodiments, the heart rate spectrum peak selection includes but is not limited to the above-mentioned methods, and each hidden state can also directly correspond to a spectrum peak type, i.e., the hidden state directly obtains the heart rate spectrum peak.

[0064] In order to reduce the large deviation of the previously output heart rate and the current calculated heart rate caused by abnormalities, and to prevent the direct output from causing a significant jump in heart rate, in step S3, a signal quality-based output post-processing method is used based on the obtained heart rate spectrum peak value, as follows:

[0065] HR t = γHR ref + (1-γ)HR current

[0066]

[0067] wherein HRt is the weighted heart rate of the heart rate values of the previous time and the current time, HR ref is the reference heart rate value, and the selection rule is as follows:

[0068] 1) Calculate the ratio β of the spectral energy of the fundamental frequency, double frequency and triple frequency of the corresponding frequency point of the real-time calculated heart rate value in the PPG spectrum within the threshold bandwidth to the total spectral energy of [0.5, 4] Hz spectrum, and the calculation method is the same as the signal concentration α;

[0069] 2) If there is a case where β is greater than the preset value within the previous 5s, then HR ref is set as the output heart rate value at this time, otherwise HR ref is set as the output heart rate value of the previous time;

[0070] γ is the heart rate confidence parameter, if the current β is greater than the preset value, γ is 0, otherwise γ = β1 / (β1+β2), β1 is the β value of the reference heart rate, and β2 is the β value of the current heart rate; HR current is the current heart rate value; HR th is the threshold value of the maximum heart rate difference allowed to appear in the adjacent time interval of the heart rate, i.e., HR tThe value exceeding the threshold range will be limited to the upper and lower limits, the selection of which is related to the current motion type and motion intensity, wherein the current motion type is obtained through device end user input or motion self-recognition module, and the motion intensity is obtained through three-axis ACC combined acceleration intensity level; the HR is the final output heart rate value;

[0071] In other heart rate output values, including but not limited to the above-mentioned manner, other models can also be obtained through kalman model.

[0072] In summary, the application adopts specific filter sets: second-order difference filtering, median filtering, mean filtering, band-pass filtering, adaptive filtering (RLS), CZT transformation, and spectral subtraction; in order to reduce the excessive heart rate deviation caused by single-time spectrum anomaly (noise not processed cleanly, or heart rate signal not collected), that is, to increase the accuracy of the scheme output value, the application adopts a hidden Markov model for spectral peak tracking; in order to reduce the excessive heart rate deviation caused by the previous time output heart rate and the current time calculation heart rate, and the direct output leading to the significant jump of the heart rate, the application adopts an output post-processing strategy based on signal quality. Through the above methods, the accuracy and stability of the heart rate output value are greatly improved.

[0073] Embodiment 2:

[0074] Referring to Figure 4 As shown in the figure, the method for measuring motion heart rate provided by the embodiment includes a processor 41, a memory 42, and a computer program 43 stored in the memory 42 and executable on the processor 41, such as a redundancy robot motion planning program. The processor 41 executes the computer program 43 to implement the steps of embodiment 1 described above, such as Figure 1 The steps shown in the figure.

[0075] For example, the computer program 43 can be divided into one or more modules / units, which are stored in the memory 42 and executed by the processor 41 to complete the application. The one or more modules / units can be a series of computer program instruction segments that can complete a specific function, which are used to describe the execution process of the computer program 43 in the method for measuring motion heart rate.

[0076] The processor 41 can be a central processing unit (CPU), and can also be other general-purpose processors, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor.

[0077] The memory 42 can be an internal storage unit of the motion heart rate measuring method, such as a hard disk or a memory of the motion heart rate measuring method. The memory 42 can also be an external storage device of the motion heart rate measuring method, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. Further, the memory 42 can include both an internal storage unit and an external storage device of the motion heart rate measuring method. The memory 42 is used to store the computer program and other programs and data required by the motion heart rate measuring method. The memory 42 can also be used to temporarily store data that has been output or will be output.

[0078] Embodiment 3

[0079] The embodiment provides a computer readable storage medium, which stores a computer program. The computer program is executed by a processor to implement the steps of the method in embodiment 1.

[0080] The computer readable medium can be any medium that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. Computer readable medium more specifically can be, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device. More specific examples (a non-exhaustive list) of the computer readable medium include the following: an electrical connection having one or more wires (electronic devices), a portable computer diskette (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber (optical device), and a portable compact disc read-only memory (CDROM). In addition, the computer readable medium can even be paper or another suitable medium upon which the program is printed, as the program can be electronically captured, via, for instance, optical scanning of the paper or other medium, then compiled, interpreted, or otherwise processed in a suitable manner, if necessary, and then stored in a computer memory.

[0081] The above embodiments are only to illustrate the technical concept and characteristics of the present application, and the purpose is to enable those skilled in the art to understand the content of the present application and to implement it, and cannot limit the protection scope of the present application. Any equivalent changes or modifications made according to the essence of the present application should be covered within the protection scope of the present application.

Claims

1. A method for measuring exercise heart rate, characterized in that, include: S1. Process the raw signal collected by the wearable device to obtain the spectral signal after spectral subtraction; the raw signal includes a PPG signal and an ACC signal; the PPG signal is an optical volumetric pulse wave signal; the ACC signal is a triaxial accelerometer signal. S2. Perform spectral peak tracking on the obtained spectral signal to obtain the peak value of the heart rate spectrum; S3. Process the peak values ​​of the heart rate spectrum to obtain the heart rate value; Step S2 includes: S21. Obtain the PPG spectrum signal and the ACC spectrum signal; S22. Calculate the strongest spectral amplitude, fundamental frequency, and spectral concentration characteristics in the [0.5,4]Hz range of the ACC spectrum signal. Calculate the frequency points, spectral concentration, and maximum amplitude of the three spectral peaks with the largest amplitude in the PPG spectrum signal. S23. Use a hidden Markov model to evaluate the current state of the signal in order to obtain the hidden state of heart rate measurement. S24. Select heart rate peaks based on the heart rate hidden state; The visible states of the Hidden Markov Model (HMM) are composed of PPG and ACC features. The ACC state is divided into strong and weak states based on its maximum amplitude, and clear and chaotic states based on the concentration of the ACC spectrum signal. Similarly, the PPG signal is divided into strong and weak states based on its maximum amplitude, and clear and chaotic states based on the concentration of the PPG spectrum signal. The HMM has 42 visible states and 6 hidden states, corresponding to the following 6 possible scenarios for a user's heart rate value: 1) The frequency point corresponding to the true heart rate value is the PPG main frequency, that is, the frequency point where the spectrum amplitude is the largest; 2) The frequency point corresponding to the true heart rate value is the harmonic or half-frequency point of the main frequency, that is, the maximum frequency point of the spectrum may be half-frequency or multiple frequency of the heart rate. 3) The frequency point corresponding to the true heart rate value is not half or overtone of the PPG main frequency, but it also corresponds to a PPG spectral peak; 4) The frequency point corresponding to the true heart rate value is the ACC main frequency; 5) The frequency point corresponding to the true heart rate value is half or double the ACC main frequency; 6) The frequency point corresponding to the true heart rate value has no spectral peak in either the PPG or ACC spectrum; The above six cases of the Hidden Markov Model include not only the normal heart rate measurement scenario, but also the scenario where ACC and PPG are at the same frequency, as well as the scenario where the heart rate signal cannot be collected. Step S24 includes: The hidden state of heart rate measurement was obtained in S23. The selection rules for the heart rate spectrum peaks in the above 6 states are as follows: 1) Select the peak of the heart rate spectrum at the point of strongest PPG spectrum; 2) Select the heart rate spectrum peaks from the half-frequency or octave peaks of the PPG main frequency that differ the least from the previous moment. 3) Select the frequency point of the heart rate spectrum peak that has the smallest difference from the previous moment; 4) Select the point with the strongest ACC spectrum peak for heart rate spectrum; 5) Select the heart rate spectrum peaks from the half-frequency or octave peaks of the ACC main frequency that differ the least from the previous moment. 6) The heart rate peak selection should be consistent with the heart rate peak of the previous moment.

2. The method for measuring exercise heart rate as described in claim 1, characterized in that, Step S1, which processes the raw signals collected by the wearable device, includes: S11. Perform differential processing on the original ACC signal with a fixed window length to obtain the differentially filtered signal; S12. Perform median filtering on the differential filtered signal to obtain the median filtered signal; S13. Perform mean filtering on the median-filtered signal to obtain the mean-filtered signal; S14. Perform bandpass filtering on the mean-filtered signal to obtain the bandpass-filtered signal. S15. Perform CZT transformation on the sum of the bandpass filtered ACC signals to obtain the spectrum of the sum of the ACC signals. Then calculate the concentration of the ACC spectrum signal α and the maximum value of the spectrum within the passband P. If the concentration of the spectrum α is greater than the threshold and the maximum value of the spectrum P is greater than the threshold requirement, then the ACC spectrum signal meets the requirements and the ACC spectrum signal is obtained. S16. If the ACC spectrum signal meets the requirements, pass the PPG signal and the triaxial accelerations x, y, z through the RLS adaptive filter respectively. S17. Perform CZT transformation on the PPG signal to obtain its passband spectrum information; S18. If the ACC spectrum signal meets the requirements of step S15, perform spectrum subtraction to obtain the subtracted PPG spectrum signal.

3. The method for measuring exercise heart rate as described in claim 2, characterized in that, In step S11, the difference processing is a second-order difference processing, and the difference formula is as follows: y i =x i+2 +x i -2x i+1 Where i is the index of the signal, y i The signal after second-order difference, x i This is the input signal for the differential filter.

4. The method for measuring exercise heart rate as described in claim 2, characterized in that, In step S15, the method for calculating the concentration α of the ACC spectrum signal is as follows: Where, p i Th1 is the frequency point where the i-th peak of the ACC spectrum signal is located, and Th1 is the peak width threshold.

5. The method for measuring exercise heart rate as described in claim 2, characterized in that, In step 18, if the ACC spectrum meets the requirements of step S15, spectral subtraction is performed. The spectral subtraction method is as follows: Among them, f PPGC (i) represents the frequency i of the spectrum after spectral subtraction, f PPG (i) represents the spectrum before processing, f ACCN (i) is the normalized sum of acceleration spectrum, f PPGN (i) is the spectrum of the normalized PPG signal.

6. The method for measuring exercise heart rate as described in claim 4, characterized in that, Step S3, processing the peak value of the heart rate spectrum, includes: HR t γHR ref +(1-γ)HR current Among them, HR t The heart rate is a weighted average of the heart rate values ​​from the previous moment and the current moment. ref The selection rules for the reference heart rate value are as follows: 1) Calculate in real time the ratio β of the spectral energy of the fundamental, second, and third harmonic frequencies of the heart rate value corresponding to the frequency point in the PPG spectrum within the threshold bandwidth to the total spectral energy of [0.5,4]Hz. The calculation method is the same as the signal concentration α. 2) If β is greater than the preset value within the previous 5 seconds, then HR ref Set to the output heart rate value at that moment; otherwise, HR ref Set to the output heart rate value from the previous moment; γ is the heart rate confidence parameter. If the current β is greater than the preset value, γ is 0; otherwise, γ = β1 / (β1 + β2), where β1 is the reference heart rate β value and β2 is the current heart rate β value. current Current heart rate; HR th The threshold for the maximum allowable heart rate difference between adjacent time intervals before and after a heart rate change, i.e., HR. t If the threshold range is exceeded, it will be limited to an upper or lower limit. The selection of the value is related to the current exercise type and exercise intensity. The current exercise type is obtained through user input on the device or the exercise self-recognition module, and the exercise intensity is obtained through the combined acceleration intensity level of the three-axis ACC. HR is the final output heart rate value.

7. A device for measuring exercise heart rate, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 6.

8. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 6.

Images