Wearable Real-time Based On Multi-sensor Fusion Research On Heart Rate Monitoring Technology

With the rapid development of sensor technology,wearable devices have been applied to many research fields.The application of wearable devices to monitor physical signs can analyze the physical and mental health of individuals objectively.Heart rate is an important physiological indicator of the human body.It can evaluate the activity of the heart and the degree of fatigue.It can be used to evaluate emotions by monitoring heart rate changes and voice recording or breathing.However,single-lead or multi-lead equipment is often used to monitor ECG in medical treatment.To calculate the heart rate,there is inconvenience in daily life.Photoplethysmography(PPG)is a method that uses PPG sensors to reflect light onto the skin to measure changes in light intensity.It can be placed in earlobes,fingertips or wrists for long-term non-invasive heart rate monitoring.PPG signals collected by wearable devices are difficult to obtain PPG signals that are not interfered by motion artifacts(MA)due to the subject’s voluntary or involuntary movement.These MAs will affect the calculation and monitoring of heart rate seriously.This paper investigates and compares the performance and real-time performance of various denoising algorithms.This paper proposes to use Recursive Least Squares(RLS)algorithm to fuse three-axis acceleration signals to remove motion artifacts from PPG signals.In order to further determine the heart rate frequency range and select the appropriate spectrum peak,a decision tree heart rate spectrum interval estimation method is proposed to provide prior information for PPG signal spectrum tracking,so as to optimize the spectrum peak,which can be processed in real time on a wearable device.The main research work of this paper is as follows:(1)Design a multi-sensor realtime heart rate monitoring bracelet,using wearable devices as the hardware platform and NB-IOT as the system architecture to build a real-time heart rate monitoring system;(2)Based on three-axis The acceleration data is adaptively filtered to the PPG signal,and the low-order RLS adaptive filtering algorithm is adopted,which has high real-time performance and is embedded in the wearable device to run;(3)The heart rate change trend is estimated based on the acceleration characteristics,and the decision tree is used to determine the heart rate interval.Make decisions to provide a priori criteria for spectrum tracking,which effectively improves the accuracy and real-time performance of spectrum selection;(4)The heart rate estimation value is obtained by fusing the results of multiple filters,which effectively avoids the difference between the single-axis acceleration frequency and the actual heart rate frequency.(5)This article uses the designed wearable real-time heart rate monitoring system to collect 12 sets of continuous heart rate monitoring data in daily life.The experimental scene is the switching of 4 states in the daily life learning area,including standing,sitting,walking and Running,mixed with some non-periodic exercises such as typing,flipping through materials,etc.In addition,12 public data sets are selected as supplementary data sets.The analysis on the above-mentioned data sets has obtained a high accuracy rate,which ensures that the realtime operation is embedded in the wearable device while being within the error tolerance range.