The Search Of Heart Rate Signal Extraction Based On The Pulse Wave

Heart rate is one of the important indexes to reflect the health of the human body.Heart rate monitoring can not only reduce the incidence caused by cardiovascular disease,but also have a good value for the people who usually exercise.In recent years,some sudden cardiac arrest and sudden death caused by tachycardia often occurs.For this reason,the heart rate monitoring can make the exerciser to understand their physical condition,and control the amount of exercise in time.At the present time,there are several methods to measure heart rate,the first one is calculated by the fluctuation of the pressure sensor.This method was applied in clinical medicine.The second one is to acquire the heart rate waveform from the electrocardiogram which is accurate and professional.The third one is to measure the variation of heart rate used by Photo Plethysmography(PPG).The former two methods were usually applied in clinic.Although the monitoring result is very precise,the method of monitoring is relatively complex,and the monitoring equipment is too huge to carry.Therefore,the method used PPG is getting more and more attention.In this paper,a real-time and efficient extraction method of heart rate was proposed.For the problem of movement interference,adaptive filter method was used.The three-axis accelerometer was used to acquire noise correlation signal,and after the preprocessing of signal,using acceleration signal as the reference signal,and gain adaptive filter parameters dynamically according to the state of motion to eliminate the noise,and monitor the heart rate in real time.The main research work is as follows:Firstly,in heart rate signal extraction technology,to test the human movement,the noise can roughly judge movement type,size,scope,and has a great influence on filtering effect of noise.But the variability of the human body characteristics and diversity making the system in the use of sensor number and contradiction between wearing comfort.How to use a small amount of perception in the premise of equipment and processing equipment and realize the motion characteristic parameters of convenient and reliable,accurate and nondestructive measurement and processing is still on the stage of research.In this paper,by using a single triaxle acceleration sensor to get the X,Y,Z three axis upward acceleration signal,the time domain,frequency domain analysis and information entropy for signal characteristics,and by using principal component analysis method to extract and identify the most relevant 25 characteristic elements,considering the wear life of equipment,choose a small amount of calculation of behavior recognition method that the decision tree method of motion recognition.Secondly,characteristics of human were diversity and complex,and were often mutual connection between multiple attributes.In the current study,often simply limit health monitoring system application environment to the human body static or resting state,even if there were for sports under the state of detection,will inevitably exist the certain error.In this paper,the use of photoelectric device photoelectric volume pulse wave were collected in the wrist,a series of pretreatment,an improved adaptive filtering method,comparative analysis of various parameters,based on the convergence property and location tracking system change ability of comparative judgment filter performance.According to the result of recognition motion and dynamic adaptive filter parameters choice,the movement to make the right judgment noise and offset,this method according to the specific situation of monitoring objects to provide users with more accurate,real-time monitoring services.The experiment result of the method we proposed has a good consistency with the medical monitor.Although the method used by decision tree cannot achieve the optimal identification rate of all kinds of sports,its relatively simple computational complexity can be more reasonable to meet the requirement of low power consumption and high endurance of the wearable devices.The adaptive algorithm we proposed in this paper has a higher accuracy than the traditional fast Fourier method.We can make a conclusion by analyzing three kinds of evaluation standard for the experiment result that the heart rate calculated by the algorithm proposed in this paper was more closely to the real heart rate,and the absolute error rate can be controlled within 3.3%.So in the allowed error range of wearable devices,it is possible to take advantage of the method proposed in this paper for noninvasive real-time heart rate monitoring.