Sleep Monitoring Recognition Research Based On Ballistocardiography

People spend one third of their lives asleep.Sleep is an essential behavior for human beings and is required for survival.At the same time,the quality of sleep is closely related to the growth and development of our human beings and their longevity.Polysomnography(PSG)is a commonly used sleep monitoring method,but it's also complex,costly and affects normal sleep.Therefore,this thesis uses a sleep monitoring method based on Ballisticocardiography(BCG)signal,which originates from blood pumped by the heart.This signal is derived from the blood flow caused by the heart pumping blood,and impact force is generated on a supporting body that is in contact with the human body.Physiological data can be collected by non-direct contact,which can include heartbeat,breathing,body movement,etc.In addition,the measurement accuracy of this method is also significantly better than that of the wristband type wearable device,so the sleep monitoring method based on the BCG signal has both user experience and accuracy.The main work of this thesis includes the following three parts:(1)Extraction of physiological parameters of BCG signal: In order to not affect normal sleep,this thesis has designed a set of BCG signal acquisition program,using non-contact,noninterfering single device piezoelectric thin film sensor to collect BCG signal,and extract by analog-digital circuit conversion.Body motion value,based on local extreme point method to extract heart rate,extract respiratory value by wavelet profile curve,and verify the collected signal.(2)Sleep position detection based on BCG signals: Sleep quality is closely related to sleeping position,and poor sleeping posture may even exacerbate the potential risk of multiple diseases.In order to monitor sleep health more precisely,this thesis proposes a sleeping posture pattern recognition algorithm based on BCG signals.BCG signals from four different sleeping positions(supine,prone,left lateral and right lateral)were collected with a piezoelectric film sensor in a non-contact way.Two kinds of characteristics were extracted from the preprocessed BCG signals after wavelet-based denoising,and the sleep position pattern was recognized using BP neural network,and compare the two features to extract the classification effect.(3)Non-EEG sleep staging based on BCG signals: Studies have shown that sleep staging has important clinical significance for the detection,prevention and treatment of sleep disorders,clinical use of PSG for sleep monitoring,but the operation is complex and costly.In order to overcome the PSG method to the subjects to bring binding issues,this thesispresents a non-EEG sleep staging method based on the BCG signal,heart rate,respiratory and body motion values obtained using the physiological parameter extraction algorithm of the above BCG signal.First step,we verified the sleep stage based on body motion values and the sleep stage based on heart rate transform;the second step,we designed a multi-physiological sleep stage based on convolution neural network classifier.