Research On Key Technology Of Non-contact Sleep Monitoring System

Sleep is an immeasurable life activity of human beings.As the pace of modern life accelerating and work pressure increasing,many people will be implanted by sleep-related diseases.Therefore,it is necessary to detect sleep conditions.Traditional sleep monitoring methods mainly rely on contact sleep monitoring instruments to monitor the human body.These monitoring instruments are complicated to use,costly,numerous electrodes,and restrict personal freedom.Kind of shortcomings.Therefore,the study of a non-contact sleep monitoring instrument is of great significance for the diagnosis and treatment of sleep-related diseases.This paper designs a non-contact sleep monitoring system based on the principle of Doppler radar.The hardware part of the system mainly includes radar acquisition module,electronic control module and communication module.According to the principle of Doppler radar,the radar acquisition module realizes the acquisition of respiration and ECG signals during human sleep.The electronic control module uses the STM32L431 microprocessor as the control core,designs the voltage conversion circuit and the communication circuit,and completes the control of the radar front end and data transmission.The software part includes BP neural network sleep staging model establishment and human-computer interaction.BP neural network sleep staging model mainly includes feature extraction of heart rate variability,sleep staging model establishment and accuracy evaluation.Human-computer interaction mainly includes two parts: user login registration and sleep monitoring.Users can register their personal information and log in to the system to check their night sleep conditions,such as falling asleep time,deep sleep time,sleep staging,etc.;make the sleep status digitized,easy for users to understand and use.Finally,the software part is generated into an installation package.Sleep staging is the standard and prerequisite for measuring sleep quality.This paper first uses statistical methods to extract 15 characteristic parameters from the time domain characteristics of heart rate variability,including the mean value of the RR interval,standard deviation,and heart rate variation coefficient;using wavelet transform methods,extracts from the frequency domain characteristics of heart rate variability 6 frequency domain indicators such as low frequency power,high frequency power and total power;and 8 characteristic parameters including sample entropy and de-trend fluctuations in the non-linear domain.Secondly,the characteristics of heart rate variability combined with the characteristics of breathing signals and the number of body movements are used as the characteristic indicators of sleep staging,and the BP neural network algorithm is used to establish a sleep staging model.Using confusion matrix and ROC curve two evaluation indicators to evaluate the accuracy of the model,comparing the staging results with the standard staging results,the final accuracy of the model is 83.2%.Through the calculation of human physiological parameters and the judgment of sleep staging by this non-contact sleep monitoring system,the results provide a certain reference value for clinical judgment of sleep-related diseases.