Design Of Sleep EEG Signal Acquisition And Processing System

Sleep is so important to our lives that the average healthy adult spends about a third of their day sleeping.Therefore,the quality of sleep is related to the health of life and the quality of life,enhance the attention to daily sleep,can effectively prevent the impact of poor quality sleep on our life.Compared with the sleep quality test in the hospital,the home-based sleep monitoring instrument is more easily accepted and selected by the public due to its convenience,immediacy and low cost.In many biological signals,eeg carries information for analyzing sleep quality has very good effect,its principle is through the obtained from the collected eeg signals can reflect the characteristic value of the characteristics of the various sleep stages,using appropriate classification algorithm of brain electric signal accurate classification,finally combining classification results to evaluate the tester sleep quality.Based on the above principles,this paper designs a sleep quality assessment system covering eeg signal acquisition,processing,feature extraction and sleep stages.The main research contents of this paper are as follows:(1)The hardware acquisition equipment of EEG signals was built to facilitate the acquisition of EEG signals,highlighting the immediacy and flexibility of the system.The hardware part of the system USES TGAM module to complete the acquisition of EEG signals,and then USES the signal acquisition board with STM32f103 as the main control chip to complete the signal analog to digital conversion,as well as data transmission and other functions of the upward computer.After the actual test of the equipment,the eeg signals collected by the hardware can truly reflect the time-frequency characteristics of human sleep EEG.(2)In the feature extraction of EEG signals,the feature vector of EEG signals is composed by the energy feature of eeg rhythm wave obtained by wavelet packet decomposition combined with the feature value obtained by KC complexity and approximate entropy in the nonlinear dynamics method.Experiments show that the time-frequency analysis combined with nonlinear analysis can more effectively show the characteristics of eeg signals at different stages.(3)Compared the classification effects of two optimization algorithms based on SVM classifier,namely C and G cross validation and particle swarm optimization,and selected particle swarm optimization algorithm to optimize SVM classifier according to the results;The training data were collected from experiments on electroencephalogram(EEG)sleep signals published by the Physio Net website.A suitable classification model was developed by combining the accurate staging results obtained from the data.According to the training results,the model generalization ability after the classifier training reaches more than 80%.(4)The fuzzy logic algorithm is used to process the staged data and provide the sleep quality of the corresponding testers.Combined with the experimental test results,the sleep quality analysis results obtained by the system also accord with the actual situation,and the system basically meets the design requirements.