Study On Feature Extraction And Recognition Of Visual Fatigue EEG Signal

Working in a fatigued state can seriously affect people's health and life safety,also have a negative impact on the people they serve.Therefore,it is of great significance to objectively understand fatigue and monitor it.The brain is the core of the central nervous system,and changes in its electrical activity reflect changes in the state of fatigue.Therefore,the use of electroencephalogram(EEG)signals to identify fatigue is an effective method.Compared with traditional fatigue identification techniques based on questionnaires and facial expressions,the identification of fatigue signals based on EEG signals is more reliable.The generation of EEG signals includes spontaneous and evoked.Steady-state visual evoked potentials(SSVEP)have become the focus of research because of their simple operation,high information transfer rate,and no need for training.The present study is based on the visual fatigue EEG signal generated by SSVEP.This article extracts features of EEG signals during the awake and fatigue phases to find feature vectors that can efficiently identify visual fatigue states.The main research work of the article is as follows:1.In this paper,a fatigue experiment based on SSVEP is designed.The two phases of awake and fatigue in the experiment are used to study the detection of fatigue EEG based on SSVEP.Twenty subjects watched repeated visual stimuli for a long time to produce visual fatigue EEG signals.Subjective fatigue scores were recorded before and after the experiment as a reference for subjective statistics.An independent component analysis(ICA)method was used to preprocess the collected EEG signals to remove artifacts such as electrooculography(EOG)and electromyography(EMG).2.Power spectral analysis and sample entropy method were used to extract the frequency domain characteristics and nonlinear characteristics of visual fatigue EEG signals.In terms of frequency domain characteristics,it was found that the power of alpha,theta rhythm and the ratio of alpha/beta,(alpha+theta)/beta had significant differences in fatigue and awake phases.In terms of nonlinear characteristics,entropy is a measure of the complexity of the chaos level that reflects the chaotic nature of timeseries signals and multi-frequency components.This paper compares the characteristics of approximate entropy and sample entropy that chooses sample entropy as the feature of fatigue EEG.Similarly,the value of sample entropy shows significant differences in the fatigue and awake phases,and the sample entropy value in the fatigue phase drops significantly.3.Using support vector machine(SVM)as a classifier of the fatigue EEG signal,the five features of the three leads were combined into a 15-dimensional feature vector and passed to the SVM.The results showed that the visual fatigue feature parameter has good separability,the classification accuracy rate is more than 90%,and it can be used as an indicator to judge the fatigue status.