Study On EEG Fatigue Detection Of Wearable Electroencephalograph Based On Machine Learning

The detection and application technology of electroencephalogram(EEG)has gradually improved in recent years.With the development of wearable electroencephalograph,breakthroughs have been made in the management of brain health.Through the monitoring and analysis of EEG,the fatigue degree of human body can be obtained,and the chronic diseases such as cardiovascular and cerebrovascular diseases can be monitored and prevented more efficiently.This paper provides most recent study on fatigue state detection based on support vector machine(SVM),deep belief network(DBN)and generalized regression neural network(GRNN)for the intelligent recognition of the fatigue state of EEG with single-electrode wearable electroencephalograph.Through this study,the intelligent analysis of EEG and the management of human health is realized by accurately identifying fatigue state and providing timely health warning and adjustment scheme.First of all,we built the data sets by using the questionnaire to investigate the Karolinska sleepiness scale and exploiting the fatigue detection bracelet to obtain the fatigue level marking of EEG data.Secondly,we did data preprocessing and then extracted features from both the time domain and the frequency domain.Principal component analysis was used to reduce the dimension of the data.Thirdly,the fatigue recognition model was established by GRNN and the recognition accuracy was calculated.Meanwhile the SVM method was used to compare withour test model.Thirdly,the fatigue detection model was established by SVM,DBN and GRNN,and the recognition accuracy was calculated,and the test results of the model were compared.Finally,the real time fatigue detection was carried out under the established GRNN model.Experiments showed that the maximum recognition rate of EEG fatigue state is 88.1% under GRNN model.Compared with SVM model and DBN model,GRNN model has higher recognition accuracy,faster calculation speed,greater stability and better discrimination for different fatigue degrees.Compared with SVM mode,DBN model has higher recognition accuracy and better discrimination for different fatigue degrees,but slower calculation speed.Wearable electroencephalograph can realize real-time detection of EEG fatigue state,which plays an important role in health management and disease prevention.