Study On Mental Fatigue Detection System Based On Wireless Brain Computer Interface

Mental fatigue is due to long-time physical,mental work,or accumulated mental pressure,People with mental fatigue show bad sleep quality,poor memory,reduced executive ability,easy weary feeling for the work and so on.Mental fatigue is a common physiological phenomenon and a protective reaction of the human body.If a person shows a long time and repetitive mental fatigue,it will seriously affect his/her health,and even cause accidents during the work in some conditions.Thus,it is urgent to prevent this kind of tragedy or alleviate fatigue problems.Currently,there is still a lack of portable and wearable mental fatigue detection system,which can effectively detect the state of mental fatigue.In this paper,a wireless brain computer interface for mental fatigue detection system is studied.The study mainly includes the following aspects:1)Feature extraction method and feature selection of the EEG signal are studied.Six different EEG signal features were extracted by using energy spectrum analysis.We examine the effectiveness of these features in three fatigue states according to the results of the three-class discrimination and the three kinds of two-class classification experiments.It turns out that the feature of(θ+α)/β could obtain the highest classification accuracy in both the three-and the two-class discrimination experiments.2)The method of classification of mental fatigue states is studied..In this paper,a semi-supervised classification algorithm is applied for the classification and recognition of mental fatigue.For semi-supervised learning method,only a few labeled EEG data is needed to train the initial classifier.Then a large number of unlabeled EEG data are used to update the classifier.It could shorten the training time of the subjects and improve the classification performance and generalization ability.In this paper,an improved self-training method is proposed by adding classifiers and filtering steps to select unlabeled EEG samples and extend the training set.The improved self-training algorithm is compared with the traditional self-training algorithm and the Fisher discriminant analysis based on the classification performance.The results show that the improved self-training algorithm is more effective than the traditional self-training algorithm.Though the classification accuracy of the improved self-training algorithm is not significantly improved compared with that of supervised FDA method,the number of labeled samples needed for the classifier training is greatly reduced.3)The mental fatigue detection system based on wireless brain computer interface is designed.In this paper,In this paper,the detection system is composed of wireless EEG signal acquisition and online detection module.The wireless EEG signal acquisition realizes real-time data acquisition and wireless transmission of multi-channel EEG signals.The online detection of mental fatigue achieves the pre-processing,storage of the EEG signal,the feature extraction,the online evaluation of mental fatigue and the display of the corresponding result of each step.Compared with the existing detection systems,the system proposed in this paper has the advantages of lower power consumption,less noise,but better portability and real-time performance,which are consistent with the requirements of the intelligent devices.4)The mental fatigue detection system is tested in real-time.Based on the previous studies of the algorithm and the study of the mental fatigue detection system,the online test of mental fatigue is designed and implemented.In this study,the mental fatigue states of two subjects are tested for five consecutive days.This system takes only 1 minute to output the results of the mental fatigue state.The experimental results show that the online detection accuracy of discriminating two-class mental fatigue state is up to 85%,which indicates that the system designed in this paper has realized the real-time detection of the mental fatigue states and reached the expected design goal.