Research Of The Mental Fatigue Classification Based On HHT And Kernel Method

In modern society, intense social competition and rapid pace of life lead to serious physical and psychological pressure which is seriously harmful to people's health. Meanwhile, with the development of modern industrial automation, the increased mental labour caused more and more industrial accidents. In addition, traffic accidents caused by mental fatigue brought great hazard to people's lives. Therefore, the research of mental fatigue does not only have important medical value, but also have the great social value and economic value. In recent years, with the rapid development of EEG analysis technique, the recognition of mental fatigue has made a great progress.Firstly this thesis designs the EEG acquisition and wireless transmission system based on PSoC3, and then finishes the EEG data collection through the fatigue test. Secondly, this thesis analyzes entropy feature of four channel EEG(O1,O2,Fp2,Cz) with the HHT method, and then brings up the EEG feature vector of approximate entropy,EMD decompositions approximate entropy,Hilbert marginal spectrum and energy spectrum entropy. At last, the Kernel Fisher Discriminant Analysis and linear SVM method are combined to recognize the feature vector, and the effectiveness of mental fatigue recognition is discussed through comparing with other classification method.