Research On Recognition Method Of Driving Fatigue State Based On Kernel Principal Component Analysis

The driver’s fatigue driving behavior has a serious impact on the traffic accidents.The Electroencephalograph(EEG)signals in the no-fatigue state and fatigue state of the drivers studied and analyzed in this paper.In view of the nonlinear characteristics of EEG signals collected in the fatigue driving state recognition research,this makes the recognition accuracy of the fatigue driving state recognition method of EEG data based on entropy characteristics is still unsatisfactory,this paper based on EEG of entropy characteristics and kernel principal component analysis(KPCA)establishes the model of fatigue driving state recognition,and proposed the corresponding algorithm;in terms of the issue of the time performance of fatigue driving state recognition,t-test to further study based on the above proposed algorithm added in this paper,and propose different entropy characteristics and t-test combined with KPCA algorithm(called ENTROPY_T_KPCA),and the research on 30-electrode data has achieved the desired effect.The main research contents and innovations of the paper are as follows:(1)In view of the recognition accuracy of the fatigue driving state recognition method of EEG data based on entropy characteristics is still unsatisfactory.First,the reconstruction of EEG data by different entropy combined with principal component analysis(PCA)studied in this paper;secondly,the reconstruction of EEG data by different entropy combined with KPCA studied in this paper;finally,support vector machine(SVM)classifier is used to tested on the reconstructed data.The test results show that the recognition accuracy of the fatigue driving state after using the KPCA algorithm is significantly improved compared with the unused KPCA algorithm.(2)In terms of the issue of the time performance of fatigue driving state recognition is still unsatisfactory,t-test based on previous proposed algorithm added in this paper.The t-test is used to find the feature vectors with significant differences,and the time performance is improved by feature reduction.The experimental results show that the t-test algorithm is effective,After adding the KPCA algorithm,theENTROPY_T_KPCA method can not only improve the time performance but also improve the recognition accuracy of the fatigue driving state when the appropriate classifier is selected.