A Driving Fatigue Feature Detection Method Based on Multifractal Theory

Driving fatigue seriously threatens traffic safety. In our work, the multifractal detrended fluctuation analysis (MF-DFA) method is proposed to detect driver fatigue caused by driving for a long time. First, the theta (4 similar to 7 Hz) and beta (14 similar to 32 Hz) subbands of subjects' electroencephalogram (EEG) signals were extracted. Furthermore, the multifractal spectrum characteristic indexes, the fluctuation function, mass exponent, Hurst exponent, spectral width values, and symmetry, were analyzed. Finally, the characteristics of different driving states of subjects were compared and analyzed. The results show that there are significant differences in the Hurst exponent width values of the subjects' EEG signals corresponding to different driving times, and the spectrum width values and symmetry of the multifractal spectrum. In addition, compared with several typical fatigue detection methods, the absolute value of the slope of the fit straight line for fatigue feature extraction by MF-DFA is larger. It means that the MF-DFA method is more effective in detecting driving fatigue.

Automated summary

Driving fatigue poses a serious threat to traffic safety. In this study, a new method called multifractal detrended fluctuation analysis (MF-DFA) is proposed to detect driver fatigue. The method analyzes the characteristics of electroencephalogram (EEG) signals and shows significant differences in the features of EEG signals corresponding to different driving times. The results demonstrate the effectiveness of the MF-DFA method in detecting driving fatigue.