In this paper, Electroencephalogram (EEG) based driving fatigue detection by using the Topological Data Analysis (TDA) was presented. The experimental obtained EEG activity signal was marked by combining the KSS (Karolinska Sleepiness Scale) results and recorded facial video for different participants, to form a binary data set of driving fatigue state and waking state. The EEG signal was preprocessed by using the EEGLAB, and the noises were carefully eliminated and a 0.3~30Hz frequency band was retained. The extracted feature of different driving states was classified by using the Support Vector Machine (SVM). The EEG data was also processed by using the Power Spectral Analysis (PSA) for the comparison of performance and robustness. The results show that the accuracy of 88.7% and the recall rate of 92.5% were obtained by using the TDA method. The performance was agreed very well with the PSA approach. The topological features were significantly robust to the EEG results, where the changes of identification accuracy of the TDA approach were less than 8%. The proposed TDA method for driving fatigue EEG signal process and analysis has good anti-interference characteristics, low data processing cost, and high economy. It is helpful for stable and efficient detection of driver fatigue state and has a great scientific practical application value.

Zhejiang University of Technology: Zhengqing Liu, Feiyang Zhou; Zhejiang Jiuzhou New Energy Technology Co Ltd: Bingbin Zhu; RMIT University: Jinhui Xu, Mohammad Fard