Exploring the Different Patterns for Generation Process of Driving Fatigue Based on Individual Driving Behavior Parameters

Zhibiao Peng,Jian Rong,Yiping Wu,Chenjing Zhou,Yuan,Xiansheng Shao
DOI: https://doi.org/10.1177/0361198121998351
2021-01-01
Transportation Research Record Journal of the Transportation Research Board
Abstract:Driving fatigue is one of the main causes of traffic accidents in monotonous environments such as grassland highways. However, the process of generation of driving fatigue on grassland highways is still not clear. A driving simulation experiment with 23 participants was performed to collect data on driving behavior, reaction time and electrocardiogram (ECG) results when driving on a grassland highway. The effective feature indicators of driving fatigue based on driving behavior data were calculated by Pearson correlation coefficient and principal component analysis method. The matter-element model based on entropy weight method was used to quantify the generation process of driving fatigue (GPDF). GPDF was classified as different patterns by the eigenvalue of GPDF curves. Reaction time and ECG data were utilized to verify the rationality of GPDF. Results show that there were 13 feature indicators of driving behavior suitable for driving fatigue description. GPDF was not completely consistent among different participants and was classified into three patterns (i.e., mild, moderate and severe fatigue). The mean similarity for GPDF in each pattern was 0.87, 0.61 and 0.50. Validation test demonstrated that driving fatigue detection accuracy by GPDF was 72%. The mean similarity of the GPDF between driving behavior and ECG was 0.72. Driving fatigue tended to occur with driving time of 19 min or 33 min. This study is helpful to understand GPDF on grassland highways from the perspective of individual driving behavior, which would provide suggestions for the reasonable setting of anti-fatigue devices.
What problem does this paper attempt to address?