Author(s)

Jie Wang, Yanan Zhao and Li Gao

Corresponding Author

Yanan Zhao

ABSTRACT

Aiming at the scientific problem of complex driving environment representation, based on the analysis of road traffic accidents in China, the influencing factors of collision accidents are dissected from four aspects of people, vehicle, road and environment. Based on the data of collision dangerous conditions extracted from the natural driving data of dangerous goods transport vehicles, 12 driving behaviors are identified in comparison with 37 types of pre-crash scenarios summarized by NHTSA. The data analysis system architecture of dangerous condition is built from three aspects: driving dangerous road, driving environment and driving operation. The K-means clustering algorithm is optimized to quantify and analyse the dangerous collision scenarios of dangerous goods transport vehicles from the perspective of time and space, and to design test parameters and correct the types and spaces by combining Euro-NCAP, C-NCAP and SAE test standards. Finally, the active crash prevention and control performance of dangerous goods transportation vehicle under different dangerous working conditions is tested in the test site of the Ministry of Transportation to verify the environmental adaptability of the system. It effectively solves the problem of difficulty in quantifying and lack of realism of the automatic driving simulation scenarios and closed road test scenarios.

KEYWORDS

Dangerous goods transport vehicles, driving behavior, collision dangerous condition, cluster analysis