Accelerating behaviour of the driver and acceleration characteristics of the dangerous event

The accelerating behaviour of the driver and the relationship between the extreme acceleration and dangerous events are studied by using the naturalistic driving data. Firstly, the bivariate accelerating patterns between the longitudinal acceleration, lateral acceleration and velocity are proposed. Next, the univariate accelerating behaviour characteristics are achieved. Then, the extreme driving data in 0.01st density contour outer area are analysed. In the last, the driving style is studied based on the extreme acceleration data. The conclusions include: the univariate acceleration distributions approximate the Pareto distribution, which is known as the heavy tail distribution. Most of the extreme acceleration data are not dangerous events. And the majority of the dangerous events are in the high brake deceleration region, whereas high lateral acceleration does not have apparent relationship with the dangerous event. Aggressive drivers account for most of the extreme acceleration data, and they are more likely to be in danger.