Traffic Flow Risk Analysis on Expressway Straight Segments Using Drone Video

In order to understand the changes of freeway traffic flow risk，drone videos was used to obtain vehicles trajectories on the freeway, analyzing the spatio-temporal interactions between vehicles, the propagation patterns of traffic conflicts, and the pattern of risk changes. Classify traffic flow states based on three-phase traffic theory. Starting from the frequency and severity of conflicts, the risk characteristics under different traffic flow states was investigated. The fuzzy C-means clustering algorithm was used to determine the risk level. Results indicate that the vehicles in the first lane on the left were more sensitive to the speed changes of the leading vehicles. The deceleration wave is highly consistent with the propagation path of traffic conflicts. When the backward propagation of deceleration waves, the collision risk also propagates backward simultaneously. In the process of transitioning from free flow to synchronized flow, high-risk state accounts for the highest proportion, about 30.5%. The findings of this paper provide quantitative technical support for active traffic safety management on freeways, holding certain significance in aspects such as traffic flow management and speed control.