The freeway free-tolling policy leads to the high traffic volume, and it also has some impact on the number of traffic accidents. With the informatization development of freeway operation and management, the traffic accidents are released to the public by the social network in time. Based on social-network information and some relevant statistical data (including vehicle ownership, freeway mileage, GDP (Gross Domestic Product) of Jiangsu Province, etc.), this study analyzes the characteristic of freeway traffic accidents, and explores different prediction methods to achieve the number of traffic accidents in future. Firstly, a sensitivity analysis on factors that might have an influence on traffic accidents is carried out. And then, two prediction models including a statistical regression model and a neural-network model are presented. Lastly, taking the freeways of Jiangsu Province as an example, the accuracy of two prediction models is evaluated. The prediction result shows that the neural-network model is more accurate comparatively. The prediction result could assist the operation and management of freeways.

CICTP