In order to accurately identify factors affecting the occurrence and severity of crashes on ice-snow covered freeways

an integrated model based on fault tree and Bayesian network was developed. Three directed arcs were added into the transformed Bayesian network. Leaf nodes were divided into three states according to the crash severity

and the conditional probability table of the leaf nodes was updated. Bayesian network reverse reasoning and sensitivity analysis were carried out based on the proposed integrated model. Results show that high risk factors including low visibility

adverse weather (rain

fog

snow)

trucks

non-lighting at night

lack of driving experience

speeding

and insufficient headway mainly induced the occurrence of crashes on ice-snow covered freeways. Overloading

trucks

and illegal driving tended to increase the crash severity under ice-snow conditions. The integrated model of fault tree and Bayesian network is expected to provide a new perspective for the analysis of crash factors.