A Situation Assessment Method of Traffic Incident Based onBayesian Networks

摘要

Severity estimate of traffic incidents is the basis of traffic incident management. This paper combines the situation assessment technology with Bayesian networks in traffic incident management. First, the concept of situation assessment and Bayesian networks theories is introduced. Then a traffic incident situation assessment model is developed based on Bayesian networks for projection of future status. Structure and parameters of this network are gathered from field data in 1-880 Freeway using K2 and EM (expectation maximization) algorithms. Finally this model is evaluated with the real data. The results indicate that Bayesian network is an effective method.