Effective Freeway Incident Response: A Bayesian Network Based Algorithm

Prompt and accurate incident response plan is an essential part of an effective freeway incident management system. However, uncertainties associated with the incident often lead to delayed or inaccurate response decision and processing plan. The core of the proposed algorithm is a Bayesian network based decision-making system that quantifies the causal dependencies between response decisions and incidents information. With input from real time incident data, the Bayesian networks update the probability of uncertainties associated with highway incidents through a two-way inference mechanism. We apply the proposed method to the incidents data recorded in Nanjing-Lianyungang freeway from 2001 to 2004 combined with more complete incidents data recorded in 2005 obtained from RK-IMS (a rule-based freeway incident management system). The model was validated using incidents data of the year 2006 from RK-IMS. The comparison results indicate that the proposed algorithm is accurate and reliable.