Vessel Collision Risk Assessment Based on Evidential Reasoning Theory

Due to the flourishing growth of the marine economy and the high density of vessel traffic, busy and narrow waterways are facing the challenge of high potential collision risks. The safety aspects of maritime navigation have become important research problems. Timely and prompt collision alert is crucial for collision avoidance to ensure maritime safety and reduce casualties. The Collision Risk Incidence (CRI) is a metric that evaluates the probability of collision with other vessels. This paper presents a CRI assessment approach based on Evidential Reasoning (ER) theory to trigger prompt alert of potential collision vessels in the surveillance area. This paper takes into account the vital factors that are relevant to the value of CRI, and accurate real-time data of collision-involved vessels are acquired by the use of the Automatic Identification System (AIS). Furthermore, the Analytic Hierarchy Process (AHP) is employed to assign weight to these factors according to their significance. The membership degree of these factors for each dangerous level are obtained via similarity computation. We then apply the ER approach to assess CRI values, where the specific value of CRI is calculated by the explicit operation. Finally, simulation results confirm the validity of the proposed approach.