A probabilistic consequence estimation model for collision accidents in the downstream of Yangtze River using Bayesian Networks

Collision is a major type of accident in maritime transportation, which in the downstream of Yangtze River is even more pronounced due to specific features that have significant impact on the collision consequence, including a special lane for small-sized ships, traffic intensity variation with the tide period, many restricted areas, and emergency resources spread along the river. This article models the collision consequences in the downstream of Yangtze River using Bayesian Networks, considering the causation factors and including a novel approach for the emergency management of maritime accidents. The graphical structure lies on domain experts and on previous works, while the conditional probability tables are developed from historical data. Both the graphical structure and parameters are validated using the well-known methods, which reflects that the developed model is reasonable. The merits of the proposed consequence estimation model that considers emergency management includes (1) a detailed description of the collision accident development and (2) consequence estimation result with good accuracy.