Modelling the Collision Risk in the Yangtze River Using Bayesian Networks

Ships navigating in the traffic separation scheme pose serious risk owing to the distinguishing characteristics of narrow channels, such as limited depth, dense traffic, and disturbance of crossing ships. This paper models the collision risk in the Yangtze River by considering both the causation factors and the emergency management of maritime accidents using historical data. First, more than one hundred collision accident data are collected for the period of 2009 and 2012. Second, a Bayesian network is proposed to model the collision risk of ships in the Yangtze River, while the qualitative part is established by domain experts and previous works, and the quantitative part is developed based on historical data. Third, the collision risk is compared with the navigational risk in Tianjin Port. The findings are beneficial for the safety management of sailing ships in the Yangtze River.