Bayesian network modelling for navigation status control of cargo ships in the Three Gorges Waterway

Abstract

This paper proposes a Bayesian network-based navigation status control model to estimate the probability of cargo ships being allowed to navigate in the waterway between the Three Gorges Dam and the Gezhouba Dam. The objective of the study is to develop a model to decide the navigation status of cargo ships by considering their attribute characteristics and the navigation risk factors related to the channel flow. Therefore, environmental condition, ship condition, and traffic complexity three modules are comprehensively considered in the proposed model. Specifically, the historical navigation data of cargo ships collected from the Three Gorges Navigation Administration, together with the navigation regulations of this waterway, are used to analyze the influencing factors of the navigation status and to construct the graphical structure of the Bayesian network. Moreover, the conditional probabilities of some nodes are determined using the Noisy-OR gate and the IF-THEN method. Finally, the navigation records are used to verify the feasibility of the navigation status control model for cargo ships, and the estimated average navigable probability of cargo ships is 89 %. The results indicate that the proposed model can reasonably estimate the navigation probability of cargo ships and assist maritime supervisors in controlling the navigation of cargo ships in the Three Gorges Waterway.