Maritime Accident Risk Probability Prediction of Sea Lanes Based on Dynamic Bayesian Network

The safety of maritime transportation has become increasingly important with the development of international economies and trade. Maritime transport poses risks that may cause damage to crew members, ships, and the environment. An improved Dynamic Bayesian Network (DBN) model is proposed in this paper to predict the dynamic probability of emergencies risk of sea lanes. It is a novel model which can efficiently represent and inference the complex stochastic knowledge. To develop the DBN-based model, the data, which were collected from emergencies investigation reports by International Maritime Organization (IMO), is used to provide guidance for the construction of BN. Second, the prior probability is determined by Evidence Theory model based on historical data. Third, the conditional probability is learned by EM algorithm and the transition probability is obtained by Markov model. Finally, Viterbi algorithm is adopted to predict emergencies risk probability. The emergencies occurred in Indian Ocean from 2009 to 2018 was used as a case study for risk probability prediction. The sensitivity analysis was carried out to identify the significant influencing factors. The results show that risk of sea lanes in Indian Ocean fluctuates within a small range but a downward trend overall. The significant influencing factors include wind speed, waves, visibility and pirate attacks. The findings can be used to provide a reference for maritime stakeholders to make proper decisions.

Language

  • English

Media Info

  • Media Type: Digital/other
  • Pagination: pp 157-171
  • Monograph Title: Proceedings of the International Forum on Shipping, Ports and Airports (IFSPA) 2019: Beyond Breakthroughs, Above Excellence

Subject/Index Terms

Filing Info

  • Accession Number: 01744567
  • Record Type: Publication
  • ISBN: 9789887408406
  • Files: TRIS
  • Created Date: May 11 2020 11:20AM