Comparative analysis of risk factors and scenarios for different types of ship contact accidents using a data-driven Bayesian network model

15 de fevereiro de 2026

This paper develops a data-driven Bayesian network model to investigate the individual and combined impact of risk factors on ship contact accidents involving terminals, bridges, reefs, and other objects. Based on a dataset of 234 ship contact accidents from China, the United States, Australia, and the United Kingdom, the model's structure and parameters are derived from the Tree-Augmented Naive Bayes and the Expectation-Maximisation algorithms, respectively. Subsequently, the developed model is assessed through a comparison between learned parameters and statistical data, axiomatic validation, k-fold cross-validation, and two subsample tests. A comparative analysis of different types of ship contact accidents is conducted by identifying high-frequency risk factors through diagnostic reasoning, high-sensitivity risk factors through sensitivity analysis, and high-risk scenarios using multi-scenario simulation. The research results indicate that (1) human factors, such as deficient safety awareness and inadequate technical experience, and insufficient risk perception and contact hazard assessment, are identified as high-frequency risk factors; (2) spatiotemporal factors, such as areas with contact accidents, and ship factors, such as operation, type, speed, and width, are identified as high-sensitivity risk factors; (3) the combined effects of key risk factors are investigated through the most probable explanation analysis and multi-factor joint scenario simulation. Moreover, the comparative analysis reveals differences in the high-frequency risk factors, high-sensitivity risk factors, and high-risk scenarios for different types of ship contact accidents. These findings have crucial implications for formulating both universal and tailored safety strategies for ship contact accidents.

Autores da comunidade :

• 

  C

  Carlos Guedes Soares

  ist11869

  Afiliações e fontes

  IST > CENTEC

  Centro de Engenharia e Tecnologia Naval e Oceânica

VoR - Versão publicada

Elsevier

https://www.sciencedirect.com/science/article/abs/pii/S0029801825036273

Ocean Engineering

123945

347

0029-8018

10.1016/j.oceaneng.2025.123945

other-engineering-and-technologies - Outras Ciências da Engenharia e Tecnologias

eng - Inglês