Abstract
The analysis of ship collision accident data plays a crucial role in ensuring maritime transportation safety. This paper proposes an intelligent visual analysis-based method focusing on the co-occurrence patterns, collision risk, similarity analysis, and classification of ship collision data. Based on data from 1573 maritime accidents that occurred globally between 2010 and 2023, this paper first designs co-occurrence patterns to identify regularities between accident attributes. Second, a fusion model combining fuzzy comprehensive evaluation and logistic regression is developed to calculate ship collision risk, and the model is applied to reconstruct and visually analyze the collision process. Third, the similarity between two collision accidents is analyzed based on Sentence-BERT algorithm, and a heterogeneous graph neural network model is constructed for accident classification. Finally, all the analysis contents are integrated into the visual analysis system of the ship collision accident data to illustrate the accident behavior and mechanism inherent in ship collision accident data. The rich content within the visualization system provides trainee crew members with a more comprehensive understanding of accidents and serves as a reference for maritime management agencies to analyze similar and related accidents, derive lessons learned, identify high-risk areas, and formulate targeted risk management strategies.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call