Risky Maritime Encounter Patterns via Clustering

The volume of maritime traffic is increasing with the growing global trade demand. effect growth especially observed in narrow and congested waterways as an increase ship-ship encounters, which can have severe consequences such collision. This study aims to analyze validate patterns risky encounters provide a framework for visualization model variables explore patterns. Ship–ship interaction database developed from AIS messages, interactions are analyzed via unsupervised learning algorithms determine using ship domain violation. K-means clustering-based novel methodology among encounters. applied long-term dataset Strait Istanbul. Findings support that length speed be used indicators understand Furthermore, results show site-specific risk thresholds ship–ship determined additional expert judgment. mid-clusters indicate violation grey zone, should treated carefully rather than bold line. approach integrated safety measure authorities use decision tool.