Multi-ship collision risk assessment based on DBSCAN algorithm

The evaluation of collision risks in maritime traffic is essential for guiding strategic evasive actions and decreasing the incidence of collision-related events. In this scholarly work, we introduce a maritime collision risk assessment model that employs the DBSCAN clustering methodology. This model is capable of assessing the potential for collisions in situations where numerous vessels encounter. In this model, the DBSCAN algorithm is used to obtain the collision cluster of ships, and a quantitative index based on collision avoidance rules and ship domain - danger sector is used to represent the collision risk of each ship in the case of multi-ship encounters. AIS data was applied to validate the proposed risk assessment method, and the findings demonstrated that the proposed model is proficient in discerning the likelihood of multi-vessel collisions, thereby offering a foundational theoretical framework for the surveillance of navigational safety in intricate maritime environments.