To address the complex and dynamic environments encountered by vessels in coastal waters, this paper proposes a dynamic group extraction method for addressing ship navigation risk based on spectral clustering.

Taking the waters of Xiamen Port as a case study, key information such as ship positions, speeds, and navigational states under different scenarios and timeframes was extracted from automatic identification system (AIS) data, enabling real-time calculation of spatiotemporal collision risk levels between each pair of vessels. A vessel conflict network was then constructed based on the computed potential risk values, forming a topological structure to describe risk distribution and vessel interactions within the waters. By integrating risk values and distances, modularity was introduced as a criterion for estimating the number of clusters. Spectral clustering was applied to partition vessels into risk groups with dense internal conflicts and sparse external interactions. Finally, a risk potential field of vessel groups was established to delineate hotspot areas, and the risk value of each group was further calculated to precisely identify these hotspots.

The results demonstrate that this method effectively reveals the spatial distribution of navigation risks in complex coastal environments through group clustering, enabling the timely and accurate identification of risk hotspots.

The findings will assist maritime authorities in comprehensively understanding real-time vessel dynamics and implementing preventive measures to enhance vessel traffic safety.