A Hierarchical Clustering and Personalized Potential Field-Based Trajectory Repair Model for Small Coastal Vessels

High-integrity vessel trajectory data are essential for navigation behavior analysis and maritime safety management. This paper proposes a hierarchical clustering and personalized potential field (HCPPF) trajectory repair model to address long-duration AIS outages in small coastal vessels operating in near-shore waters. AIS voyages are first grouped through OD-based DBSCAN and in-cluster spectral clustering to obtain representative navigation modes. A mode-specific potential field and a gradient-guided APF-A* search are then used to reconstruct spatial gaps, and temporal information is subsequently recovered using mode-level speed profiles to ensure spatial–temporal consistency. Experiments on AIS data from Luoyuan Bay indicate that the proposed method outperforms conventional baselines in Hausdorff Distance (HD), Dynamic Time Warping (DTW), and path length deviation (DL), demonstrating improved geometric plausibility and behavioral stability.