A Dimensionality Reduction-Based Multi-Step Clustering Method for Robust Vessel Trajectory Analysis.

Area of Science:

• Maritime technology
• Data science
• Navigation safety

Background:

• Shipboard Automatic Identification System (AIS) data is vital for maritime surveillance and navigation safety.
• Analyzing AIS trajectories aids in identifying abnormal patterns and customary routes.
• Traditional trajectory clustering methods struggle with outliers and complexity.

Purpose of the Study:

• To propose a robust multi-step trajectory clustering method for AIS data.
• To enhance the accuracy and reliability of maritime traffic monitoring and analysis.
• To overcome the limitations of existing clustering techniques in handling trajectory data.

Main Methods:

• Utilized Dynamic Time Warping (DTW) for trajectory similarity measurement.
• Applied Principal Component Analysis (PCA) for dimensionality reduction of the distance matrix.
• Developed an improved center clustering algorithm with automatic selection of the number of clusters (k).

Main Results:

• The proposed multi-step clustering method demonstrated superior performance compared to traditional spectral clustering and fast affinity propagation.
• Experimental results on realistic AIS datasets showed significant improvements in both quantitative and qualitative evaluations.
• The method effectively handles outliers, leading to more accurate AIS trajectory clustering.

Conclusions:

• The developed multi-step clustering approach offers a robust solution for analyzing spatio-temporal AIS trajectories.
• This method enhances maritime surveillance capabilities and contributes to improved navigation safety.
• The findings suggest a promising direction for data mining and pattern analysis in maritime transportation.