Historical AIS Data-Driven Unsupervised Automatic Extraction of Directional Maritime Traffic Networks

Vessel experience route analysis can provide empirical support for maritime traffic management. Recently, the application of the Automatic Identification System (AIS) provides multi-dimensional data about voyages and vessels. However, traditional route extraction methods do not take into account information such as the vessel traffic pattern and the density distribution in the channel. The experience routes obtained is not accurate enough. This paper proposes an unsupervised method for extracting vessel experience routes from historical AIS data. The method consists of three parts: vessel traffic pattern extraction, channel boundary extraction, channel triangle network construction and hottest route extraction. The method comprehensively considers the spatiotemporal information and density distribution of the vessel trajectories, and constructs a directional maritime traffic network which can effectively convert historical data into information supporting decision-making.