Analysis of Vessel Anomalous Behavior Based on Bayesian Recurrent Neural Network

Nowadays, with the development of the maritime transportation industry, the safety of vessel navigation has become an important issue. Vessel anomalous behavior analysis can detect the problematic vessels in time, so that supervisors can take timely measures to avoid accidents. How to detect the anomalous vessel behavior based on Automatic Identification System (AIS) data has become an important research issue. In this paper, an architecture (Bayesian Recurrent Neural Network for Vessel Behavior) is proposed. In the architecture, we propose an encoding method to represent AIS trajectory data and a neural network structure combined with VAE to learn the vessel normal behavior. Finally, we perform experiments on the Nanjing dataset to verify how the model outperforms other anomalous behavior analysis methods.