Research of Ship Trajectory and Navigation Behavior Characteristics Based on CALS Algorithm

This study examines the navigation trajectory data of over AIS ships in the Chinese sea area during 2021, proposes a CNN - Attention - LSTM neural network deep learning algorithm for constructing a predictive and early warning model for navigation trajectories based on ship navigation trajectory training, and introduces the sliding window algorithm to effectively address concept drift in streaming data, wherein the underlying patterns of data change with time. Furthermore, real - time detection of abnormal navigation states is achieved. Additionally, ship behavior abnormalities are categorized into three groups and ten types based on AIS information parameters related to navigation trajectories. This approach successfully addresses the challenge of detecting and providing early warnings for abnormal navigation trajectories while offering valuable support for safe navigation and military decision - making.