Ship Trajectory Prediction based on AIS Data Using a Hybrid Deep Learning Model

Abstract

The emergence of Shipping 4.0 has significantly transformed the maritime industry by advancing data acquisition and processing technologies. One of the most impactful developments is the availability of high-frequency Automatic Identification System (AIS) data, which enables continuous tracking of vessel positions, speeds, and directions in real time. This shift from traditional static route planning to dynamic, data-driven analysis has allowed researchers to extract complex movement patterns from historical AIS records and apply them to predict future ship trajectories more accurately and efficiently. Such innovations are increasingly critical for improving navigational safety and operational performance at sea. To address the challenges of modeling complex sequential dependencies in AIS data, we propose a hybrid deep learning model that combines neural network components to better capture the complex patterns in vessel movement. Each part of the model is designed to handle different aspects of the data, such as local features and long-term features. By integrating these elements, the model can learn both detailed and overall patterns in ship trajectories more effectively. The performance of the proposed hybrid model was assessed using widely adopted trajectory prediction metrics, including mean squared error (MSE) and mean absolute error (MAE). Our results demonstrate that the proposed model consistently outperforms existing models, including standalone LSTM and Transformer models, in terms of prediction accuracy. The proposed approach overcomes the individual weaknesses of LSTM and Transformer architectures. Overall, this study contributes a solution for ship trajectory prediction, with strong potential for application in real-world maritime navigation and planning systems.