A TDV attention-based BiGRU network for AIS-based vessel trajectory prediction

Area of Science:

• Maritime technology
• Artificial intelligence
• Data science

Background:

• Automatic Identification System (AIS) provides vessel data but suffers from noisy trajectories and prediction inaccuracies due to flexible routes and unconfirmed broadcasts.
• Existing methods struggle with the inherent variability and unreliability of raw AIS data, impacting trajectory prediction performance.

Purpose of the Study:

• To develop an advanced vessel trajectory prediction model that enhances accuracy and stability.
• To address the limitations of raw AIS data, including noise and route variability, for more reliable predictions.

Main Methods:

• Proposed a novel Trajectory Direction Vector (TDV) to integrate positional data (latitude, longitude) with navigational parameters (course, speed).
• Introduced an attention mechanism to dynamically adjust the influence of TDV at different stages, filtering out erroneous trajectory points.
• Developed a hybrid model combining the TDV attention mechanism with a bidirectional gate recurrent unit (BiGRU) network for trajectory prediction.

Main Results:

• The integrated TDV and attention mechanism effectively filters noisy data and improves the relevance of trajectory information.
• The BiGRU network, enhanced by the TDV attention mechanism, demonstrates superior performance in predicting vessel trajectories compared to traditional methods.
• The proposed model achieves higher accuracy and stability in vessel trajectory prediction using AIS data.

Conclusions:

• The combination of TDV, attention mechanism, and BiGRU offers a robust solution for accurate and stable vessel trajectory prediction.
• This method effectively mitigates issues associated with noisy AIS data and dynamic vessel movements.
• The findings contribute to improved maritime safety and operational efficiency through enhanced predictive capabilities.