Vessel trajectory prediction based on modified LSTM with attention mechanism

Trajectory prediction of moving targets on the sea surface is crucial for nautical safety, maritime traffic management and military applications. In recent years, with the development of deep learning techniques, the use of recurrent neural networks, especially LSTM (Long Short-Term Memory) networks, for analyzing and predicting temporal data has become increasingly popular. In this study, an improved LSTM model based on the attention mechanism is proposed for a more accurate prediction of the trajectory of a moving target on the sea surface. In traditional LSTM models, old information may be forgotten as time advances, leading to a decrease in prediction accuracy. To overcome this limitation, this paper integrates an attention mechanism that aims to strengthen the model's ability to capture important information in time series data. The attention mechanism enables the model to focus more on the part of the sequence that is most relevant to predicting the target trajectory by assigning different weights to different time steps. In this paper, the model is trained and tested using historical trajectory data of real sea surface moving targets. The experimental results show that compared to the standard LSTM model, our proposed improved LSTM model based on the attention mechanism significantly improves the prediction accuracy, showing better prediction accuracy and robustness.