Research on autonomous route generation method based on AIS ship trajectory big data and improved LSTM algorithm

Abstract

The autonomous generation of routes is an important part of ship intelligence and it can be realized by deep learning of the big data of automatic identification system (AIS) ship trajectories. In this study, to make the routes generated by long short-term memory (LSTM) artificial neural network more accurate and efficient, a ship route autonomous generation scheme is proposed based on AIS ship trajectory big data and improved multi-task LSTM artificial neural network. By introducing an unsupervised trajectory separation mechanism into LSTM, a fast and accurate separation of trajectories with similar paths is realized. In the process of route generation, first of all, a clustering algorithm is used to cluster the trajectories in massive AIS data according to the density of trajectory points, so as to eliminate the trajectories in the routes that do not belong to the target area. Furthermore, the routes are classified according to the type of ships, and then the classified trajectories are processed and used as datasets. Based on these datasets, an improved LSTM algorithm is used to generate ship routes autonomously. The results show the improved LSTM works better than LSTM when the generated route trajectories are short.

Large-Scale Long-Term Prediction of Ship AIS Tracks via Linear Networks with a Look-Back Window Decomposition Scheme of Time Features

“…To identify the areas with the densest ship trajectories, we used an algorithm called clustering analysis to process all ship trajectory data [20]. Cluster analysis is an important algorithm for analyzing vessel trajectories.…”

“…All models utilized in this experiment were specifically designed for time series forecasting of ship trajectories. Seven different models were used: Autoformer [23], Informer [24], Transformer [25], Linear, NLinear, DLinear, and PatchTST [20]. The input consisted of a time series of length seq_len, which contained both multivariate and univariate feature data.…”