A Ship Trajectory Prediction Model Based on ECA-BiGRU

It is essential to forecast the ship's next track with the ship's historical track and real-time motion status information to ensure safe navigation. With the large-scale application of the Chinese satellite navigation system Beidou, the total management of ship trajectory is feasible, and its data sequence characteristics are more robust. This thesis proposes an ECA-BIGRU vessel track forecast model based on Beidou vessel sequence data using the BIGRU model and makes a preliminary track forecast. Then it uses the ECA attention mechanism to optimize BiGRU. The measurement results of 112 ships running continuously for 20 days show that this dissertation proposes higher real-time performance and more accurate ship trajectory prediction in this dissertation than the ordinary BiGRU and LSTM algorithms, and it has significant application value.