基于Transformer的海上船舶轨迹预测方法OACSTPCD

The trajectory prediction of marine ship has an important application prospect in the safe operation and marine su-pervision.Aiming at the problem of ship trajectory prediction,this paper proposes Transformer based marine ship trajectory predic-tion method.First,the trajectory data is preprocessed to eliminate the mooring points and abnormal points,and then the trajectory resampling technology is used to achieve the equal time interval sampling of the trajectory,so as to reduce the impact of time series characteristics on trajectory prediction.Then,drawing on the concept space subdivision of GeoHash technology,this paper divides longitude and latitude equally,and then converts it into high-dimensional embedded space by equidistant segmentation of longitude and latitude and one pot representation.Finally,the Transformer technology is used to predict the trajectory.The results show that compared with RNN,LSTM and Kalman filtering methods,the method in this paper has a greater performance improvement in 30 min and 60 min trajectory prediction accuracy.