Research On Ship Trajectory Prediction Based On PSO-BiLSTM

With the acceleration of the global economic integration process,maritime transport has become the main channel for foreign trade of various countries.However,the maritime traffic environment is complex and unstable,and accidents such as ship collisions and groundings occur frequently.Therefore,the importance of accurate,efficient and real-time prediction of ship trajectories is self-evident in order to alleviate maritime traffic congestion,reduce accidents and promote intelligent and efficient operation of water traffic management.In this paper,based on the BP(Back Propagation)neural network,Long Short-term Memory(LSTM)and Bidirectional Long Short-term Memory(Bi LSTM)networks,we combine the particle swarm algorithm(Particle Swarm Optimization(PSO),a hybrid PSO-Bi LSTM prediction model is developed,and the following conclusions are drawn:(1)The prediction model is validated by the historical trajectory data of Automatic Identification Systems(AIS),and it is found that the improved BP,LSTM and Bi LSTM models based on PSO algorithm have better prediction accuracy,and their average absolute errors in longitude are reduced by 79.5%,33.3% and The average absolute errors in latitude were reduced by 8.7%,64.9% and 53.8%,which verified the effectiveness of the improved particle swarm algorithm and significantly improved the prediction accuracy of the models.(2)The PSO-Bi LSTM model is compared with PSO-BP and PSO-LSTM models for ship trajectory prediction experiments,and the experimental results show that the PSO-Bi LSTM model has the best prediction accuracy,and its average absolute error in latitude and longitude is(0.0070,0.0091),which is reduced(18.6%,2.1%)compared with the PSO-LSTM network model;compared with the PSO-BP network model(78.2%,39.3%),thus obtaining that the PSO-Bi LSTM model has higher accuracy in predicting ship trajectories compared with other models,which verifies the effectiveness of the PSO-Bi LSTM model proposed in this paper in ship trajectory prediction.(3)The experimental results show that the PSO-Bi LSTM model still has higher prediction accuracy compared with the PSO-LSTM model through the trajectory prediction comparison experiments for ships in different scenarios,which verifies the general applicability of the PSO-Bi LSTM model in the prediction of different ship trajectories.