Predicting the dynamic information of ship navigation is the basic work of ship abnormal behavior analysis, so improving the performance of prediction models is of great significance for advancing intelligent monitoring at sea.Aiming at the problem of insufficient accuracy of existing ship trajectory prediction models, a long-term and short-term memory network (LSTM) ship track prediction model based on adaptive particle swarm optimization (PSO) optimization is proposed in this paper.The particle swarm algorithm was used to optimize and improve the number of hidden layer nodes, learning rate, maximum number of iterations, and the number of input layer steps in the LSTM network, so as to obtain a better ship track prediction model.A PSO-LSTM model was constructed using AIS data, and experiments were performed using ship AIS data in the VTS waters Wusong Traffic Management Center . Comparing the experimental results with several other models, it was found that the PSO-LSTM prediction model has higher accuracy.