Research On Ship Trajectory Prediction Based On Data Mining

With the rapid development of modern economy,maritime transport has ushered in a huge opportunity for development.How to ensure the safety and economics of maritime transport has become a research hotspot in modern shipping.The huge amount of AIS data on ships contains important ship navigation rules,which can provide necessary information support for ship trajectory prediction.This topic studies the analysis and processing methods of AIS data,digs the inherent laws of ship navigation,and effectively predicts the ship’s track,so as to improve the safety of ship navigation.First of all,researches were made on route coding extraction,data missing repair,and outlier processing in AIS data preprocessing.Aiming at the problem of missing original data,a missing data interpolation method based on bidirectional iterative calculation of interpolation points was designed.Data loss repair test results show that the method can effectively solve the problem of imbalance between the time and accuracy of data repair.Secondly,it focuses on the issue of AIS data compression.Based on the analysis of the trajectory compression evaluation index and the trajectory compression threshold selection principle,an improved ship trajectory data compression method based on the DP algorithm is designed.Compared with the traditional compression method based on DP algorithm,this method can better maintain the main characteristics of the original trajectory while compressing the trajectory data.Thirdly,the problem of ship trajectory clustering is studied.The trajectory similarity measurement method is analyzed,and a ship trajectory clustering method based on adaptive density(DBSCAN)is designed.This method can effectively solve the imbalance problem of the combination of distance threshold ε and density threshold Minpts in the trajectory clustering algorithm.To achieve better clustering of ship trajectories.Finally,a naval Bayesian ship trajectory prediction method is designed,which predicts the ship trajectory based on the clustered ship trajectory data.Based on the AIS data in Zhoushan Archipelago,a verification test for the prediction method was designed.Experimental results show that the method has a higher prediction rate than the trajectory prediction algorithm based on weighted KNN.