Research And Application Of Liner Track Prediction Based On AIS Data And Bi-LSTM Algorithm

Today,with the mutual promotion and symbiosis of the global economy,the economic exchanges between countries are becoming more and more frequent.As an important link connecting the global economy and trade,shipping is the lifeline of the global economy.With the growth of global trade,the number of ships and routes in various sea areas has also increased,and the risk of ship navigation has also increased,and marine accidents have occurred from time to time.To this end,the International Maritime Organization requires that sailing ships must be equipped with AIS systems.By studying a large number of ship historical track data to predict the ship’s track,we can increase our monitoring of the ship’s navigation,thereby reducing the risk of ship navigation and ensuring the safety of navigation.Ship navigation is affected by various uncontrollable factors at sea,and there are generally large errors in track prediction.While the liner route is fixed,the ship is affected by a large number of the same factors during navigation,and the movement characteristics of the ship have more the same laws.For this reason,this paper selects the liner track with a fixed navigation route for research,and implements the liner track prediction model based on the historical track design of the liner route to improve the accuracy of the liner track prediction.The main work of this paper is as follows:(1)According to the longitude and latitude of arrival and departure of the liner route,combined with the longitude and latitude,speed and navigation status of the AIS trajectory point,the liner navigation trajectory data set of the liner route is extracted,and the extracted navigation trajectory data is preprocessed.First,the abnormal data of the track is cleaned;then,an improved hybrid interpolation method is used to interpolate and repair the missing data of the track,so as to obtain the complete liner navigation track data set.(2)Combined with the characteristics of the liner track,an improved DTW similarity measurement method is studied,and the cluster analysis of the liner track data set is carried out to extract the typical route track clusters of the liner and form its typical navigation route.(3)Three track prediction models(Stacked-LSTM prediction model,CNNLSTM prediction model,Bi-LSTM prediction model)are constructed based on LSTM,and the distances from the voyaged track segment to each cluster center of the typical track cluster are calculated for the predicted ship.The calculation of,takes the closest track cluster as the prior data of the predicted ship.This paper compares the prediction accuracy of these three track prediction models,and finally confirms that the model trained by Bi-LSTM has a relatively good prediction effect.(4)In order to display the historical track and predicted track more intuitively,this paper develops a visualization system based on the Bi-LSTM track prediction model,which maps the geographic location of the ship to the global chart,not only can intuitively see the ship at sea The historical navigation trajectories and predicted trajectories are also added to the relevant data analysis to improve the use value of the trajectory data.