Research On Arrival Time Prediction And Trajectory Prediction Of Oil Tankers Based On AIS Data

Over 80% of the world trade is carried by seas and oceans,making maritime transportation one of the most important means of transportation.In maritime transportation,port operation plays a vital role as it bridges maritime transportation with terrestrial transportation,which ultimately brings merchandise and services to our daily life.Compared to land transport,maritime transport has complex sea conditions and more difficult to perceive the sea situation,which leads to various problems in maritime transport.Two of the major themes are the timely arrival of ships and the knowledge of their future trajectory.On the one hand,for port,effective ship arrival time prediction can be used to assist the decision making of berth allocation,quay crane management,storage and terrestrial transportation arrangement so as to gain high efficiency.Especially in the past few years,the whole world has been hit by the COVID-19 epidemic,and the operation of ports and terminals has also suffered a great test;On the other hand,both collision avoidance and abnormal behavior detection of ships are inseparable from the prediction of ship trajectory.Accurate trajectory prediction can greatly improve the safety and efficiency of sea transportation.This paper explores the development of solutions to improve the efficiency and safety of maritime transport operations by using data-driven research methods.Prediction of arrival time of oil tanker and tanker trajectory prediction are the two themes of this paper.The main work and contributions are as follows:For the prediction of arrival time of oil tanker: we propose a data-driven approach for estimating the time of arrival(To A)of ships based on Automatic Identification System(AIS)data.The proposed approach exploits a novel trajectory clustering approach to extract representative routes,from which training trajectories are grouped into clusters and To A estimation models based on Support Vector Regressor(SVR)are trained for each cluster.The proposed To A estimation approach adopts a short period of the latest AIS data as input and performs SVR model selection and To A estimation in realtime.A historical AIS dataset provided by the Danish Maritime Authority is used to evaluate the proposed approach.Numerical results for tanker ships travelling to the port of Skagen demonstrate that the proposed approach can achieve To A estimations that are 40% to 70% more accurate than state-of-the-art To A estimation methods.For trajectory prediction: we propose a long-term trajectory prediction algorithm.The algorithm is still based on AIS data.Different distance and angle parameters are calculated according to the current position and destination of the target tanker,and then the navigation mode that may exist at the current position in the historical data is fused to predict the remaining route to the destination.Compared with the prediction method implemented by combining graph theory and pathfinding algorithms,it achieves more than 25% improvement in the similarity with real trajectories.Finally,as a supplement to the work,we develop a HTML-based user interface to display the research results.The interface shows the results of arrival time prediction and trajectory prediction produced by the algorithms developed.