Trajectory Prediction of Vessels based on Data Mining and Machine Learning

With the development of marine traffic information digitalization, proactive information service has become increasingly important in maritime intelligent traffic systems (ITS). Trajectory prediction is one of the kernel problems that must be addressed to realize proactive information service. This study proposes an intelligent model to solve the issue of the trajectory prediction of vessels based on data mining and machine learning methods. The spatial clustering algorithm of data mining is used to cluster the historical trajectories of vessels, and the cluster results represent the distribution patterns of these historical trajectories. The support vector machine algorithm of machine learning is used to train the classifiers. The classifiers define the pattern of the new trajectory of the vessel, which must be predicted. In the experiment, the information on 2862 trajectories is used as input to the model in chronological order to simulate the data flow in realtime situation. The predicted trajectories are compared with the actual trajectories of vessels. Experimental results show that future trajectories can be predicted efficiently and accurately. The intelligent model can also solve prediction problems with little human intervention and can automatically adapt to dynamic applications. The prediction results can provide accurate and reliable data for proactive information service. The model promotes the development of maritime ITS. Subject Categories and Descriptors K.2.8 [Database Applications]: Data mining; B.2.4 [HighSpeed Arithmetic]: Algorithms General Terms: Data Mining, Machine Learning, Intelligent Traffic Systems