Classification of Ship Trajectories by Using Naive Bayesian algorithm

Abstract

In order to automatically classify the ship’s historical trajectory and predict the class of a ship’s trajectory, a ship’s track classification algorithm based on naive Bayesian method is proposed. Using the Automatic Identification System (AIS) data of the Yangtze River in Wuhan section, the AIS data is first preprocessed to extract valid trajectory data. Then the trajectory data is analyzed and the characteristics of average speed, average heading, maximum heading, minimum heading, heading variance and maximum turning rate are extracted. The Naive Bayes classifier is trained and verified. The results show that the accuracy of classification is as high as about 98.59%. It takes only 0.165s to extract features from 709 ship trajectories. The Naive Bayesian classification method can effectively classify inland ship trajectories.