The Research Of Ship Abnormal Behavior Based On AIS Data

With the rapid development of China’s shipping industry,the huge vessel traffic volume makes the maritime traffic environment become increasingly sophisticated,and the security problem is becoming more prominent.The wide application of the AIS has created growing ship’s space-time data.Therefore,it is a research hotspot to improve the utilization efficiency of ship’s space-time data,obtain the characteristic of maritime traffic and realize the intellectualization of the marine traffic management system.As an important part of maritime traffic supervision,abnormal behavior detection is significant for the navigation safety,marine protection and illegal activity.In view of this,based on the historical information of AIS,the paper uses Hive,SparkR and data mining technology to research the motion feature and law from OD trajectory segmentation,port mooring mining and ship type identification.And combining the model establishment and case study,ship abnormal behavior detection is realized.By analyzing the related research of anomaly behavior detection in various fields,the paper summarizes the research methods of abnormal behavior and puts forward the following three approaches for detecting ship abnormal behavior.First,based on OD segmentation model,a track anomaly detection method is established.Taking the “COSCOSTAR” as an example,the actual sailing date and route of the ship are obtained.By comparing actual and scheduled sailing date and route,the correctness of the model is verified.Secondly,this paper establishes the port mooring mining method based on grid density clustering,selects six ports at home and abroad to obtain the mooring location and area,and validate the correctness of model by analyzing the result and real-time mooring behavior in dynamic ship map.Finally,this paper establishes a ship type recognition model based on random forest.This paper takes Zhoushan fishing ground as an example to obtain the fishing boat recognition model,and validate the correctness of model by comparing the type of model and the type of ship reported.The results show that three ship behavior models can reflect the characteristic and law of marine motion.Based on the analysis of the characteristic,it can be used to detect the abnormal behavior of the ship.Among them,the OD segmentation model can be used to detect ship route anomaly,mooring mining model can be used for real-time detection of mooring behavior,and ship type identification model can be used to detect abnormal behavior of non fishing boat in fishing ground.Finding the ship abnormal behavior by mining the characteristic of ship’s motion can effectively regulate the traffic safety,lay the foundation for the intellectualization of ship traffic management system,and also promote the healthy development of navigation.