| Along with the booming development of navigation industry,the contradiction between the demand of ship behavior supervision and traffic service resources is increasingly prominent.How to dynamically monitor the maritime behavior of ships has become a difficult problem in maritime transportation supervision.With the continuous popularization of the Automatic Identification System,the static and dynamic information of ships contained in it provides new opportunities for the research and monitoring of ship anomalies.Among them,abnormal ship behavior is the focus of ship behavior detection,which is helpful to assist maritime management personnel in ship monitoring and relevant departments in maritime traffic planning,and plays an important role in supporting national economic development,promoting foreign trade and maintaining soundness.With the increasingly extensive and in-depth application requirements of ship abnormal behavior detection,the requirements for the characterization and detection accuracy of abnormal behavior are getting higher and higher.The combination of geographic information science and traffic engineering knowledge provides a new idea and method for detecting abnormal ship behavior.In view of this,based on the basic knowledge of ships in the field of maritime traffic,combining the theoretical methods of spatial behavior modeling,spatiotemporal analysis and spatiotemporal anomaly detection,the abnormal behaviors of ships are analyzed and detected from the group and individual levels respectively,so as to provide scientific law knowledge and theoretical model for real-time monitoring,timely prevention and control and optimization suggestions of abnormal behaviors of ships.The main contents of this paper include:(1)Recognition of abnormal ship behavior modeling.First of all,based on the research results of ship abnormal behavior,we expound the relevant theories and representative methods of ship abnormal behavior research at home and abroad;Furthermore,we introduce and summarize the classic and cutting-edge ship abnormal behavior framework in chronological order;Then,based on the abstract and induction of ship behavior characteristics,we propose a ship abnormal behavior detection framework based on group-individual behavior law cognition.(2)Anomaly detection of ship trajectory based on cognition of group behavior law.Based on the ship’s trajectories,the network topology structure of the ship’s routes is extracted by using the strategy of main graph structure learning.On this basis,the Rayda’s criterion is used to detect and describe the abnormal behavior in the ship’s space;in terms of dynamic information,we model the multi-dimensional characteristics of ships based on the Isolation Forest Algorithm,and detect and describe abnormal behaviors in the dynamic information of ships.(3)Anomaly detection of ship trajectory based on cognition of individual behavior law.On the basis of mining the ship behavior pattern with the cognition of the group behavior law,we integrate the nonspatiotemporal information of the ship,and detect the difference of the behavior law on the space information and the dynamic information of different types of ships with fine granularity.In space,we use ST-Apriori algorithm to detect the ship’s spatial channel pattern and find the irregular space movement pattern;On the dynamic information,we combine KPCA algorithm and SVM algorithm to mine the multi-dimensional characteristics of ships,and find the differences and anomalies of ship behavior patterns under different categories.(4)Design of ship trajectory anomaly detection system based on cognition of group-individual behavior law.Based on the anomaly detection algorithm proposed in this paper,we use the real ship trajectory data to realize the visual analysis of maritime situation awareness analysis,target behavior analysis and meteorological visualization functions.Finally,the main research results and innovation points are summarized,and the problems to be further studied are prospected. |
