Research On Visual Analysis Of Ship Behavior Based On AIS Data

With the development of waterway transportation,the issue of water security has become more and more important.It has much theoretical value and application value to analyze and study the ship behavior characteristics targeting on the safety of ship navigation.As a common ship identification and tracking facility,ship automatic identification system(AIS)are widely equipped on ships.The AIS data generated by shipsactan important role in ship behavior study.Traditional research of AIS data and ship behavior only do data mining for specific aspects,resulting in rigid,complex and less exploratory outcome.Whereas visual analysis technology combined with visualization and human-computer interactioncan provide users with intuitive information view and interactive analysis.Therefore,based on AIS data,using visual analysis technology of ship behavior research will provide a new solution to study ships sailing Security.This thesis do research on ship behavior characteristic modeling method and interaction model of the ship behavior visual analysis from three aspects namedship trajectory distribution,ship density distribution and ship speed distribution,combined with a variety of visualization methods.As a result,a ship behavior visual analysis prototype system was designed.The main work of this thesis is as follows:(1)Modeling method of ship’s behavior characteristicsTo do visual analysis on the three aspects of the ship behavior attributes,feature modeling must be done first.The trajectory clustering algorithm based on tree clustering was improved combined with the characteristics of inland river ship trajectory,and thus the trajectory distribution modeling method based on tree cluster aggregation was proposed.Oriented to heatmap visualization method,vessel density distribution feature modeling method wasstudied based on kernel density estimation(KDE).Oriented to grid color coded visualization method to study the feature modeling method of ship speed spatial distribution and probability distribution.(2)Research on visualization method of ship behavior characteristicsBased on the modeling methods of ship behavior,the visualize method of the ship behavior wasstudied.On the ship trajectory distribution,use the aggregated visualization method to draw feature modeling visualization view.On vessel density distribution,heat map of vessel density distribution was visualized based on kernel density estimation matrix.On ship velocity distribution,at first waterway were divided to meshes,then visualize ship speed distribution by color coding method of grid.(3)Research on interaction model of ship behavior visual analysisBased on feature modeling outcome and visualization methodof ship behavior features,firstly according to the user analysis demand,do research on visibility analysis interactive task modeling.Interactive tasks include underlying interaction tasks such as changingthreshold in algorithms of the trajectory and density distribution feature modeling and top-level interaction tasks targeted to the data set and view surface.All human-computer interaction models were designed based on interaction tasks proposed previously.(4)Prototype system designing and experimental analysisOn the basis of the ship behavior feature modeling methods,visualization methods and interaction models,a prototype system of ship behavior visual analysis was designed.Then worked on experiment and analysis of the proposed visualization method and interaction model in the visual analysis,and do analysis to evaluate the performance of the improvement on trajectory clustering algorithm.As a result experimentsshow that visualization method and interaction model of ship behavior visual analysis proposed in this thesis showed good effect,is effective.And the improved trajectory clustering algorithm is feasible.