基于轨迹的内河船舶行为模式挖掘

从内河海量的船舶AIS数据中提取出有用的交通知识,辅助水上安全监管,对于研究日益复杂的水上交通安全形势具有重要意义.基于内河船舶行为特征,构造由船舶位置、航速和航向4个维度组成的船舶航行状态空间来描述船舶行为.针对传统DBSCAN聚类算法提取状态空间中相似船舶轨迹存在计算复杂高的问题,提出增量式算法改进DBSCAN算法用以高效地计算不同船舶的行为模式;然后利用核密度估计等统计方法对不同模式的船舶行为特征进行数据挖掘,得到船舶航速、航向和位置的时空分布特征规律,进一步挖掘不同行为模式下的船舶微观特征.以武汉航段的汉江分叉航道水域作为研究案例,利用所提的方法对该水域分析研究,得到了6类不同行为模式,挖掘出不同模式下分叉航道内船舶静态属性信息(船舶类型、船舶尺寸)、空间分布特征(轨迹点分布、航速分布、航向分布)、船舶到达规律等信息.利用该模型所提取的知识有助于水上监管人员迅速获取水域交通态势,从而提高水上交通安全监管的水平和效率. 

Behavior Pattern Mining of Inland Vessels Based on Trajectories

In view of complex situations of marine traffic safety, it is of great significance to investigate AIS data mining methods for useful traffic information.On basis of behavior patterns of inland vessels, a four-dimensional state-space model including temporal and spatial locations, speed, and course is proposed to describe behavior patterns of vessels.Considering high time complexity of extracting similar ship trajectories in the state space model, an incremental DBSCAN algorithm is thus introduced for effective calculations of different behavior patterns of vessels.Statistical methods such as kernel density estimation are further applied to derive vessel behavior characteristics under different modes, and spatial-temporal distributions of microscopic characteristics (i.e.vessel speed, heading angle, and position).Six different kinds of behavior patterns are analyzed through a case study in bifurcation waterways of Hanjiang River in Wuhan, China.Static information (types and sizes of ships), spatial distribution characteristics (trajectories, speeds, and heading angles), and arrival patterns of vessels are successfully extracted.The model can be helpful to improve supervision efficiency of maritime traffic safety.