基于位置可达域的船舶轨迹噪声去除方法

A Noise Reduction Method for Ship Trajectory Based on Position Reachable Domain

船舶的AIS轨迹数据对研究船舶交通流和交通行为具有重要意义。然而通信链路及卫星定位信号等方面的干扰,导致AIS轨迹含有较多的噪声数据,严重影响了数据分析和数据挖掘的质量。根据船舶运动规律,研究得出一种基于位置可达域的船舶轨迹噪声数据去除方法。通过计算某时刻船舶运动的所有可能状态,估计船舶下一时刻位置可达的范围作为可达域判别条件,并利用滑动窗口对船舶AIS轨迹点进行滚动处理,将不满足可达域判别条件的点作为噪声点进行滤除,形成轨迹去噪算法流程。选取含有噪声的真实AIS轨迹数据进行方法验证,并与当前常用的轨迹去噪方法进行对比分析。相比于速度阈值法和基于密度的聚类算法,所提出的方法在召回率上分别提升了27.22%和23.14%,在F1指标上分别提升了14.31%和24.03%,说明算法对噪声点的识别较为准确,并能有效针对不同类型噪声进行处理。同时,通过多次重复试验和时间复杂度分析,验证该算法不仅满足海量AIS数据的离线去噪需求,还可用于轨迹在线去噪,满足实时应用的需要。

The trajectory data of AIS is of great significance for studies of ship traffic flow and traffic behavior.However,interference from communication links and satellite positioning signals causes the AIS trajectory to contain a high level of noise data,which seriously affects the quality of data analysis and data mining.According to the ship motion pattern,a method for removing noise from ship trajectory data based on position reachable domain is developed.By calculating all possible states of ship motion at one certain moment,and estimating the reachable range of the ship's position at the next moment as the reachable domain discriminating condition,the sliding window method is used to process the points of the ship's AIS trajectory,and the points that do not meet the reachable domain discriminating condition are filtered out as noise points to form the trajectory denoising algorithm flow.The actual AIS trajectory data containing noise data is selected to verify the method,and compared with the currently used noise reduction methods.Compared with the velocity threshold method and DBSCAN,the proposed method improves the recall rate by 27.22%and 23.14%,and improves the F1 by 14.31%and 24.03%,respectively,indicating that the algorithm is more accurate in identifying the noise points and can effectively handle different types of noise.Meanwhile,through repeated tests and time complexity analysis,it is verified that the algorithm not only meets the offline denoising requirements of massive AIS data,but also can be used for online noise reduction of trajectories to meet the needs of real-time applications.