内河船舶缺失轨迹修复方法

An approach for restoring the lost trajectories of vessels in inland waterways

为解决船舶自动识别系统常出现数据丢失妨害海事监管和交通流采集的问题,本文利用内河中船舶时空轨迹具有相似性的特点,提出基于相似轨迹回归的残缺轨迹修复方法。通过插值法对时间周期不规律的AIS数据进行同步,提出相似轨迹快速搜寻方法从数据库中搜寻出相似轨迹。采用改进Hausdorff距离评估和挑选丢失轨迹的最相似轨迹,并利用粒子群优化算法和相似轨迹优化最小二乘支持向量机修复模型,进而修复残缺轨迹数据。实验结果表明:对于长距离残缺轨迹,本文提出的轨迹修复方法比BP神经网络和样条插值法精度更高,平均误差小于30 m;在时效性方面,本方法次于插值法,但优于BP神经网络。

Interference from fragmentary Automatic Identification System(AIS)data has always been problematic in maritime supervision and traffic flow data collection.We therefore propose a restorative method for fragmentary trajectories based on similar trajectory regression.This method utilizes the similar characteristics of ship spatial temporal trajectories in inland rivers.Firstly,the AIS data with irregular period was synchronized by interpolation.A parallel trajectory,fast-searching method was proposed to investigate the similar trajectories from the trajectory database.The improved Hausdorff distance was introduced to evaluate and select the track most similar to the fragmentary trajectory.Finally,a least-squares post-vector mechanisms(LS-SVM)repairing model optimized by a particle swarm optimization(PSO)algorithm and similar trajectory were used to restore the fragmentary trajectory.The experimental result showed that,for the long distance miss track points,the method proposed in this paper was more accurate than a back propagation(BP)neural network and spline interpolation,with an average error of<30 meters.In terms of timeliness,the elapsed time of this method was longer than that of simple interpolation,but superior to that of the BP neural network.