考虑对地航速和航向的船舶典型轨迹提取方法

基于船舶自动识别系统(Automatic Identification System, AIS)数据的船舶典型轨迹挖掘需要经过两个重要步骤，一是压缩 AIS 数据，二是聚类压缩后的 AIS 数据。传统的DP(DouglasPeucker)压缩算法，只考虑船舶轨迹的压缩形状，忽视了船舶航行中其他重要信息。为解决此问题，把对地航速和航向加入到DP算法的压缩过程中。在AIS轨迹聚类方面，传统谱聚类方法只对船舶轨迹的位置进行相似性度量，没有考虑船舶轨迹的其他维度，针对此问题，提出多属性轨迹相似性度量方法。由于不同的输入参数影响着最终的聚类质量，引入Calinski-Harabasz指标评价谱聚类算法，实现聚类参数的自适应选择。利用山东威海水域的实际AIS数据进行实例研究，并与传统谱聚类算法做比较实验。实验结果表明，利用该方法提取到的典型轨迹符合真实水域的交通情况，相较于传统谱聚类方法具有更高的聚类质量。

Typical Ship Trajectory Extraction Method ConsideringGround Speed and Heading

Typical trajectory mining of ships based on Automatic Identification System (AIS) data needs to go throughtwo important steps, which include compressing AIS data and then clustering the compressed AIS data. The traditionalDouglas-Peucke (DP) compression algorithm only considers the compressed shape of ship trajectory, but ignores otherimportant information in the ship navigation. To solve this problem, the ground speed and heading are added to thecompression process of the DP algorithm. In the AIS trajectory clustering, the traditional spectral clustering methodonly measures the similarity of ship trajectory position, without considering other dimensions of ship trajectory. Tosolve this problem, a multi-attribute trajectory similarity measurement method is proposed. Since different inputparameters affect the final clustering quality, the Calinski- Harabasz index is introduced to evaluate the spectralclustering algorithm, and then the adaptive selection of clustering parameters is realized. The actual AIS data of theWeihai water area in Shandong Province are used for a case study to compare the proposed algorithm with thetraditional spectral clustering algorithm. The experimental results show that the typical tracks extracted by this methodare consistent with the traffic conditions of real water areas, and the clustering quality is higher than that of traditionalspectral clustering methods.