基于改进谱聚类算法的航路辨识

为解决船舶自动识别系统(automatic identification system, AIS)数据挖掘不够充分,对航路辨识分析不够全面等问题,提出一种基于改进谱聚类算法的数据挖掘方式。利用Sliding Window算法对船舶轨迹AIS数据进行压缩,减少数据冗余提高聚类效率。改进亲和距离函数,提出新的亲和矩阵的标准,提高聚类的稳定性,进一步对数据去噪,减少噪声敏感。通过优化初始中心对k均值算法进行改进,优化全局搜索能力,缓解初始值的选取对聚类效果的影响。以天津港AIS数据为样本进行算法验证。结果表明,该聚类算法能准确提取和划分某水域船舶主要航迹段,算法消耗系统资源少,计算速度快。改进后的算法可为航路辨识、分道通航制定等提供理论支持。

Route identification based on improved spectral clustering algorithm

In order to solve the problem that automatic identification system (AIS) data mining is not enough and the analysis of route identification is not comprehensive enough, a data mining method based on the improved spectral clustering algorithm is proposed. The Sliding Window algorithm is used to compress AIS data of ship trajectory so as to reduce data redundancy and improve the clustering efficiency. The affinity distance function is improved, a new affinity matrix standard is proposed to improve the stability of clustering, and the data denoising is carried out to reduce the noise sensitivity. The k means algorithm is improved by optimizing the initial center so as to optimize the global search ability and alleviate the impact of the initial value selection on the clustering effect. The AIS data of Tianjin Port is taken as a sample to verify the algorithm. The results show that, the clustering algorithm can accurately extract and divide the main track segments of ships in a certain water area, and the algorithm consumes less system resources and has faster calculation speed. The improved algorithm can provide theoretical support for route identification, traffic separation, and so on.