摘要： 针对挖掘营运车辆超速点过程中存在的问题，提出一种基于密度的聚类方法。该方法依据车载GPS实时监控数据，挖掘超速多发点段，通过区域查询搜索超速点邻域内所有超速事件，寻求超速密度大于阈值的点或地段，并创建密度可达最大值的超速点聚类。同时利用简单直观的邻接表替代R*-树，简化了数据结构的建立过程，减少内存占用。实验结果表明，该方法有效。

关键词: 数据挖掘, 聚类, 营运车辆, 安全管理, 超速

Abstract: Aiming at the problems existed in the process of mining overspeed spots for commercial vehicles, a density-based clustering method is proposed, which searches for the spots where the overspeed usually happens in high frequency according to the GPS real-time data. The overspeed spots with high density can be found by searching in all overspeed in the neighborhood of each spot. The maximum overspeed spots can be identified by clustering. To simplify the data structure building process and reduce the memory space it occupied, the method is improved with the adjacency list replaced R*-Tree. Experimental results show this method is effective.

Key words: data mining, clustering, commercial vehicles, safety management, overspeed

中图分类号: 

• TP301.6