基于DBSCAN算法的营运车辆超速点聚类分析  

Clustering Analysis of Overspeed Spots for Commercial Vehicles Based on DBSCAN

作　　者：刘卫宁[1] 曾婵娟[1] 孙棣华[2] 

机构地区：[1]重庆大学计算机学院,重庆400044 [2]重庆大学自动化学院,重庆400044

出　　处：《计算机工程》2009年第5期268-270,共3页Computer Engineering

基　　金：重庆市智能交通系统示范工程及其关键技术基金资助项目(7742)

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。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.

关 键 词：数据挖掘 聚类 营运车辆 安全管理 超速 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]