动态聚类法在车辆实际行驶工况开发中的应用  

Application of Dynamic Cluster to the Development of Vehicle Real World Driving Cycle

作　　者：马志雄[1] 朱西产[2] 李孟良[2] 乔维高[1] 张富兴[2] 

机构地区：[1]武汉理工大学汽车工程学院,武汉430070 [2]中国汽车技术研究中心,天津300162

出　　处：《武汉理工大学学报》2005年第11期69-71,81,共4页Journal of Wuhan University of Technology

基　　金：国家"863"计划(2003AA501993)

摘　　要：在车辆实际行驶工况的开发过程中,为了得到相应交通特征道路上的实际行驶工况,需要将运动学片断按照道路交通特征分类。针对人工分类的不足,研究了动态聚类法在车辆实际行驶工况开发中的应用。以主成分分析得到的前3个主成分为聚类因子,并对运动学片段总体样本采用动态聚类法分类。研究表明:动态聚类法分类获得的2类运动学片段分类清晰,分别代表了车辆在交通拥挤和交通畅通的道路上的行驶特征。In the course of the development of vehicle real world driving cycle, in order to develop vehicle real world driving cycle on the relevant traffic characteristic road, the kinematic sequences must be classified according to road traffic characteristic. Taking into account the deficiency of artificial classification, the purpose of this paper was to report the application of dynamic cluster to the development of vehicle real world driving cycle. The first three principal components, which came from principle component analysis （PCA）, were regarded as the cluster factors, and using the method of the dynamic cluster, the kinematic sequence＇s overall sample was classified. The study showed two groups of kinematic sequences, which was obtained by classification, were very distinct and respectively represent travel characteristic that vehicle was on the crowded road and on the unblocked road.

关 键 词：动态聚类法 运动学片段 行驶工况 

分 类 号：U491.2[交通运输工程—交通运输规划与管理]