基于改进K-均值聚类算法的合肥市电动客车行驶工况构建  

Construction of Hefei Electric Buses Driving Cycle Based on Improved K-Means Clustering

作　　者：孙骏[1] 方涛[1] 张炳力[1] 李傲伽 朱鹤[2] Sun Jun;Fang Tao;Zhang Bingli;Li Aojia;Zhu He(Hefei University of Technology,Hefei 230009;Anhui Ankai Automobile Co.,Ltd.,Hefei 230051)

机构地区：[1]合肥工业大学,合肥230009 [2]安徽安凯汽车股份有限公司,合肥230051

出　　处：《汽车技术》2020年第8期56-62,共7页Automobile Technology

基　　金：安徽省科技重大专项项目(JZ2018AKKZ0321);安徽省智能汽车工程实验室项目(PA2018AFGS0026)。

摘　　要：为构建合肥市电动客车行驶工况,选取合肥市4条典型线路的12辆电动公交车进行连续一周的数据采集,基于短行程分析法将经过处理的有效数据分割成24595个片段并计算特征值,采用主成分分析法和改进K-均值(K-Means)聚类算法对短行程特征值进行降维与聚类,根据类中心距离从聚类结果中选出类代表短行程,从而构建出合肥市电动客车的行驶工况。将所构建的工况与实车采集数据及国内外典型行驶工况进行对比,结果表明,构建的行驶工况能更准确地反映合肥市电动客车的行驶特征。In order to construct the driving cycle of electric buses in Hefei,12 electric buses on 4 routes in Hefei are selected to collect data for one week.Based on the short-range analysis method,the processed effective data are divided into 24595 segments and the characteristic values are calculated.The principal component analysis method and the improved K-Means Clustering are used for dimension-reduction and clustering of the short journey characteristic values.Based on the class center distance,the representative short journeys are selected from the clustering results,and then the driving cycle of Hefei electric bus are constructed.The constructed driving cycle is compared with the data collected from real vehicle driving and the typical driving conditions in China and foreign countries,the results show that the constructed driving conditions can more accurately reflect the driving characteristics of Hefei electric bus.

关 键 词：合肥市 电动客车 行驶工况 改进K-均值聚类 

分 类 号：U491.2[交通运输工程—交通运输规划与管理]