基于全局K-means聚类算法的汽车行驶工况构建
|
摘要
为构建符合郑州市交通特征的乘用车典型行驶工况,选取60辆乘用车进行数据采集,将采集的有效行驶数据划分成120 744条运动学片段,采用主成分分析法对构建的特征参数矩阵进行降维、非线性化处理。选取戴维森-堡丁指数来确定最佳聚类数,并通过全局K-means聚类算法将主成分分析得到的4个主成分聚成3类,然后采用相关系数法从各类片段库中选取具有代表性的运动学片段,从而构建出郑州市乘用车典型循环工况(ZZ_DC)。通过与传统K-means聚类算法构建的工况比较,采用全局K-means聚类构建的工况更加准确地反映了郑州市乘用车在实际道路上的行驶状态。将ZZ_DC工况与国内一直采用的乘用车测试工况(NEDC工况)比较,结果表明,ZZ_DC工况与NEDC工况差异显著,因此,有必要开发出适用于郑州市交通特征的乘用车行驶工况。
In order to construct the typical driving cycle of the passenger cars to meet the traffic characteristics of Zhengzhou city,60 passenger cars were selected to collect the data,then the effective driving data were divided into 120 744 kinematics fragments,and principal component analysis method was used to reduce the dimensions and to execute the nonlinear processing of the constructed characteristic parameter matrix. DaviesBouldin index( DB) was chosen to determine the optimal cluster number,and the four principal components obtained by the principal component analysis method were clustered into three categories by the global Kmeans clustering algorithm,then the representative kinematics fragments were selected from fragment libraries by means of the correlation coefficient method,thus the typical driving cycle of Zhengzhou passenger cars( ZZ_DC) was constructed. Compared with the driving cycle constructed by the traditional K-means clustering algorithm,the driving cycle constructed by the global K-means clustering could more accurately reflect the driving status of the passenger cars on the actual roads in Zhengzhou. The ZZ_DC was compared with the test driving cycle of the domestic passenger cars( NEDC),and the results showed that the differences between the ZZ_DC and the NEDC were significant. Therefore,it was necessary to develop the driving cycle suitable for the traffic characteristics of Zhengzhou.
In order to construct the typical driving cycle of the passenger cars to meet the traffic characteristics of Zhengzhou city,60 passenger cars were selected to collect the data,then the effective driving data were divided into 120 744 kinematics fragments,and principal component analysis method was used to reduce the dimensions and to execute the nonlinear processing of the constructed characteristic parameter matrix. DaviesBouldin index( DB) was chosen to determine the optimal cluster number,and the four principal components obtained by the principal component analysis method were clustered into three categories by the global Kmeans clustering algorithm,then the representative kinematics fragments were selected from fragment libraries by means of the correlation coefficient method,thus the typical driving cycle of Zhengzhou passenger cars( ZZ_DC) was constructed. Compared with the driving cycle constructed by the traditional K-means clustering algorithm,the driving cycle constructed by the global K-means clustering could more accurately reflect the driving status of the passenger cars on the actual roads in Zhengzhou. The ZZ_DC was compared with the test driving cycle of the domestic passenger cars( NEDC),and the results showed that the differences between the ZZ_DC and the NEDC were significant. Therefore,it was necessary to develop the driving cycle suitable for the traffic characteristics of Zhengzhou.
引文
[1]万霞,黄文伟,强明明.深圳市乘用车道路行驶工况构建[J].深圳大学学报(理工版),2016,33(3):281-287.WAN X,HUANG W W,QIANG M M. Construction of driving cycle for passenger vehicles in Shenzhen[J].Journal of Shenzhen University(Science and Engineering),2016,33(3):281-287.
[2] FOTOUHI A,MONTAZERIGH M. Tehran driving cycle development using the K-means clustering method[J]. Scientia Iranica A,2013(20):286-293.
[3]彭育辉,杨辉宝,李孟良,等.基于K-均值聚类分析的城市道路汽车行驶工况构建方法研究[J].汽车技术,2017(11):13-18.PENG Y H,YANG H B,LI M L,et al. Research on the construction method of driving cycle for the city car based on K-means cluster analysis[J]. Automobile Technology,2017(11)13-18.
[4]胡宸,吴晓刚,李晓军,等.哈尔滨城市公交工况的构建[J].哈尔滨理工大学学报,2014,19(1):85-89.HU C,WU X G,LI X J,et al. Construction of Harbin city driving cycle[J]. Journal of Harbin University of Science and Technology,2014,19(1):85-89.
[5]胡志远,秦艳,谭丕强,等.基于大样本的上海市乘用车行驶工况构建[J].同济大学学报(自然科学版),2015,43(10):1523-1527.HU Z Y,QIN Y,TAN P Q,et al. Large-sample-based car-driving cycle in Shanghai city[J]. Journal of Tongji University(Natural Science),2015,43(10):1523-1527.
[6]石琴,仇多洋,周洁瑜.基于组合聚类法的行驶工况构建与精度分析[J].汽车工程,2012,34(2):164-169.SHI Q,QIU D Y,ZHOU J Y. Driving cycle construction and accuracy analysis based on combined clustering technique[J]. Automotive Engineering,2012,34(2):164-169.
[7]公安部交通管理局. 2016年全国机动车和驾驶人保持快速增长. 2017-01-10. http://www. mps. gov. cn/n2255040/n4908728/c5595634/content. html.TRAFFIC MANAGEMENT BUREAU OF THE PUBLIC SECURITY MINISTRY. The nation’s motor vehicles and drivers maintained rapid growth in 2016. 2017-01-10. http://www. mps. gov. cn/n2255040/n4908728/c5595634/content. html.
[8]马志雄,李孟良,朱西产,等.乘用车实际行驶工况开发方法的研究[J].武汉理工大学学报(信息与管理工程版),2004,26(3):182-184.MA Z X,LI M L,ZHU X C,et al. Study of the methodology for car driving cycle development[J]. Journal of Wuhan University of Technology(Information&Management Engineering),2004,26(3):182-184.
[9]黄万友,程勇,李闯.基于车辆能耗状态的济南市道路行驶工况构建[J].西南交通大学学报,2012,47(6):989-995.HUANG W Y,CHENG Y,LI C. Driving cycle construction of city road based on vehicles energy consumption in Jinan[J]. Journal of Southwest Jiaotong University,2012,47(6):989-995.
[10]赵丽.全局K-均值聚类算法研究与改进[D].西安:西安电子科技大学,2013.ZHAO L. Research and improvement of the global Kmeans clustering algorithm[D]. Xi’an:Xidian University,2013.
[11] ARISTIDIS L,NIKOS V,JAKOB. The global K-means clustering algorithm[J]. Pattern Recognition,2003,36:451-461.
[12]周士兵.聚类分析中的最佳聚类数确定方法研究应用[D].无锡:江南大学,2011.ZHOU S B,Research and application on determining optimal number of clusters in cluster analysis[D].Wuxi:Jiangnan University,2011.
[13]秦大同,詹森,漆正刚,等.基于K-均值聚类算法的行驶工况构建方法[J].吉林大学学报(工学版),2016,46(2):383-389.QIN D T,ZHAN S,QI Z G,et al. Driving cycle construction using K-means clustering method[J]. Journal of Jilin University(Engineering and Technology Edition),2016,46(2):383-389.
[2] FOTOUHI A,MONTAZERIGH M. Tehran driving cycle development using the K-means clustering method[J]. Scientia Iranica A,2013(20):286-293.
[3]彭育辉,杨辉宝,李孟良,等.基于K-均值聚类分析的城市道路汽车行驶工况构建方法研究[J].汽车技术,2017(11):13-18.PENG Y H,YANG H B,LI M L,et al. Research on the construction method of driving cycle for the city car based on K-means cluster analysis[J]. Automobile Technology,2017(11)13-18.
[4]胡宸,吴晓刚,李晓军,等.哈尔滨城市公交工况的构建[J].哈尔滨理工大学学报,2014,19(1):85-89.HU C,WU X G,LI X J,et al. Construction of Harbin city driving cycle[J]. Journal of Harbin University of Science and Technology,2014,19(1):85-89.
[5]胡志远,秦艳,谭丕强,等.基于大样本的上海市乘用车行驶工况构建[J].同济大学学报(自然科学版),2015,43(10):1523-1527.HU Z Y,QIN Y,TAN P Q,et al. Large-sample-based car-driving cycle in Shanghai city[J]. Journal of Tongji University(Natural Science),2015,43(10):1523-1527.
[6]石琴,仇多洋,周洁瑜.基于组合聚类法的行驶工况构建与精度分析[J].汽车工程,2012,34(2):164-169.SHI Q,QIU D Y,ZHOU J Y. Driving cycle construction and accuracy analysis based on combined clustering technique[J]. Automotive Engineering,2012,34(2):164-169.
[7]公安部交通管理局. 2016年全国机动车和驾驶人保持快速增长. 2017-01-10. http://www. mps. gov. cn/n2255040/n4908728/c5595634/content. html.TRAFFIC MANAGEMENT BUREAU OF THE PUBLIC SECURITY MINISTRY. The nation’s motor vehicles and drivers maintained rapid growth in 2016. 2017-01-10. http://www. mps. gov. cn/n2255040/n4908728/c5595634/content. html.
[8]马志雄,李孟良,朱西产,等.乘用车实际行驶工况开发方法的研究[J].武汉理工大学学报(信息与管理工程版),2004,26(3):182-184.MA Z X,LI M L,ZHU X C,et al. Study of the methodology for car driving cycle development[J]. Journal of Wuhan University of Technology(Information&Management Engineering),2004,26(3):182-184.
[9]黄万友,程勇,李闯.基于车辆能耗状态的济南市道路行驶工况构建[J].西南交通大学学报,2012,47(6):989-995.HUANG W Y,CHENG Y,LI C. Driving cycle construction of city road based on vehicles energy consumption in Jinan[J]. Journal of Southwest Jiaotong University,2012,47(6):989-995.
[10]赵丽.全局K-均值聚类算法研究与改进[D].西安:西安电子科技大学,2013.ZHAO L. Research and improvement of the global Kmeans clustering algorithm[D]. Xi’an:Xidian University,2013.
[11] ARISTIDIS L,NIKOS V,JAKOB. The global K-means clustering algorithm[J]. Pattern Recognition,2003,36:451-461.
[12]周士兵.聚类分析中的最佳聚类数确定方法研究应用[D].无锡:江南大学,2011.ZHOU S B,Research and application on determining optimal number of clusters in cluster analysis[D].Wuxi:Jiangnan University,2011.
[13]秦大同,詹森,漆正刚,等.基于K-均值聚类算法的行驶工况构建方法[J].吉林大学学报(工学版),2016,46(2):383-389.QIN D T,ZHAN S,QI Z G,et al. Driving cycle construction using K-means clustering method[J]. Journal of Jilin University(Engineering and Technology Edition),2016,46(2):383-389.