中文核心期刊
CSCD来源期刊
中国科技核心期刊
RCCSE中国核心学术期刊

Journal of Chongqing Jiaotong University(Natural Science) ›› 2022, Vol. 41 ›› Issue (06): 126-132.DOI: 10.3969/j.issn.1674-0696.2022.06.19

• Transportation Equipment • Previous Articles     Next Articles

Driving Construction of Electric Vehicles Based on t-SNE and Fuzzy Clustering

WANG Jingang1,2,3, XU Hang1,2, LIU Hai1,2, YU Hanzhengnan4, LIU Yu4   

  1. (1. School of Mechanical Engineering, Hebei University of Technology, Tianjin 300130, China; 2. Tianjin Key Laboratory of Power Transmission and Safety Technology for New Energy Vehicles, Tianjin 300130, China; 3. School of Mechanical and Power Engineering, Cangzhou Jiaotong College, Cangzhou 061199, Hebei, China; 4. China Automotive Technology and Research Center Co., Ltd., Tianjin 300300, China)
  • Received:2020-11-23 Revised:2021-03-04 Published:2022-06-22

基于t-SNE与模糊聚类的电动汽车行驶工况构建

王金刚1, 2, 3,徐航1, 2,刘海1, 2,于晗正男4,刘昱4   

  1. (1. 河北工业大学 机械工程学院,天津 300130; 2. 天津市新能源汽车动力传动与安全技术重点实验室,天津 300130; 3. 沧州交通学院 机械与动力工程学院,河北 沧州 061199; 4. 中国汽车技术研究中心有限公司,天津 300300)
  • 作者简介:王金刚(1961—),男,河北沧州人,教授,博士,主要从事车辆工程方面的研究。E-mail:wangjg1961@sina.com 通信作者:刘海(1980—),男,天津人,副教授,博士,主要从事车辆工程方面的研究。E-mail:liuhai@hebut.edu.cn
  • 基金资助:
    国家重点研发计划项目(2018YFB0106403)

Abstract: By constructing the driving conditions of electric vehicles in typical cities, the laws of driving and energy consumption can be predicted. Taking Tianjin as an example, 20 feature parameters and 24 distribution parameters were extracted from the actual road data of 10 battery electric passenger cars for one month. And t-distributed domain embedding algorithm, fuzzy c-means clustering and grey correlation analysis were used to construct the working condition. The validity of the driving conditions was verified by MAPE and K-S tests. Moreover, the typical driving conditions at home and abroad were compared and analyzed. The research results show that the MAPE error between the constructed working condition and the actual road data is 3.82%; K-S test values are 0.0471 and 0.0126. The constructed working conditions are consistent with the actual driving conditions.

Key words: vehicle engineering; pure electric vehicles; driving conditions; t-distributed domain embedding; clustering analysis

摘要: 通过构建典型城市的电动汽车行驶工况,可对其行驶与能耗规律进行预测。以天津市为例,采集10辆纯电动乘用车1个月的实际道路数据,基于短行程片段提取20个特征参数和24个分布参数,使用t分布领域嵌入算法、模糊C均值聚类及灰色关联分析等方法进行工况构建;利用MAPE和K-S检验验证构建工况的有效性;并对比分析了国内外典型的行驶工况。研究结果表明:已构建工况与实际道路数据MAPE误差为3.82%;K-S检验值为0.047 1和0.012 6;所构建工况符合实际的行驶情况。

关键词: 车辆工程;纯电动汽车;行驶工况;t分布领域嵌入;聚类分析

CLC Number: