• CN:11-2187/TH
  • ISSN:0577-6686

机械工程学报 ›› 2021, Vol. 57 ›› Issue (14): 129-140.doi: 10.3901/JME.2021.14.129

• 特邀专栏:电源系统设计、管理与大数据 • 上一篇    下一篇

扫码分享

基于用户大数据的电动汽车驱动系统可靠性试验循环工况构建方法

赵礼辉1,2,3, 王震1, 冯金芝1,2,3, 郑松林1,2,3, 宁欣4   

  1. 1. 上海理工大学机械工程学院 上海 200093;
    2. 机械工业汽车机械零部件强度与可靠性评价重点实验室 上海 200093;
    3. 上海市新能源汽车可靠性评价公共技术平台 上海 200093;
    4. 河南科技学院机电学院 新乡 453003
  • 收稿日期:2020-06-06 修回日期:2021-05-06 出版日期:2021-09-15 发布日期:2021-09-15
  • 通讯作者: 赵礼辉(通信作者),男,1985年出生,博士,副教授。主要研究方向为车辆可靠性评价与轻量化设计。E-mail:Pheigoe@126.com
  • 作者简介:王震,男,1997年出生。主要研究方向为电驱动系统可靠性分析与测试。E-mail:wangzhenares@yeah.net
  • 基金资助:
    国家重点研发计划(2018YFB0104802)和国家自然科学基金(51705322)资助项目

Construction Method for Reliability Test Driving Cycle of Electric Vehicle Drive System Based on Users' Big Data

ZHAO Lihui1,2,3, WANG Zhen1, FENG Jinzhi1,2,3, ZHENG Songlin1,2,3, NING Xin4   

  1. 1. School of Mechanical Engineering, University of Shanghai for Science & Technology, Shanghai 200093;
    2. CMIF Key Laboratory for Strength and Reliability Evaluation of Automotive Structures, Shanghai 200093;
    3. Public Technology Platform for Reliability Evaluation of New Energy Vehicles in Shanghai, Shanghai 200093;
    4. School of Mechanical and Electrical Engineering, Henan Institute of Science and Technology, Xinxiang 453003
  • Received:2020-06-06 Revised:2021-05-06 Online:2021-09-15 Published:2021-09-15

摘要: 为构建与用户载荷关联的电动汽车驱动系统可靠性试验循环工况,基于300个实际用户累计3 540 000 km载荷数据,提取工况片段并关联电驱动系统失效主导载荷构造工况特征参数,运用主成分分析与K-Means聚类分析法将用户使用条件下电驱动系统运行片段分成五种典型工况;基于损伤累积模型,分析不同工况载荷对电驱动系统部件造成的损伤,并确定各工况最优单位损伤强度分布模型,并以损伤分布连续性转折点,筛选单位损伤强度较高的片段作为可靠性试验循环工况;基于马尔科夫状态转移概率矩阵,采用马尔科夫链蒙特卡洛法产生伪随机数对工况拼接,构建电动汽车驱动系统可靠性试验加载谱。通过构造的可靠性试验载荷谱与标准循环工况及用户损伤对比,表明构建的电驱动系统可靠性试验加载谱具有较高的损伤强度,能够有效涵盖90%以上的用户使用强度,从而为电动汽车驱/传动系可靠性设计与验证提供参考和依据。

关键词: 用户大数据, 电驱动系统, 可靠性试验, 工况构建, 马尔科夫过程

Abstract: To construct the reliability test driving cycle of the electric vehicle drive system correlated with users' load, based on 300 actual users load data with 3.54 million kilometers, fragment of conditions are extracted and characteristic parameters are constructed with the failure dominant load of the electric drive system. Using principal component analysis and K-Means clustering analysis methods, the operating segments of the electric drive system under user conditions are divided into five typical conditions. Based on the damage accumulation model, the damage caused to the components of the electric drive system under different conditions are analyzed, and the optimal unit damage intensity distribution models are determined for each condition. Based on the continuity turning point of the damage distribution, the segments with higher unit damage strength are selected as reliability test driving cycle. Based on Markov state transition probability matrix, using Markov chain Monte Carlo method to generate pseudo-random numbers for stitching of conditions, then the reliability test load spectrum of the electric vehicle drive system is constructed. The comparison of the constructed reliability test load spectrum with the standard cyclic conditions and customers damage shows that the reliability test loading spectrum of the electric drive system is constructed with higher damage strength, which can effectively cover more than 90% of the strength under user conditions, so as to provide reference and basis for the reliability design and verification of electric vehicle drive or transmission system.

Key words: users' big data, electric vehicle drive system, reliability test, construction of conditions, Markov process

中图分类号: