交通运输系统工程与信息

交通运输系统工程与信息 ›› 2019, Vol. 19 ›› Issue (5): 120-127.

• 系统工程理论与方法 • 上一篇    下一篇

疲劳驾驶转向特征指标的个体差异敏感度分析

吴超仲a, b，孙一帆a, b，张晖* a, b，肖逸影a, b，李徐义a, b   

  1. 武汉理工大学a. 智能交通系统研究中心；b. 水路公路交通安全控制与装备教育部工程研究中心，武汉 430063
  • 收稿日期:2019-05-24 修回日期:2019-07-16 出版日期:2019-10-25 发布日期:2019-10-25
  • 作者简介:吴超仲(1972-)，男，湖北天门人，教授，博士.
  • 基金资助:

    国家重点研发计划项目/ National Key Research and Development Program of China(2017YFC0804802)；国家自然科学基金/ National Natural Science Foundation of China(61603282)；国家自然科学基金联合基金/ Joint Funds of the National Natural Science Foundation of China(u1624262).

Sensitivities of Fatigue Driving Steering Features to Individual Difference

WU Chao-zhonga, b, SUN Yi-fana ,b, ZHANG Huia, b, XIAO Yi-yinga, b, LI Xv-yia, b   

  1. a. Intelligent Transportation Systems Research Center; b. Engineering Research Center for Transportation Safety, Ministry of Education,Wuhan University of Technology,Wuhan 430063, China
  • Received:2019-05-24 Revised:2019-07-16 Online:2019-10-25 Published:2019-10-25

PDF

摘要：

个体差异是影响疲劳驾驶识别的重要因素. 为研究基于转向行为的疲劳驾驶识别受个体差异的影响，本文对疲劳驾驶转向特征指标的个体差异敏感度进行分析. 通过实车试验获得自然驾驶数据，对正常和疲劳状态下的指标进行Kruskal-Wallis(KW)检验，以H 统计量表示指标有效性；以H 统计量最大的单被试为基础逐一与其他被试组成双被试组合，采用线性模型拟合双被试组合的H 统计量和指标个体差异度，以斜率绝对值表征指标个体差异敏感度. 研究获得9 个转向特征指标的个体差异敏感度，结果表明，敏感度越低，指标有效性受个体差异影响越小，其中方向盘转角标准差的个体差异敏感度最低为2.056. 本研究可为转向特征指标的性能评估及疲劳驾驶识别模型的特征选择提供参考.

关键词: 智能交通, 疲劳驾驶转向特征指标, 敏感度分析, 自然驾驶, 个体差异

Abstract:

Individual difference is important factor affecting identification of fatigue driving. For analyzing the effects of individual differences on fatigue driving identification based on steering driving behaviors, sensitivities of fatigue driving steering features to individual differences were analyzed. Naturalistic driving data was obtained by field driving experiments, features under normal and fatigue status were analyzed by Kruskal-Wallis (KW) test, H -statistics was used to indicate the effectiveness of features. Single participant with maximal H -statistics was used as basis to form double participants group with other participants one by one, H - statistics and individual difference degree for features of double participants group were fitted by linear model, and absolute value for slope was used to indicate sensitivities of features to individual differences. Sensitivities of nine steering features to individual differences is obtained, the lower sensitivity is, the less effectiveness of feature is affected by individual differences, and sensitivity for standard deviation of steering wheel angle to individual differences is the lowest whose value is 2.056. This study can offer references for evaluation of fatigue driving steering features' performances and feature selection for recognition model of fatigue driving.

Key words: intelligent transportation, steering features of fatigue driving, sensitivity analysis, naturalistic driving, individual difference

中图分类号: