驾驶行为险态辨识方法  

A Method for Identifying Risk Status of Driving Behavior

作　　者：马国忠[1] 李燕[1] 郭孜政[1] 杨露 

机构地区：[1]西南交通大学交通运输与物流学院,四川成都610031 [2]四川省交通投资集团有限公司,四川成都610031

出　　处：《公路交通科技》2013年第7期113-118,共6页Journal of Highway and Transportation Research and Development

基　　金：国家自然科学基金项目(51108390);国家自然科学基金项目(51108040);中国博士后基金项目(2012M510051)

摘　　要：为实现险性驾驶行为状态的有效辨识,提出了一套驾驶行为险态辨识方法。以单位时间误操作率为依据,采用自底向上的分段算法实现了行为险态分级,采用因子方差分析,选取听觉感知、动视野、动视力、色觉、暗适应、注意力、判断能力、反应时这8项因子构成驾驶行为状态因子集,构建驾驶行为险态辨识特征向量,然后再对行为状态指标数据予以预先分级的前提下,采用单因子分析法对试验数据予以分析。并设计出BP神经网络行为险态辨识模型,最后进行了实例分析与计算。分析结果表明:反应时、注意力、判断能力3项指标在各分级间差异显著,故可作为驾驶行为险态辨识主因子,行为状态错判误差率为2.5%。In order to realize the effective identification of dangerous driving behavior, a risk status identification method of driving behavior is proposed. Based on the rate of misoperation in unit time, behavior risk status are classified by bottom-up sectional algorithm. By analysis of variance, the significance factors, which are hearing, dynamic visual field, dynamic vision, colour vision, dark adaptation, attentiveness, judgment ability and reaction time, are selected to form the status factor set of driving behavior, and the characteristic vectors of driving behavior risk status identification is proposed. After classifying the driving behavior status indexes, the experiment data are analyzed by single factor method. The BP neural network model of driving behavior risk status identification is designed and a case study is performed. The result shows that there are obvious differences among reaction time, attentiveness and judgment ability on different classes. Thus these indexes can be used as main factors for driving behavior risk status identification, and the error vate of misidentification is 2. 5 %.

关 键 词：交通工程 交通安全 险态辨识 自底向上的分段算法 驾驶行为 BP神经网络 

分 类 号：U491.254[交通运输工程—交通运输规划与管理]