基于BP网络的驾驶员黄昏空间距离判识规律  

BP Neural Network-Based Space Distance Cognition of Drivers at dusk

作　　者：赵炜华[1] 刘浩学[1] 陈昊[1] 

机构地区：[1]长安大学汽车学院,西安710064

出　　处：《武汉理工大学学报（交通科学与工程版）》2012年第4期744-747,共4页Journal of Wuhan University of Technology（Transportation Science & Engineering）

基　　金：国家自然科学基金项目(批准号:50778023);中央高校基本科研业务费专项资金项目(批准号:CHD2011JC031)资助

摘　　要：为研究黄昏时段行车时,驾驶员对空间距离判识随环境照度的变化规律,进行实际道路试验.选用32名男性驾驶员,在不同自然环境照度下,判识不同空间深度距离时红、绿色障碍物的绝对距离和相对距离.统计分析被试判识结果,分析距离判识差异,获得空间距离判识特征值.运用BP神经网络,模拟距离判识结果,分析距离判识变化规律.结果表明:红、绿色障碍物黄昏时距离判识差异显著,BP网络可以很好拟合距离判识变化规律.绝对距离和相对距离判识结果,均随环境照度降低而增加;判识距离随深度距离增加也增加,相对距离增加较小.In order to explore the laws of distances cognition variant with illumination decreasing in dusk, real road experiments were carried trough. Randomly selected 32 drivers percept absolute space distances and relative ones of red and green obstacles under different environment illumination and depth distance in the real road. Statistical methods was utilized to analyze cognition values and differ- ence significance and character values. Variation results of cognition were simulated by BP neural net- work and the laws were studied. Results showed that there is significant difference between red and green obstacles. Variation trend can well be simulated by BP network. With the environment illumi- nation decreasing absolute and relative distance cognitive values continuously augment. Absolute dis- tance cognition values increase with depth distance increasing but relative ones increase a little.

关 键 词：驾驶员 交通安全 判识距离 照度 黄昏 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] B842.2[交通运输工程—道路与铁道工程]