交通信息与安全2017，Vol.35Issue(6)：19-24,53,7.DOI:10.3963/j.issn.1674-4861.2017.06.003
山区公路弯道预警方法研究
An Early Warning Method of Curve Roads in Mountain Areas
摘要
Abstract
Accident rate of curve roads in mountain areas is often higher than which in general sections.Due to sight distance and some reasons,drivers cannot easily identify risks of curve roads.A method for early warning of curve roads is developed.A simulation is set up based on driving modes of front vehicles and road conditions in mountain areas,in order to provide valid information and complete chain transfer of information.The relations between velocity,acceleration,and curves are analyzed.Frequency diagrams and distribution of correlation coefficients are computed.The results show that four variables taken into this model as independent variables is feasible.An early warning model is developed by using the historical data and Elman recurrent neural network.Compared with two binary logistic regression and BP neural network,the accuracy rate of this model is 85.59％.The validity of the model is verified.关键词
交通安全/弯道预警/驾驶状态/Elman神经网络/ROC曲线/阈值Key words
traffic safety/early warning of curve roads/driving mode/Elman neural net/ROC curves/threshold分类
交通工程引用本文复制引用
许多,方守恩,陈雨人..山区公路弯道预警方法研究[J].交通信息与安全,2017,35(6):19-24,53,7.基金项目
国家科技支撑计划课题项目(2014BAG01B06)、国家重点研发计划项目(2016YFC0802701)资助 （2014BAG01B06）
