摘要
为分析高速公路交通流状态参数对交通事故的影响,建立了基于广义线性模型的高速公路交通事故预测模型。采用美国8号州际高速公路的实时交通流数据和交通事故数据,建立了实时交通流状态下的高速公路泊松分布预测模型和负二项分布预测模型,并给出了模型变量弹性系数的计算方法,用以确定高速公路交通事故的突出诱导因素。研究结果表明:广义线性模型能够很好的拟合高速公路交通事故,并且负二项分布预测模型的预测精度高于泊松分布预测模型;交通量、占有率、大车比例和速度标准差是高速公路交通事故的显著影响参数,并且与之呈正向关系;交通量是诱导高速公路交通事故发生最突出因素,交通量增长1%,可导致交通事故增长5.8%。
In order to analysis the impact of traffic state parameters to the freeway crash,a freeway crash prediction model was proposed based on the generalized linear regression models （GLM）.Using the traffic flow data and crash data from I-8E freeway in the United States,Negative binomial （NB）prediction model and Poisson prediction model based on the real-time traffic flow condition was developed, and the elasticity coefficient calculation method used to confirm the prominent traffic induction parameter was proposed.The results show that the freeway crash can be fitted by the GLM,and the predication accuracy of NB model is higher than that of Poisson prediction model.The parameters of traffic volume, share ratio and speed standard deviation have significant affected freeway crashes,and they are in a positive relationship with freeway crash.Traffic volume is the most induction factor that influences freeway crash and an increase of 1 % in traffic volume leads to an increase of 5.8% in freeway crash.
出处
《公路工程》
北大核心
2015年第5期115-119,共5页
Highway Engineering
基金
国家自然科学基金项目(51208337)
天津市市政公路行业科技创新计划项目(2013-04)
关键词
交通工程
高速公路
广义线性模型
弹性系数
traffic engineering
freeway
generalized linear regression models
elasticity coefficient
