高速公路追尾碰撞事故起数预测模型研究

发布时间：2018-09-03 15:57

【摘要】：为分析显著影响追尾碰撞事故的因素,分别采用负二项(NB)回归模型和零堆积负二项(ZINB)回归模型建立追尾碰撞事故起数预测模型。通过模型的拟合优度检验和准确性检验,对比2种模型的优劣,采用弹性分析确定自变量对因变量的影响程度。以2006—2008年某高速公路发生的追尾碰撞事故进行实证性研究。结果表明:ZINB回归模型比NB回归模型具有更好的拟合效果和预测准确性;在ZINB回归模型中,曲度、曲线比例、车道宽度与追尾碰撞事故起数显著正相关,曲率变化率、相邻路段坡差和弯坡组合与追尾碰撞事故起数显著负相关;车道宽度、曲率变化率和曲度对追尾碰撞事故的发生影响尤为明显。
[Abstract]:In order to analyze the factors that influence the collision accident significantly, the negative binomial (NB) regression model and the zero stacking negative binomial (ZINB) regression model are used to establish the prediction model of the initial number of rear-end collision accidents. Through the goodness of fit test and veracity test of the model, the advantages and disadvantages of the two models are compared, and the degree of influence of independent variables on dependent variables is determined by elastic analysis. An empirical study was carried out on a rear-end collision accident on a highway from 2006 to 2008. The results show that the NB regression model has better fitting effect and prediction accuracy than the NB regression model, and in the ZINB regression model, the curvature, curve ratio, lane width and the initial number of collision accidents are significantly positively correlated, and the curvature change rate is significant. The slope difference and the combination of curved slope of adjacent sections are significantly negatively correlated with the number of rear-end collision accidents, and the impact of lane width, curvature change rate and curvature on the occurrence of rear-end collision accidents is particularly obvious.
【作者单位】：长安大学汽车学院;
【基金】：国家自然科学基金资助(51208052,51308058) 陕西省科学技术研究发展计划项目(2013K13-04-02) 陕西省交通科技项目(10-22K)
【分类号】：U491.3

【参考文献】