为鉴别山区高速公路交通事故影响因素并确定相邻路段间安全效应的空间关联作用，以云南省某山区高速公路为研究对象，从交通运行条件、道路几何线形两方面收集事故影响因素，整理了 3 258起交通事故数据，利用同质法划分出686个路段(即事故建模时的基本路段单元)。在现有负二项(Negative Binomial，NB)模型和混合效应负二项(Mixed-Effects Negative Binomial，MENB)模型的基础上加入空间关联效应项，构建了山区高速公路事故空间关联效应(Space Related Partie，SP)模型，并与NB和MENB进行了对比分析。通过模型估计及敏感性分析，识别出对事故风险有显著影响的年平均日交通量、路段长度、平曲线偏角与曲率、纵坡坡度、竖曲线坡差绝对值、平均坡度七个因素。模型检验及对比分析结果表明:SP模型整体表现优于NB模型，拟合优度优于MENB模型，可对事故影响因素进行安全效应分析;立交路段事故风险高于基本路段;路段前后的平曲线偏角、纵坡坡度等越大，对路段产生的风险溢出越大。
Study on accident analysis model of mountainous highways considering Space Related Partie effect
The article conducts an extensive investigation into the causes of traffic accidents on mountainous highways,specifically focusing on a freeway located in Yunnan Province.By analyzing data from 3 258 traffic accidents(including information on accident timing,location,climatic conditions,etc.)and evaluating factors such as traffic conditions and road structure,the research aims to provide insights into the safety relationship between road segments.The researchers utilized a homogeneous method to segment the highway based on actual traffic conditions and road characteristics,resulting in 686 segments as the fundamental units for accident modeling.Employing advanced statistical techniques,the study enhanced existing models such as the Negative Binomial(NB)model and Mixed-Effects Negative Binomial(MENB)model by incorporating spatial correlation effect terms.This development led to the creation of a Space Related Partie(SP)model,enabling a more comprehensive analysis of traffic accidents on mountainous highways.This innovative model was compared and evaluated against NB and MENB to assess its effectiveness in analyzing accident safety.Through model estimation results and sensitivity analysis,this study examined the safety impacts of accident influencing factors.It identified seven significant factors:annual average daily traffic volume,road section length,deflection and curvature of flat curves,longitudinal slope,slope difference of vertical curves,and average slope,all of which contribute significantly to accident risk.Based on the results of model testing and comparative analysis,the SP model demonstrates superior overall performance compared to the NB model,with a better fit than the MENB model,making it suitable for analyzing the safety impacts of factors influencing accidents.The study identifies higher accident risks at interchange sections than on basic road segments.Furthermore,greater deflection angles of flat curves and longitudinal slopes before and after road sections correspond to increased risk on those segments.
safety engineeringmountainous highwaysaccident influencing factorssafety effectsNegative Binomial(NB)modelSpace Related Partie(SP)model
万方数据
维普
