考虑空间相关性和交通环境影响的宏观事故建模  

Macro-level accident modeling considering spatial correlation&traffic environment impacts

作　　者：靳文舟[1] 包胜男 裴晓航 汤左淦 JIN Wen-zhou;BAO Sheng-nan;PEI Xiao-hang;TANG Zuogan(School of Civil Engineering and Transportation,South China University of Technology,Guangzhou 510640,China;Shenzhen Urban Transport Planning Center Co.,Ltd.,Shenzhen 518057,Guangdong,China)

机构地区：[1]华南理工大学土木与交通学院,广州510640 [2]深圳市城市交通规划设计研究中心,广东深圳518057

出　　处：《安全与环境学报》2023年第3期846-854,共9页Journal of Safety and Environment

基　　金：国家自然科学基金项目(52072128)。

摘　　要：为准确分析宏观尺度下各项因素对发生交通事故的影响,从人口岗位等社会经济特征、交通环境和城市重要设施分布等角度出发,基于深圳市61.4万警情事故和892个交通分析小区建立带空间自回归误差项的空间自回归模型(SARAR),采用广义空间二段最小二乘法进行模型参数估计后,计算出各要素的空间溢出效应。模型结果表明:人口数量、岗位数量、区域面积、学校数量、主次干道长度和主次干道衔接不足均会显著增加本区域和周边区域内的交通事故数量;拥堵路段长度与交通事故的发生不存在显著关系。拟合优度对比结果发现,SARAR模型优于OLS、SAR和SEM模型,是分析宏观尺度下区域安全水平的有效方法。This article studies traffic safety at the macro and the traffic community level.From the perspectives of social and economic characteristics such as population positions,traffic environment,and the distribution of important urban facilities,it considers the spatial autocorrelation of data to analyze the impact of various factors in traffic accidents accurately.Based on the data of the 614 thousand accidents from traffic police and 892 traffic analysis districts in Shenzhen,firstly,this paper adopts the Moran global test method to test the spatial autocorrelation of the sample data;secondly,it establishes the spatial autoregressive model(SAR)and it is estimated by generalized spatial two-stage least square.The results of the model parameter estimation show that the spatial lag parameters of the SAR model and the SEM model are significant at the 1%significance level.It indicates that there is not only spatial autocorrelation in the data but also an error space effect.In this case,a spatial autoregressive model with spatial autoregressive disturbances(SARAR)should be established;Finally,after calculating the spatial spillover effect of each element,the results from SARAR show that the number of populations,the number of job positions,the area,the number of schools,the length of main and secondary roads and the insufficient connection of main and secondary roads would significantly increase the number of traffic accidents in the TAZ and nearby zones.The model results also indicate that it doesn’t have a significant relationship between the length of congested road sections and the occurrence of traffic accidents.Alternatively,the results of the goodness of fit comparison show that the SARAR model is superior to OLS,SAR,and SEM models,and the SARAR model is an effective method to analyze traffic zone safety at the macro level.

关 键 词：安全工程 空间相关性 带空间自回归误差项的空间自回归模型 交通环境 宏观安全分析 

分 类 号：X951[环境科学与工程—安全科学]