基于OBD数据的危险驾驶行为与建成环境空间关联研究

摘要

With applications of new generation of information technologies in the field of transportation,such as broadband mobile communication and Internet of Things,studies of driving behavior for traffic safety in the environment of Mobile Internet attract more attention.In order to complement previous studies on the spatial analysis of risky driving behavior and Connected Vehicle,risky driving behaviors are identified and extracted on basis of On-board Diagnosis (OBD)data.Spatial distribution of risky driving behavior is thus analyzed in terms of Traffic Analysis Zone (TAZ).To study the mechanism of spatial discrepancy on risky driving behavior,Point of Interests (POI)data are utilized to meas-ure the built environment of cities.Variables of significant effects on risky driving behavior are identified by an ordinary least square (OLS)regression model.Moreover,the coefficients of different environmental variables on risky driving be-haviors are evaluated by a geographically weighted regression (GWR)model.The results show that the GWR model is superior to the OLS regression method,which is able to effectively present spatial-temporal characteristics of the relation-ship between built environment and risky driving behavior.Study results can be used to support decision-making related to traffic safety improvement program of traffic management and planning agencies.%随着宽带移动通信、物联网等新一代信息技术在交通领域中的应用,面向交通安全的移动互联环境下驾驶行为研究成为热点课题.为弥补现有研究中对车联网数据分析较少或对危险驾驶行为空间分析不足等缺陷,基于车辆自诊断系统(OBD)数据对危险驾驶行为进行了空间识别与提取,并基于交通小区(TAZ)分析了危险驾驶行为的空间分布差异.研究揭示了危险驾驶行为空间分布差异的内在机理,利用百度兴趣点(POI)数据度量了城市建成环境因素,通过最小二乘法(OLS)回归模型识别出建成环境对危险驾驶行为显著影响的变量,在此基础上采用基于地理加权回归(GWR)模型得出了不同建成环境变量对危险驾驶行为的空间影响系数.模型显示,采用GWR模型拟合结果优于OLS一倍,并且可以有效地揭示出空间建成环境对危险驾驶行为影响的时空特征,为交通管理与规划部门制定措施或政策提供了决策支持.