新能源汽车与行人交通事故严重程度分析  

Analysis of the severity of traffic accidents between new energy vehicles and pedestrians

作　　者：张道文 董鑫驰 雷毅 黎华惠 罗晶 张诚龙 赵成一 汤楷文 ZHANG Daowen;DONG Xinchi;LEI Yi;LI Huahui;LUO Jing;ZHANG Chenglong;ZHAO Chengyi;TANG Kaiwen(School of Automobile and Transportation,Xihua University,Chengdu 610039,China;Vehicle Measurement Control and Safety Key Laboratory of Sichuan Province,Chengdu 610039,China;Provincial Engineering Research Center for New Energy Vehicle Intelligent Control and Simulation Test Technology of Sichuan,Chengdu 610039,China;College of Intelligent Manufacturing and Automotive,Chengdu Industrial Vocational Technical College,Chengdu 610218,China)

机构地区：[1]西华大学汽车与交通学院,成都610039 [2]汽车测控与安全四川省重点实验室,成都610039 [3]四川省新能源汽车智能控制与仿真测试技术工程研究中心,成都610039 [4]成都工业职业技术学院智能制造与汽车学院,成都610218

出　　处：《安全与环境学报》2024年第3期1061-1069,共9页Journal of Safety and Environment

基　　金：国家市场监督管理总局项目(202248);四川省道路交通安全隐患省级挂牌督办技术支持服务(225118);四川省重点实验室课题(QCCK2021-011)。

摘　　要：为探究影响新能源汽车与行人交通事故严重程度的诱因,基于英国近4 a发生的1819条新能源汽车与行人交通事故数据,从驾驶员、行人、新能源车辆、道路、时空和环境5个方面选择了19个影响因素作为协变量,并根据数据特征将因变量分为轻伤和严重伤害两类。考虑到数据异质性,首先通过因子分析法与k-means聚类进行聚类分析,然后采用Logistic模型对聚类分析得到的两类集群进一步分析,从预测准确率和协变量共线性指标两个方面确认了模型的适用度。结果表明:基于聚类分析构建的Logistic模型预测准确率均达到80%以上;变道或超车、车龄、支路及环形交叉口、双幅路(有中央分隔带)对事故伤害严重程度有异质性影响;行人年龄大于46岁和夜间行车会明显增加事故严重程度;其中,变道或超车和行人年龄对新能源汽车事故严重程度的影响比传统人-车事故更加显著。To investigate the causal factors influencing the severity of traffic accidents between new energy vehicles and pedestrians,this paper selects 19 influencing factors as covariates from a total of five aspects:pedestrians,new energy vehicles,road characteristics,spatial and temporal,and environmental characteristics.This study is based on 1819 new energy vehicle and pedestrian traffic accident data in the last four years in the UK and classifies the dependent variables into two categories of minor injuries and serious injuries according to the data characteristics.Considering the data heterogeneity,firstly,a binomial logistic severity model was established for new energy vehicle and pedestrian accidents by cluster analysis with k-means clustering,and the two types of clusters obtained from the cluster analysis were analyzed separately.To avoid strong correlations among the 19 selected variables,which may affect the accuracy of the analysis results,the selected variables were tested for covariance.The test results show that the variance inflation factor Variance Inflation Factor(VIF)values of the selected 19 covariates are significantly less than 10,which proves that there is no multicollinearity among the covariates and can be used as the basis for subsequent model studies.Besides,the applicability of the model was confirmed by the prediction accuracy.The results show that:the Logistic model constructed based on cluster analysis has a high prediction accuracy,all reaching more than 80 percent.Changing lanes or overtaking,vehicle age,branch roads and roundabout intersections,and double-width roads with segregation zones have heterogeneous effects on accident injury severity.Pedestrians older than 46 years old and driving at night would significantly affect accident severity.Among them,there are differences in vehicle driving intention and pedestrian age than traditional human-vehicle accident severity influencing factors:new energy vehicles increase accident severity when changing lanes to overtake,which is a facto

关 键 词：安全工程 新能源汽车 行人交通事故 严重程度 聚类分析 二项Logistic模型 

分 类 号：X951[环境科学与工程—安全科学]