货车交通事故严重程度影响因素分析  

Analysis on factors influencing severity of truck traffic accidents

作　　者：胡焱松 王长君 张勇刚[2] 褚宇航 HU Yansong;WANG Changjun;ZHANG Yonggang;CHU Yuhang(School of Traffic Management,People’s Public Security University of China,Beijing 100038,China;School of Public Security and Traffic Management,Guangdong Police College,Guangzhou Guangdong 510230,China;Research Institute for Road Safety of MPS,Beijing 100176,China)

机构地区：[1]中国人民公安大学交通管理学院,北京100038 [2]广东警官学院治安与交通管理学院,广东广州510230 [3]公安部道路交通安全研究中心,北京100176

出　　处：《中国安全生产科学技术》2025年第4期142-150,共9页Journal of Safety Science and Technology

基　　金：公安部部级课题(2022JSZ16);广东省教育厅项目(2023KQNCX050);广东警官学院校级课题项目(2024YBZB03)。

摘　　要：为探究货车交通事故严重程度的主要影响因素及其耦合关系,基于我国南方某城市2021—2023年货车交通事故数据,采用特征分类、K-means聚类以及特征融合的方法提取初始影响因素,并构建XGBoost模型、有序Logit回归模型和树形贝叶斯网络模型,对影响因素的重要性、单因素和因素耦合进行量化分析。研究结果表明:碰撞对象、道路类型、超载等7个因素对事故严重程度影响显著,其中碰撞对象最为重要;碰撞对象分别与重型车辆、超载、箱式货车、工程运输车辆、转弯变道、有隔离等12个影响因素耦合会大幅提高事故严重程度,当碰撞行人与阴天、超载、工程运输车辆以及转弯变道等耦合时,发生重大事故的概率超过40%。研究结果可为货车交通事故防控提供决策参考。To investigate the main influencing factors and their coupling relationship in the severity of truck traffic accidents,the initial influencing factors were extracted through the feature classification,K-means clustering,and feature fusion,based on the data of truck traffic accidents from a southern city in China from 2021 to 2023.The XGBoost,ordered Logit regression,and tree-structured Bayesian network models were used to quantitatively analyze the importance of these factors,as well as single-factor and factor coupling effects.The results show that seven factors,including collision object,road type,and overload,significantly influence the accident severity,with the collision object being the most critical.Furthermore,the coupling of the collision object with 12 other factors,such as heavy vehicles,overload,box trucks,engineering transportation vehicles,lane changing or turning,and with barriers,substantially increase the accident severity.When the collision with pedestrians is coupled with overcast weather,overload,engineering transportation vehicles,and lane changing or turning,the probability of major accidents exceeds 40%.The research results can provide decision-making reference for the prevention and control of truck traffic accidents.

关 键 词：交通安全 货车 事故严重程度 影响因素 有序Logit回归 树型贝叶斯网络 

分 类 号：X951[环境科学与工程—安全科学] U491.3[交通运输工程—交通运输规划与管理]