基于时间序列聚类的交通事故黑点识别与分析

Black spot identification and analysis of traffic accidents based on time series clustering

【目的】鉴别道路交通事故黑点以及探究事故高发区域的致因。【方法】结合地理信息软件与可解释性机器学习算法,提出一种考虑交通事故时空属性的事故黑点识别及致因分析的方法。【结果】事故高发区域主要聚集在大型商业圈、客运车站与工业区附近,其事故密度为事故低发区域的6.61倍。在事故高发区域,起决定性影响的因素是碰撞形态、天气、能见度、车道类型及两轮车类型。而在事故低发区域,碰撞形态、道路等级、能见度、路面材料以及两轮车类型为主要影响因素。此外,路口路段类型、车道类型、两轮车类型以及肇事逃逸等因素在不同事故区域的影响不同。【结论】交通事故在城市内存在事故黑点,且部分道路环境因素在不同事故区域的影响不一致。研究成果可为交管部门针对事故黑点区域制定防范措施提供指导。

[Purposes]Identify the black spots of road traffic accidents and explore the causes of the accidents in black spots area.[Methods]Combining geographic information software and interpretable machine learning algorithm,this paper proposes a method of black spot identification and analysis considering temporal and spatial attributes of traffic accidents.[Findings]The accident density in the area with high accident incidence was mainly concentrated in the vicinity of large commercial area,passenger station and industrial area.The accident density in the area with high accident incidence was 6.61 times that in the area with low accident incidence.In areas with high accident rates,collision patterns,weather,visibility,lane type and two-wheeler type play a decisive role.In the low accident area,collision form,road class,visibility,road material and two-wheel vehicle type are the main influencing factors.In addition,factors such as intersection section type,lane type,two-wheeled vehicle type and hit-and-run have different impacts in different accident areas.[Conclusions]There are black spots in urban traffic accidents,and the influence of some road environmental factors is inconsistent in different accident areas.The results can provide guidance for the traffic management department to formulate preventive measures for the black spot area.