城市拥塞环境下交通事故风险传播及控制研究

Study on the traffic accident risk propagation and control in urban congestion environment

为了解城市拥塞环境下交通事故风险因子间的逻辑相互作用以及事故风险传播,提升拥塞环境下风险控制能力,依据拥塞实际调查数据,首先对交通拥塞环境下的事故风险形成及传播过程机理进行分析,以交通冲突量化事故风险水平,标定对于每次交通冲突起绝对作用的一个或几个风险因子。基于复杂网络理论,最终筛选35个风险因子作为网络节点,将拥塞环境形成到消散过程分为9个时间区间,构造风险时间序列,应用Pearson相关性和Granger因果检验相结合的方法构建交通事故风险传播有向网络;引入节点重要度Ki概念,对选取的可人为干预节点改进感染率β′和恢复率λ′,构建一种适用于城市拥塞环境下交通事故风险传播的改进SIR模型。结果表明,对中央分隔带设置、交通信息发布、路面状况、驾驶员路内不规范停车风险节点控制后,感染节点峰值下降23.37%;对考虑节点重要度Ki的4个风险节点控制后,感染节点峰值下降达43.43%,交通事故风险传播规模抑制明显。

This paper aims to study the logical interaction between traffic accident risk factors and the propagation of accident risk in urban congestion environments,to improve the ability to control traffic accident risks in congested environments.The vehicle trajectory of the selected urban congested sections was collected through aerial photography by UAV,and we conducted surveys of the congested sections.Based on the traffic conflict technology,traffic conflicts were used to quantify the level of accident risk and calibrated one or more risk factors that played an absolute role in each traffic conflict.Then,the paper analyzed the formation and propagation process mechanism of accident risk in a traffic congestion environment.Based on the complex network theory,35 risk factors were selected as network nodes.The process of congestion forming to dissipation was divided into 9 intervals.The traffic accident risk efficacy function was calculated for the corresponding period of each risk factor based on the expert scoring method improved by the cloud model.After that,the paper constructed a risk time series.The directed network was constructed by applying a combination of Pearson correlation and Granger causality testing.Node degree value,betweenness centrality,and closeness centrality,the three topological structure attribute indicators reflect the importance of nodes in the network to varying degrees,the concept of node importance Ki was introduced according to these three indicators.This paper improved the infection rateβ'and recovery rateλ'of selected human-intervention nodes and constructed an improved SIR model suitable for the propagation of traffic accident risk in an urban congestion environment.The results show that the traffic accident risk propagation network has a small-world character.By controlling the central separation belt setting,traffic information release,road surface conditions,and irregular parking risk nodes,the peak of the infected nodes drops by 23.37%.After controlling the nodes that consider the importance of nodes Ki in the top 4,the peak of infected nodes decreases by 43.43%,and the scale of traffic accident risk propagation is obviously suppressed.