城市道路超车特征分析与高风险超车识别

Overtaking Characteristic Analysis and High-Risk Overtaking Recognition on Urban Roads

摘要:

为高效、有针对性地管理城市道路超车行为，提出基于量化特征在线识别高风险超车的方法。首先以交通波理论划分城市路段超车类型，并根据其特性构建多维度指标体系以描述关键特征；然后提出使用车牌识别数据在线计算各项指标的方法；最后基于真实数据，验证该指标体系用于高风险超车识别的有效性。结果表明，可以多项式分别拟合路段超车数和超车幅度总和与流量的关系，以筛选超车频次和强度对流量敏感的路段；K-means算法可根据计划行程速度、单车超车当量速度差将主超车聚为3类，将该类型与车辆超车后行程速度、单车超车当量速度差相结合可实时识别高风险超车；高风险超车常见于相邻交叉口信控相位协调不利的周期，且多发于上游交叉口周期初期和下游交叉口绿灯末期。

Abstract:

To take proper measures to manage the overtaking behavior on urban roads， a method is proposed to recognize high-risk overtaking vehicles online based on quantitative indicators. First， the traffic wave theory is employed to classify overtaking on urban roads. Enlightened by the types， a multi-dimensional index system is constructed to illustrate the essential features of urban-road overtaking. Then， based on license plate recognition data， a method is introduced to calculate these indexes online. Finally， with the real data， the effectiveness of the index system and its calculation are verified to recognize high-risk overtaking. According to the analysis results， both the correlations between the overtaking number and volume and the overtaken number and volume can be fitted by the polynomial， which would help identify volume-sensitive segments. The K-means algorithm is used to cluster the overtaking vehicles into three types according to the difference between the planning speed and the overtaking speed. Given the type， the actual speed and the overtaking speed difference are suitable for evaluating the overtaking risk. It is found that high-risk overtaking frequently occurs in the unfavorable phases resulting from inappropriate signal control coordination between adjacent intersections. Moreover， overtaking with high risk is also prone to be found among the vehicles passing by the end of green light at downstream intersections and the first vehicles passing during green light at upstream intersections.