张鑫

摘 要：基于昆明市20个两相位十字交叉口167小时的交通冲突、交通流量以及交叉口设计要素等数据，检验不同交通量状态下 15min直左交通沖突分布，应用直左交通冲突模型分析交叉口安全影响因素及冲突产生原因。利用广义线性模型分别构建总体预测模型和分交通状态的预测模型，结合两种模型的拟合优度和预测精度进行模型优选。结果表明：15min直左交通冲突在不同交通量状态下均服从负二项分布，考虑交通状态的直左冲突模型拟合优度和预测精度均优于总体模型。不同交通状态下冲突流量对交通冲突影响程度不同；左转导流线能够有效降低交通冲突，且在左转车流饱和度较高状态下的降低效果更明显。交通冲突模型的建立可进一步拓展交通冲突技术在交通安全分析与评价中的应用。

关键词：交通安全；交通冲突；交通安全分析；广义线性模型；信号交叉口

中图分类号：U491.31 文献标志码：A

Analyzing the safety of signalized intersections with permitted left-turn phases using traffic conflict-predictive models

ZHANG Xin（Government Service Center of Beijing Municipal Transport Commission，Beijing Ships Inspection Center，Beijing100161，China）

Abstract： The traffic conflict data， traffic volume data and intersection design elements were extracted from 167 hours video data at 20 signalized intersections where the permitted left-turn phases were used. Using the data， the underlying distributions of the conflict frequency for different volume regimes in 15-min time intervals were examined. And the opposing left-turn traffic conflict-predictive models were developed to analyze the safety performance of signalized intersection and find the influence factors of traffic conflict. Using Different conflict- predictive models were developed， including an overall negative binomial model and the separate models developed for different traffic scenarios which were defined based on the volume to capacity ratio of the conflicting traffic flows. The goodness of fit and prediction performance of different models were compared. It was found that the conflict frequency in 15-min time intervals generally followed a negative binominal distribution. And the generalized linear regression models developed for different traffic scenarios outperformed the overall NB model for the goodness of fit and prediction performance. The coefficients of the conflict prediction models showed that the conflicting traffic volumes had different effects on the occurrences of opposing left-turn conflicts under different traffic scenarios. The coefficients of the indicator variables for the presence of white line extension indicated that using the white line extension was helpful to reduce the opposing left-turn conflicts where the reduce effect was more obvious with the higher vL/cL of the left-turn volume. The use of conflict predictive models has potential to expand the uses of surrogate safety measures in safety estimation and evaluation.

Keywords：Traffic safety；traffic conflict；safety analysis；generalized linear model；signalized intersection

传统的交通安全评价采用交通事故数据作为评价指标，通过建立交通事故模型用以评价某段时间内期望事故频次与安全影响因素之间的关聯关系，然而交通事故数据采集时间长、获取困难且质量不稳定，特别是在交通事故数据库不完备、不精准的情况下，交通事故模型的广泛应用受到制约。交通冲突作为替代交通安全评价指标，因其与交通事故均具有离散性、随机性并受交通量、交通设计、交通控制等因素影响，理论上可将交通事故建模方法应用于交通冲突建模，从而利用冲突数据易采集、易观测等优势，建立交通冲突预测模型进行交通安全评价。……