为了分析影响信控交叉口安全的显著因素,利用分层贝叶斯模型关联交叉口特征和事故率,考虑位于同一主干道上交叉口间的相互关联,并用所建立的模型对交叉口层面和主干道层面的安全影响因素进行了深入分析。结果表明:分层贝叶斯模型比传统模型拟合结果更好;在交叉口层面,高架下的、相互间距近的、十字的、转向车道比例大的、相位多以及日均流量大的交叉口事故率较高;在主干道层面,位于双向行驶的、平均车速高的主干道上交叉口安全性较差。
To analyze the significant risk factors associated with accidents at signalized intersections, a Bayesian hierarchical model correlating intersections features with accidents was developed. With the correlations between intersections along one main road considered the impacts of risk factors were analyzed with the Bayesian hierarchical model. The results show that the proposed model yields better fitting results than the traditional model. At the intersection leve lane leve 1, intersections under the elevated road, in closer proximity, 4-legged, with more ratio, with more phases and greater ADT, 1, intersections locating on the two-way road are associated with more accidents tend to have more accidents. At the main roa and with higher average speed on the main roa g d d