利用混合线性回归的城市快速路隧道运行速度建模

Modelling Operating Speed of Urban Expressway Tunnel Using Mixed Linear Regression

摘要: 随着大型城市地上空间日趋饱和，越来越多的城市为缓解交通阻塞，加强交通运行效率而修建的快速路采用了隧道的建设形式。然而隧道中发生的事故往往更加危险，产生的影响也更加严重。准确预测隧道中的运行速度，对隧道的交通安全性和几何设计的评估非常重要。因此本文利用了考虑空间自相关的混合线形回归模型建立了道路上任意位置的运行速度预测模型。该模型不仅考虑了车辆当前所处路段的几何特征，还考虑了上下游路段的几何特征的影响。此外还通过了在不同的实验路段上获取的数据验证了模型的泛化能力，以保证获取的运行速度影响因素和规律具有普适性。研究发现对当前断面运行速度影响较大的特征是上游250米范围内的平均曲率，绝对值平均曲率和平均坡度，直到当前所处断面最远可见处的下游路段和上游路段间的绝对值最小曲率差值以及下游路段的右转路段占比。该模型能为在设计阶段评价隧道设计和更好地理解隧道几何特征对运行速度产生的影响提供帮助。

Abstract: With the increasing saturation of ground space in major cities, many cities have adopted tunnel construction to enhance traffic efficiency. Accurate prediction of operating speed in tunnels is essential for evaluating traffic safety and alignment design of tunnels. This paper uses a mixed linear regression model considered spatial autocorrelation to establish an operating speed prediction model at any position in the tunnel. Data obtained on different experimental tunnels verified the model’s generalization ability to ensure that the rules summarized by the model are universal. We found that the features that have a significant influence on operating speed are the average curvature, the absolute average curvature and the average grade within 250 meters upstream, the absolute minimum curvature difference between 250 meters upstream and downstream until the farthest visible point of the current section, and the proportion of the right-turn section of the downstream segment. The model can be used to evaluate tunnel design and better understand the influence of tunnel geometric features on operating speed.