为提高道路运行速度预测模型的准确性,提出将传统时序分析的向量自回归（Vector Autoregression, VAR）模型用于公路逐桩线形单元序列分析的方法。对道路线形的二维设计指标进行三维转换,得到完整且唯一确定公路曲线形状、走向及弯扭曲方向的三维线形指标参数,即线形的三维曲率、挠率、切向量、法向量等;对公路逐桩步长序列构建VAR模型,将三维指标与实车驾驶试验得到的运行速度构成联合内生变量从而建立联系。试验结果表明:线形的曲率、挠率、切向量竖向分量、切向量平面分量差分、主法向量竖向分量、副法向量平面分量差分均有90%置信度以上对运行速度造成显著影响;运行速度随线形的曲率、切向量竖向分量、切向量平面分量差分的增大而发生负响应,随线形的挠率、副法向量平面分量差分的增大而发生正响应。最后,根据运行速度对道路三维几何特征的脉冲响应结果,描述了200m内的具体影响效应,并分析了变量呈现其响应结果的原因,验证了道路线形三维特征与运行车速的密切相关性以及驾驶人对道路条件的认知规律。

In order to improve the accuracy of operating speed prediction model, the traditional sequential VAR(Vector Autoregression) model was applied to the alignment unit sequence analysis of highway pile by pile. The 2D design parameters of highway alignment were transformed into 3D geometric parameters. These parameters could completely and uniquely determine the shape, trend, bending and twisting direction of highway, i.e. curvature, torsion, tangent vectors and normal vectors. The sequences of highway piles were used to construct VAR model, and the relationships between each 3D feature and operating speed from the driving test were established. Results of tests indicated that curvature, torsion, vertical component of tangent vector, horizontal component increment of tangent vector, vertical component of main normal vector and horizontal component increment of sub normal vector of alignment were significantly related to operating speed with the confidence level of more than 90%. The operating speed had a negative response with the increase of curvature, vertical vector and difference of horizontal component of tangent vector of alignment, and a positive response with the increase of torsion and difference of horizontal component of sub normal vector of alignment. At last, according to the impulse response results of the operating speed on 3D geometric characteristics of highway, the specific influence was described, and the reasons of the responses occurred were analyzed. It verified the close correlation between 3D characteristics of highway alignment and operating speed, as well as the drivers′ cognitive law of road conditions.