基于视觉信息的高速公路运行车速预测模型

Visual Information-Based Freeway Operating Speed Prediction Model

根据人的视觉特性,提出了一个全新的概念"空间视距",用于量化驾驶员的视觉信息,通过空间视距的概念,可以将某一时刻驾驶员在高速公路某一个位置处的视觉信息抽象为一个S值.文中给出的空间视距理论算法同时考虑了视线的横向偏角限制、纵向偏角限制和地形的起伏,能够计算出在路线任意位置处的S值.结合试验,采用回归方法得到了期望车速与空间视距的非线性关系,并通过美国双车道公路微观仿真模型(TWOPAS)进行逐点纵坡修正,最终得到高速公路运行车速的预测模型.试验验证的结果表明,本文的车速预测模型与美国InteractiveHighway Safety Design Model(IHSDM)具有相近的预测结果,并且在自由流状态下,以空间视距S为基础的运行车速预测模型与实际情况有较好的相关性.

Based on the features of vision, this paper proposes a novel concept - spatial sight-distance （SCSD） to quantify the visual information of the driver. An algorithm is presented to compute the SCSD value abstracted from the visual information of the driver at any location of the road at anytime. Three factors including the horizontal deviation angle limits, vertical deviation angle limits and terrain undulation are taken into the account of the algorithm. Combining with the experiment＇s data, the nonlinear relation of the expected speed and SCSD is obtained through regression procedure, and finally the freeway operating speed prediction model is established by the point-by-point grade calibration of that nonlinear relation through TWOPAS model. The experiment results show that the speed predicted by the model is highly close to that predicted by Interactive Highway Safety Design Model （ IHSDM） of American, and the predicted operating speed also matches the reality fairly well in the case of free flow.