交通出行方式离散选择模型的效用随机项结构研究综述

多项Logit模型效用随机项独立同分布的假定为多项Logit模型带来IIA属性,降低了多项Logit模型的行为解释能力。以多项Logit模型的基本假定为基础,以开发能够真实反映不同选择肢效用随机项的异方差性和相关性的效用随机项结构为目标,逐步放松多项Logit模型效用随机项独立同分布的刚性假定,系统综述了不独立同分布、独立不同分布、不独立不同分布的效用随机项结构的拓展。     

基于Logit模型的城市道路交通事件检测仿真

以Logit模型为基础,利用效用函数与概率的概念,建立分时段的城市道路交通事件检测算法。由PARAMICS软件产生模拟交通流数据,将数据输入LIMDEP软件并标定效用函数的系数,同时还输出最大概率预测表。仿真试验结果表明:(1)基于Logit模型的检测算法不仅能够用于城市道路的事件检测,还可判断事件发生所在的车道。(2)在路段长度、车道数、流量相等的模拟条件下,交叉口信号超过仿真所设定的1 min时段长度时,检测效果降低。若将模型时段长度由1 min提高至超过最大信号周期,即可解决检测效果降低的问题。     

The effects of lighting on driver's injury severity at highway‐rail grade crossings

This study estimates the safety effect of illumination on accidents at highway‐rail grade crossings in the United States, using data from exhaustive data from Federal Railroad Administration database covering the period 2002–2011. Using mixed logit modeling approach, the study explores the determinants of driver injury severity at unlighted highway‐rail grade crossings compared with lighted highway‐rail grade crossings in the United States. Several key issues are explored including availability of relevant highway‐rail grade crossing accident inventory data; relevant data element structures; specification and estimation of models to estimate driver's injury severity with lighting and without lighting; and techniques to interpret model parameters. Overall, highway‐rail grade crossing lighting improves safety by reducing the probability of high‐level injury severity through improvements in driver's visibility compared with unlighted intersections. Copyright © 2015 John Wiley & Sons, Ltd.     

Assessing carbon reduction effects toward the mode shift of green transportation system

Plans are underway to introduce green transportation systems at Taiwan's famous scenic spot, Sun Moon Lake, to reduce carbon emissions. Carbon reduction effects after changing the current lake tour modes are assessed using a nested logit model with a stated preference survey, which investigated tourists' choice preferences. The empirical results reveal that the modes of electric bus, electric vehicle sharing, and bike are clustered in the same competitive group. Travel cost, in‐vehicle travel time, and out‐vehicle travel time are found to be statistically significant with the expected negative effects. In particular, the attribute of carbon reduction is only significant for green tourists, who are willing to pay US$ 3.5/kg of carbon saved. The result indicates that more efficient carbon reduction is possible by restricting the usage of gasoline vehicles rather than improving the service levels of low‐carbon modes. Notably, the effect of parking charges on emission reduction is equal to the amount of carbon dioxide absorbed by 4511 trees per day. Copyright © 2016 John Wiley & Sons, Ltd.