科研之友
微信
新浪微博
Facebook
分享链接
摘要
This paper examines commuters' departure time and route choices in the morning commute problem when a true distribution of travel time is unknown but belongs to a bounded distributional uncertainty set. The travel preferences towards risk and ambiguity are distinguished by adopting the criterion of ambiguity-aware Constant Absolute Risk Aversion (CARA) travel time. We first examine the dynamic user equilibrium for a single route model with a homogeneous preference towards risk and ambiguity. Compared with risk-neutral commuters, we find that departure time window is shifted earlier for the risk averse commuters and shifted later for the risk-seeking commuters. We also study the single bottleneck with a risk-averse class and a risk-seeking class. We show that with a larger gap between the two classes' preferences, the congestion pattern will change from one peak to two peaks. It implies that preference heterogeneity may stagger the departure time choices and thereby relieve the average congestion. Last, we examine a two-route problem with homogeneous preference. Commuters choose between a fast and risky route (highway) and a slow and safe route (local arterial). We prove the monotonicity of the traffic flow distribution between the two routes with respect to the maximum variation in travel time. Furthermore, we find that reducing the uncertainty on the highway by providing information will reduce the total system cost and the total expected congestion simultaneously for risk-averse commuters. However, it will reduce the total expected congestion but increase the total system cost for risk-seeking commuters. In the numerical section, the price of anarchy is analyzed by varying the risk preference and the ambiguity preference.
- 出版日期2018-11
- 单位西南交通大学
