考慮異質使用者與旅行時間感知誤差於動態道路定價下之動態使用者均衡模式

A Dynamic User Equilibrium Model Considering User Heterogeneity and Perception Error for Dynamic Road Pricing

道路收費定價策略的設計與評估須要將道路收費對於異質性用路人之路徑選擇行為的影響列入考量。然而，傳統動態使用者均衡交通量指派模式在估計或預測道路收費下之車流量分布時，假設所有的用路人具有相同的時間價值與旅行時間感知誤差。為了改善這個問題，本研究發展一個以普羅比模式為基礎的異質性動態隨機使用者均衡模式，在模式中假設用路人的旅行時間感知誤差以及其時間價值服從已知的多變量常態機率分配，且每位用路人在交通路網中會選擇一條對於自身而言在旅行時間與成本皆最佳的路徑。此模式能夠貼切地描述在動態道路收費情境下，用路人反應不同收費水準的路徑選擇行為，並且可以作為評估不同動態道路收費策略的工具。本研究將此問題定式為一個無限維度固定點模式，首先利用參數分析方法將問題近似為有限維度的多類別動態隨機使用者均衡指派問題，接著發展以變數產生法為基礎的模擬式啟發式解法來求解此問題。數值實驗結果發現，本研究中所發展的演算法能夠有效地找到近似最佳解，而且也證實了時間價值與旅行時間感知誤差的假設對於求解結果有明顯的影響。

Design and evaluation of road pricing strategies require modeling the response of users with non-identical preferences and differing perceptions. To this end, this paper presents a probit-based heterogeneous dynamic stochastic user equilibrium (HDSUE) model, which captures path choice behavior of heterogeneous users with distinct value of time (VOT) and different perceptions of travel time. Across the population of travelers, VOT is assumed to be continuously distributed, and the perception errors of travel times in a choice set are multivariate normally distributed. The problem is formulated as an infinite-dimensional fixed point problem. To solve the HDSUE problem, a parametric analysis method is employed to facilitate defining a multi-class dynamic stochastic user equilibrium problem, which transforms the fixed point problem from infinite-dimensional to finite-dimensional spaces. A simulation-based and column generation-based heuristic is developed to solve the problem. A set of numerical experiments are conducted to examine the solution quality and convergence pattern of the algorithm, and explore the impacts of user heterogeneity in VOT and perception error in travel time on users' path choice response to dynamic road tolls.