Robustness-based Models Of Traffic Congestion Management Strategies In Stochastic Networks

Traffic congestion has become one of the most noticeable problems worldwide. Different kinds of strategies are proposed to solve the problem but fall apart from expectations in stochastic networks. Due to this, the paper raises the notion of robustness about traffic congestion management strategies in stochastic networks.This paper first discusses the equivalent of congestion pricing and tradable credit scheme in deterministic networks which are different in stochastic networks at the same time. For tradable credit charging scheme, it is assumed the network uncertainty derives from a set of uncertain factors under which travelers choose the minimal cost path. Correspondingly, there is a set of solution to the traffic assignment with different uncertain factors. The smoothing projected gradient algorithm will be used to will find a robust solution with the expected residual minimization (ERM) method. For cordon pricing, this paper aims to find a robustness-based cordon pricing scheme, which would control the cordon-in traffic demand to be less than a threshold with a predetermined probability which is noted as reliability. Two implementation scenarios are considered. For the scenario without complete information, travelers are assumed to choose paths with the minimum expected travel time. Solution algorithms and numerical examples are conducted to demonstrate both models. Sensitivity analysis found that the higher the reliability of the congestion pricing higher rates, people more intense contradiction. In addition, congestion pricing increased much faster at higher reliability. So that this paper raises the growth rate of congestion pricing rates against reliability as the evaluation standard congestion pricing to balance cordon pricing rates and reliability.