Bounding The Inefficiency Of Multi-class Stochastic Transportation Network Equilibrium

Urban traffic management and control strategies are presented to guide the travelers’ behavior to improve the efficiency of the traffic system.Therefore,bounding or estimating the inefficiency of transportation network is meaningful for analyzing the non-cooperative travel behavior,and significant on the development and implementation of traffic management and control strategies.Based on the traffic network equilibrium theories,with the consideration of the traffic network types and the multi-class travel behaviors,this dissertation addresses the problem of bounding the inefficiency of multi-class stochastic user equilibrium(SUE)both in deterministic and stochastic networks,and discusses the influence factors on the upper bounds.Meanwhile,the validities of the analytical results are verified by numerical examples.The main contents are listed as follows.(1)In the deterministic traffic network under ATIS and congestion pricing,the equivalent variational inequality models of the multi-class SUE are proposed under time-criteria and cost-criteria,respectively.The inefficiency of the multi-class SUE against the system optimum(SO),and the inefficiency against the stochastic system optimum(SSO)are investigated.The upper bounds are analytically derived by variational inequality method,and the parameters influencing the bounds are discussed.(2)In the stochastic traffic network with link capacity degradation,the inefficiency of the single class SUE against the SO,and against the SSO are investigated,respectively.The free flow travel time parameter and its calculation method are introduced to directly reflect the influence of the degradation degree on the travel time,and indirectly reflect the influence of the degradation degree on the upper bound of the efficiency loss.The upper bound of the inefficiency can be obtained based on the free flow travel time.Then,the problem with single class is extended to the multi-class situation under ATIS,and the influences of the link degradation degree,the ATIS market penetration and information conformance rate are analyzed.(3)In the traffic network with stochastic demand,the inefficiency of the single class SUE is investigated.This dissertation pays more attention to the inefficiency when travel demand follow normal distribution and link travel time function is polynomial,and establishes the upper bound of the inefficiency based on the free flow travel time.Then,the problem with single class is extended to the multi-class case under ATIS,and the influences of the stochastic demand,ATIS market penetration and information conformance rate are discussed on the upper bounds through a numerical example.(4)In the stochastic traffic network with link capacity degradation,the inefficiency of the multi-class multi-criteria SUE is investigated under congestion pricing.The analytic expression of the upper bounds of inefficiency is conducted with the polynomial link travel time based on the free flow travel time under time-criteria and cost-criteria,respectively.The influences of the link degradation degree,road pricing on the upper bounds are further discussed.(5)In the traffic network with stochastic road supply,based on the travel time reliability theory,the inefficiency of the single class SUE is investigated.A smaller upper bound is derived than that of existing literatures with the introduction of the free flow travel time parameter.The problem with single class is extended to the multi-class situation with travelers’ different risk preferences.Further discussions are made on the influences of the stochastic road supply,the travelers’ risk attitudes,and system manager’s risk attitude on the upper bounds by a numerical example.