An Analysis Of Stochastic Assignment Paradox Considering Congestion Effect

The traffic paradox plays an important role in the traffic planning and design.This paper analyzes how congestion and stochastic perceived error in route choice behavior simultaneously affect traffic paradox using the traditional logit stochastic user equilibrium model and the traditional weibit stochastic user equilibrium model.Traffic paradox will be discussed in three traffic networks which have three different structures: a simple two-link network,the Braess Network,multiple origindestination with multipath network.Some general conclusions can be obtained by analyzing the paradox features of three different network.First of all,this paper analyzes the similarities and differences of traffic paradox corresponding to the total/average travel cost between different traffic network in the case of marginal improvement and adding a link.The influence of congestion and stochastic perceived error on the stochastic traffic paradox is analyzed,and compare them with those in the case without congestion.Secondly,how the stochastic perceived error affect the average travel cost in a fixed network and demand level is analyzed in this paper.The information paradox in which higher stochastic perceived error lead to lower average travel cost is discussed in three different network.The traffic demand interval that information paradox happens will be given.Whether stochastic user equilibrium model can be equivalent to SO model absolutely by adjusting the stochastic perceived error is researched in this paper.Finally,this paper studies the stochastic traffic paradox corresponding to the total/average perceived travel cost based the on multinomial logit model in three kinds of transportation networks.The influence of congestion effect and stochastic perceived error on the boundary conditions of paradox happen is analyzed in the case of marginal improvement and adding a link,and compare them with those corresponding to the total/average travel cost.Based on the application of multinomial logit model and multinomial weibit model to discuss the characteristics of stochastic traffic paradox,this paper finds that the stochastic assignment paradox corresponding to the total/average travel cost for marginal improving/adding a link can be affected by congestion effect significantly.Besides,the condition of information paradox happen has been given.Furthermore,clarifying that the SO solution can be achieved by adjusting the stochastic perceived error in the stochastic user equilibrium model.At last,this paper find that the paradox corresponding to the total/average perceived travel cost which will never happen in an uncongested network can happen in a congested network.The results obtained in this paper can help to carry out scientific and reasonable traffic management and control,and improve the traffic efficiency.