Congestion Pricing Model And Algorithm Based On Anticipated Regret Theory

With the rapid development of economy and the expanding scale of city,the traffic jam is increasingly serious.Traffic congestion pricing has become an effective tool for managing traffic demand.Traffic congestion pricing adjusts travelers*route choice behavior by charging a certain amount of fees for vehicles entering congested areas or roads.The traffic demand of OD pairs is reasonably allocated to each path to reduce traffic load and ease traffic congestion.Based on the anticipated regret theory,this paper establishes two congestion pricing bi-level programming models for the fixed demand situation in the symmetric network and the elastic demand in the asymmetric network.For the congestion pricing model with fixed demand in the symmetric network,the upper-level programming problem is to minimize the overall running time of the traffic network.The lower-level programming problem is the user equilibrium assignment problem based on the anticipated regret theory.The bi-level model is solved by combining Frank-Wolf algorithm and genetic algorithm.For the different charging schemes,the Frank-Wolf algorithm is used to solve the lower-level model to get the corresponding equilibrium flow and then bring it into the upper-level model to get the overall running time of the road network,so as to use the genetic algorithm to obtain the optimal charging scheme.For the congestion pricing model of elastic demand in asymmetric networks,the upper-level model is to maximize the social and economic benefits brought by the traffic travel on the traffic network.The anticipated regret theory is considered in the path selection within the lower-level model,and then a new user equilibrium model is established.which is formulated as a nonlinear complementarity problem.Similar to the solution of the previous model,the power penalty function is combined with genetic algorithm to solve the bi-level programming model.Finally,a numerical example is given to demonstrate the validity of the model and the effectiveness of the algorithm.