Study On Stochastic User Equilibrium Model Based On Ant Colony Optimization

Stochastic User Equilibrium(SUE)model proceeds from perspective of travelers,does traffic assignment for travelers according to travelers' impedance degree of sensibility to path,making the distribution result closer to the distribution of car flow in the real world.The essence of SUE is traffic assignment problem whose key link is routing selection.In order to describe travelers' route-choosing behavior more accurately,this paper studies the SUE model based on three kinds of discrete route choice methods: Logit,Multinomial Weibit(MNW),and Marginal Exponential Model(MEM).Because route choice is affected by the uncertainty of traffic,the traditional Method of Successive Averages(MSA)has a low equilibrium in solving such problems.Ant Colony Optimization(ACO)algorithm is a kind of natural calculation method which has been paid more and more attention in recent years.Especially when it solves traveling salesman problem,the properties of node selection influenced by pheromone and route choice problem of this study has intrinsic similarity.Therefore,this paper proposes three kinds of SUE model based on ACO algorithm,and compare the results with MSA algorithm.The specific research contents and achievements are as follows:1.In the Nguyen&Dupuis network model,Logit-SUE model is solved with MSA algorithm and ACO algorithm.Each link flow,link impedance,and path impedance which distributed by these two algorithms are compared.Traffic assignment performances of these two algorithms are analyzed with sensitivity under different ? conditions.2.An ACO algorithm pheromone update strategy which bases on MNW discrete route choice model is raised.MNW-SUE model is solved with MSA algorithm and new ACO algorithm.Each link flow,link impedance,and path impedance which calculated by these two algorithms are analyzed.Traffic assignment performances of these two algorithms are analyzed with sensitivity under different ? conditions.3.In the Sioux Falls network model,according to the similitude of path selection probability formula between Logit model and MEM model,an ACO algorithm pheromone update strategy which bases on MEM discrete route choice model is raised.New ACO algorithm and MSA algorithm with a bisection-search method are respectively used to solve MEM-SUE model.Each link flow,link impedance,path impedance and degree of sensibility of these two algorithms under different ? conditions are acquired by comparing and analyzing.This paper detailedly introduces implementation steps and detailed processes on solving these three models with MSA algorithm and ACO algorithm.The simulation results show that when travelers' perception error for path impedance is larger,ACO traffic assignment algorithm will excel MSA algorithm.It has a stronger flexibility and is more conducive to reducing traffic congestion and user travel.