Extending Bi-level Programming Model on Origin-destination Estimation

The bilevel programming model can be used to estimate origin-destination.On the basis of present achievement on the bilevel programming,the new bilevel programming is extended.The upper optimization model,which combines Bayes and muti-variable normal model,can eliminate the random errors and time effects on link accounts.The lower optimization model,following Stochastic User Equilibrium,the route choice behavior is considered.The bilevel programming model is suitable to such OD estimation problem that the traffic is mixed and the demand is elastic.At the same time,this programming is suitable to not only congested road network,but uncongested road network.Finally the extended bilevel programming is used to small-scale network,good outcome can be obtained.