The traditional deterministic process of trip assignment does not account for uncertainties in traffic demands. These point-estimate based solutions often results in large differences between forecasted and actual traffic volumes thereby imposing huge financial burdens upon development agencies. In this work, stochastic treatment has been given to the trip assignment problem, specifically the network user equilibrium problem solved using the variational inequality method, under demand uncertainties modeled as random inputs. Smolyak sparse grid interpolation technique was successfully applied to the problem and compared to Monte Carlo sampling. Performance of constructed interpolant was evaluated through output distribution recovery , statistical moment estimation, and computation time comparisons. Ability of sparse grid to efficiently handle demand uncertainties using as many as 5 times fewer points than Monte Carlo sampling in pragmatically sized transportation networks was demonstrated.
【 预 览 】
附件列表
Files
Size
Format
View
Stochastic numerical approximation approaches for estimation of traffic volume under travel demand uncertainties