Research On Uncertain Discrete Transportation Network Design And Dynamic Discrete Transportation Network Design

The discrete network design problem deals with the selection of the links to be added to the existing road network, in which the traffic demand and the construction cost are given. The objective of the problem is to make the optimal decision to minimize the total system costs. In this thesis, a bi-level programming model is used to describe this problem. The upper-level model is to minimize the total travel cost in the network under the budget constraint. The lower-level model is a user equilibrium (UE) model to describe the equilibrium distribution of traffic flow. Based on the achievements on the discrete network design problem, a bi-level chance constrain model with uncertain construction cost and a dynamic discrete transportation network design model are proposed. Then some algorithms are designed to solve the models. Some comprehensive computational experiments are conducted to illustrate the correctness and efficiency of the models and the algorithms proposed in this paper.The main contributions made by this study are:(1) The discrete network design problem with uncertain construction costs. Based on the interrelationship of the uncertain construction costs, separate the discrete network design problem with uncertain construction costs into two types. Two stochastic models are built. The algorithms are designed based on genetic algorithm theory and branch and bound algorithm theory respectively. Some comprehensive computational experiments are conducted to illustrate the correctness and efficiency of the models and the algorithms. A result can be illustrated that the scheme got from uncertain model is more feasible than the scheme got from deterministic model.(2) The dynamic discrete transportation network design problem. The period of the network construction is divided into several intervals. The construction in a stage is based on the construction of the end of the prior stage. A bi-level dynamic time-dependent model is built, and the branch and bound algorithm is designed to solve it. A comprehensive computational experiment is conducted and the result shows that the solution of the dynamic time-dependent simulation model is better than the solution of the static sequential design.