Relevant Optimization Models And Methods For Urban Road Network Design Problem

Based on analysis and review of the literatures, this thesis designs two solution algorithms for classical urban road network design problems (NDP), and then studies continuous network design and intersection signal timing problem under traffic emission reduction objective, variable-lane road network design problem, and two extensions of traditional NDP respectively. Optimization models for these problems are established by bilevel programming approach and demonstrated with numerical experiments. The main works of this thesis are as follows:1. Three solution algorithms for classical transportation continuous/discrete/ mixed network design problems are designed based on particle swarm optimization approach, in which the continuous NDP is studied in detail, discrete and mixed NDP are only described the frameworks of the solution algorithms.2. Generally, road capacity constraints must be added to the lower level problem of classical discrete NDP's bilevel programming model in order to avoid traffic assignment on candidate roads. By handling the travel cost function, this thesis formulates the transportation mixed NDP as a mixed-integer, nonlinear bilevel programming model, in which the lower level problem is a standard user equilibrium assignment problem. Then, the bilevel programming model for mixed NDP is transformed into an equivalent single-level nonlinear programming problem by the continuation method and the optimal-value function tool. And then a locally convergent algorithm is proposed by applying penalty function method.3. So-called 'improvements' to the congested transportation network may cause increases in the total emissions. This thesis proposes optimization method to avoid the emission paradoxical phenomena. Then, the thesis optimizes continuous NDP and signal timing simultaneously to minimize the total emissions with bilevel programming approach.4. In order to reduce transportation congestion and make full use of the existing road resources, this thesis establishes a variable-lane road network design bilevel model under tide-like traffic flows, in which transportation management section at the upper level reallocates the lanes of two-way roads to minimize the total system cost, while transportation users at the lower level follow stochastic user equilibrium assignment principle to select travel routes. 5. The thesis proposes two extensions of traditional NDP from the perspective of road network expansion and evolution. The first extension is to design the supporting road network for new established traffic nodes, and the second extension also consider new nodes and links to be added to the original road network, but different with the first extension, the numbers and locations of these to-be-added nodes are to be decided. The second extension is to determine the optimal location pattern so that new road network may reach the best system performance.