| Dynamic User Optimal (DUO) is a special problem of Dynamic Traffic Assignment (DTA), for the purpose of obtaining the dynamic flow equilibrium patterns in a given transportation network by assuming that each motorist is trying to minimize his/her individual travel cost; while sensitivity analysis aims to establish a connection between changes of model inputs and changes of model outputs. Specifically, the objective of sensitivity analysis of network equilibrium flow problems is to find the gradients of equilibrium flow variables with respect to perturbation parameters in travel time functions and travel demands. By conducting sensitivity analysis for DUO problems, it helps to identify the links which are most sensitive to traffic flow or travel time changes and find the critical links with respect to network flow.; In the thesis, we first propose a DUO route choice model based on DUO conditions, and then introduce perturbation parameters into the link travel time functions and the travel demand vector. The major difficulty in applying standard sensitivity techniques here is that the incidence matrices used to relate different flow variables are themselves functions of the network flows and these functions can not be expressed in a closed form. This difficulty is overcome by formulating an equivalent DUO problem of the original problem under small perturbations and then conducting sensitivity analysis for this equivalent problem. Through this method, the gradients of dynamic equilibrium flows with respect to the perturbation parameters can be obtained, and then used to estimate the equilibrium traffic flows under small perturbations. These results are also illustrated by two numerical examples.; Results of sensitivity analysis can be utilized to quantify the impact of model input errors and identify the most critical input variables; the sensitivity analysis process can also be embedded in bi-level programs to optimize some design variables. Sensitivity analysis based methods, which are applied in the transportation information system for location design, will be introduced in this research. |
