Loading...
- Type of Document: M.Sc. Thesis
- Language: Farsi
- Document No: 39203 (09)
- University: Sharif University of Technology
- Department: Civil Engineering
- Advisor(s): Zokaie Aashtiani, Hedayat
- Abstract:
- In the transportation litreture, many different algorithms have been proposed for solving traffic assignment problem based on the user-equilibrium principal.These algorithms are devided into two main groups: Path-based algorithms and arc-based algorithms. In the past, arc-based algorithms has attracted quite attention, but recent scientific progress in the computer science changed this viewpoint to the state of path-based algorithms. Much of the attention has been focused on three particular algorithms: the complementarity (C) algorithm, the gradient projection (GP) algorithm and the disaggregate simplicial decomposition (DSD) algorithm. In this thesis, these three algorithms have been implemented in the same programming environment, and performance of the algorithms has been compared using two small networks, some randomly generated networks and one realistic network. Other research and this research results show that GP algorithm is faster than DSD algorithm. Also this research results show that GP algorithm is a little faster than C algorithm, but as the size of networks grows this difference decreases. For realistic networks, performance of these two algorithms is almost same. Sensitivity analysis is performed for C and GP algorithms to examine their performance with respect to network sizes, congestion levels and accuracy levels
- Keywords:
- Evolutionary Algorithm ; Gradient Projection ; Path-Based Algorithm ; User Equilibrium Problem ; Disaggregate Simplicial Decomposition (DSD)
- محتواي پايان نامه
- view