چكيده به لاتين
Abstract:
Nowadays, the development of cities and increasing population are caused congestion and displacement problems. That’s why it is important to find a reasonable solution for efficient use of existing infrastructures. Traffic assignment models make it possible for Analysts to predict future situation of transportation network. According to dynamic feature of traffic flow, static traffic assignment isn’t suitable for traffic researches, and applying dynamic traffic model is needed. On the other hand traffic assignment must be done according to control data for more Actual results. in other word, signalized intersection effect on dynamic traffic assignment should be considered. In this thesis, first of all the previous studies on dynamic traffic assignment and traffic control reviewed and a dynamic traffic assignment model is presented with respect to traffic control in Tehran. The main purpose of this research is to reduce travel time and convergence time. Because implementing real time traffic data must be considered. The presented model in this thesis is implemented on a real network in Tehran, and the real O-D demand data is used. Since genetic algorithm isn’t suitable for real-time implementation, PSO and SA algorithm is used instead and the resultant travel times are compared together. Furthermore the research results is compared with existing situation of the network in Tehran. The resultant travel time with PSO and SA algorithm is reduced 11% and 14% respectively. According to this study, SA algorithm is better than PSO algorithm and the convergence time is halved. The resultant travel time is reduce too. Since the Extent of network affects on convergence time of model, this model is applied on a sub-network again, with both algorithm. The convergence time using 15 minutes volume data is reduce significantly.
Keywords: dynamic assignment, traffic control, PSO algorithm, SA algorithm