Effects Of Network Structure On The Performance Of Urban Traffic Flow

As we know, traffic congestion has become one of the major bottlenecks, which badly restrict the prosperous, orderly and rapid development in our cities. How to make reasonable traffic planning, control strategies to alleviate urban traffic congestion has become a hot topic. Furthermore, the urban traffic is a classic, exoteric and complex system, and its running rule is also intricate. Many domestic and foreign scholars have made a lot of researches on the generation and propagation of urban traffic congestion from different perspectives, but there is also a lack of sufficient research between network structure and urban traffic.In recent years, with the development of complex network theory, it provides a new view and method for studying the complex system. Complex network theory has been widely used in sociology, biology, physics, economics, computer science, as well as transport and other areas. A great deal of research results in transportation, urban transportation system and so on by using the complex network theory have been obtained. However, many of these studies are based on static traffic equilibrium assignment, without considering the dynamic process of traffic flow on the network. Therefore, in this paper, by introducing the cell transmission model (CTM), which can clearly describes the physical effects of queuing, simulates the shock wave, queue forming, queue dissipation, and the impact among links, we study effects of network structure on the performance of urban traffic flow.In this thesis, we study the relationships between network structure and urban traffic flow based on the CTM and the complex network theory. The main contents are summarized as follows:(1) Using regular network, small-world network, random network, small-world networks with different probability of re-training as underlying network topologies, we investigate the performance characteristic of traffic flow under different demand. The results show that traffic performance efficiency in the random network is higher.(2) By analyzing the performance characteristics, such as the evolution of traffic congestion, the temporal and spatial distribution of congestion bottlenecks, it can be found that the random network can withstand the biggest traffic demand. Under the same level of demand, the random network is not easy to generate congestion bottlenecks. When the bottlenecks appear, the dissipation time is the shortest in random network.(3) Using average journey velocity of links as measure indicators, we analyze the spatial correlation among links under different networks base on Moran's I coefficient. Results show that the similarity of average journey velocity among adjecnt links is the most obvious in small-world network.