Dynamic Sub-Control-Area Division For Regional Urban Traffic Signal Control

The sub-control-area division for regional urban traffic signal control is the basement of performing signal control on a large-scale urban roads network. Because of the complexity of the traffic flow and its'time-varying characteristics in road network, the fixed sub-areas greatly reduce the real-time strain capacity of signal control system. Therefore, a dynamic sub-area division method gradually becomes the focus of research. Fousing on the deficiencies of the present selection of correlation index and sub-area division method, this paper proposes a method based on the self-organizing mapping neural network algorithm combining with dynamic index selection and the real-time identification of key intersection to realize the dynamic sub-area division for the control areas.In terms of the analysis of traffic flow operation characteristics in urban roads network, this paper analyzed the characteristics of traffic flow based on the traffic flow data of intersections and sections in road network.The traffic flow which is spread in time and space in road network is the carrier of traffic demand transmission, the analysis of traffic demand's transfer characteristics on the space in this paper reveals that there exists correlation characteristic between areas. In addition, this paper introduced the principle of macroscopic fundamental diagram which can be used to study the macroscopic transportation characteristics of road network. The study on the characteristic of traffic flow in the urban road network can lay the foundation for the dynamic sub-area division.In the research of dynamic sub-control-area partition method, this thesis mainly studied on the relevant influence factors between intersections and calculation method of dynamic area division for the regional signal control. Through out the research on the basic problems of the dynamic sub-control-area division which includes the prediction of key intersection in road network, simulation analysis of the factors which influences the intersection's associated characteristics and the self-organizing map neural network algorithm, this paper puts forward an algorithm which is based on the self-organizing mapping neural network for the dynamic sub-area division.This paper selected the kunshan city center as the research object to evaluate the effect of dynamic sub-control-area partition method which was based on the self-organizing mapping neural network algorithm and adopted the method in this paper and the method which was based on the "merger index" to calculate the division of the areas respectively.At last,this paper adopted two methods which were based on the traffic network space distribution and the performance of regional signal control to evaluate the two sub-area division methods.The Results shows that the method presented in this paper can well divide the areas to sub-areas with high similarities and can achieve well control performance if this sub-areas are used to regional signal coordination control.