Research On The Run State Evaluation Of Urban Transportation System And Control Strategies

Currently, the traffic congestion has become a widespread international problem, congestion alleviation and prevention has become an urgent task for urban development in our country. However, human has always been the lack of adequately, accurately and comprehensive estimate and understood for transportation system run states of urban, which brings about blindness when people make prevention and alleviation measures. Meanwhile the urban traffic congestion is the outstanding contradiction between traffic demand and traffic supply. Therefore, it is necessary to carry out the study work of urban transportation system state evaluation which is the performance of the interaction between supply and demand (two and three and four chapters in this dissertation), and based on this from different angles to explore the characteristics of traffic congestion and the propagation in depth (five chapter in this dissertation), furthermore the traffic congestion control and prevention strategies can be proposed (six chapter in this dissertation). The main contents of the dissertation are summarized as follows:1. Research on dynamic OD estimation models considering deviationand capacity constraints The important input data for urban transportation system run states evaluation is time-varying traffic demand. A Generalized Least Square (GLS) dynamic OD estimation model based on floating car data and considering capacity constraints is proposed, at the same time the Kalman filter dynamic OD estimation model considering deviation and the algorithm are established in this dissertation. The estimated or predicted dynamic OD is loaded on the traffic network and then the assessment or prediction of the transportation system run states can be achieved which can be used to pre-judge the traffic congestion, so the traffic guidance information can be published to drivers ahead of traffic congestion or to be develop the appropriate control strategies, and play a proactive role in traffic congestion, attempts to prevent it in the first place.2. Real-time path searching algorithm for large traffic-networkEach vehicle must follow a specific path from its current location to its destination in the simulator. So how to make the traffic simulator run efficiently online and reduce the huge computation and improve the system performance in real time, the path generation and storage is a challenging problem. Here, a new idea of effective path generation and storage and an effective path generation algorithm of large traffic network are proposed. Additionally, a new data structure that can store large count of paths and support high performance searching is given. By using the recursive property, the new algorithm can reduce the size of problem which can derive the paths with a small space required. As a result, it will improve the compute speed and save the computer storage resources greatly.3. Research on system run states evaluation based on improved traffic flow modelThe method of combining the road traffic flow and network traffic flow is applied, at the same time the analytical integrated with simulation tools are applicated to explore the urban transportation system run states evaluation in this dissertation. The basic theory of urban transportation system run states evaluation which is based on the interaction of the demand simulation and supply simulation is described in detail. The traffic flow as the role content is improved and the detail simulation process for urban transportation system run states evaluation is designed, and then proposed the method which dealed with traffic flow model's anisotropy, so the basic platform is constructed for system simulation and provides a theoretical method for more accurately evaluation to urban traffic system run states.4. Research on the complex properties of traffic congestion based onnetwork structures To be deeply understand the complex mechanism and characteristics of urban traffic congestion, the propagation of traffic congestion and its complexity features are studied and the dynamic traffic characteristics of the network is analysised based on the complexity of the theory in this dissertation. Simulation results show that the large intermediaries edge in complex network prone to traffic jams. At the same time, the traffic congestion and its propagation characteristics have a greater influence on the network structure and the regular network has a great effect to ease traffic congestion. For regular network especially bi-direction grid network, the propagation and dissipation of traffic congestion with accidents is studied and the traffic congestion states is evaluated with the averaged flow, velocity, density of each section of the network.5. Research on the control and prevention strategies of traffic congestion For the incidented traffic jams on the bi-direction grid network the tree control feedback strategy is proposed and the prediction information induced strategy as well as regional signal optimal control strategy which can ease the regular traffic congestion are developed in this dissertation. Among them, the signal optimal control strategy use the idea of bi-level programming, for the general traffic network, The regional network signal optimization model and the SPSA algorithm are proposed with the mean delay, length of queue and the amount of traffic flow in the intersection as the optimization goals. Simulation results show that control strategies can alleviate traffic congestion in some extent and the overall network health has been some improvement.