Mesoscopic Traffic Flow Model Based On Cell Transmission Mechanism

Traffic flow model system in transportation science consists of microscopic,mesoscopic and macroscopic traffic flow models. The thesis focuses on mesoscopictraffic flow models, which is mainly related to the evolution of velocity distribution.The thesis adopts the cell trasnmission mechanism as a modeling technique, whichdiscretizes and analyzes traffic and also utilizes the maximum information principleand active particles theory as iccroporating modeling methods. The proposed modelscan be used to simulate traffic flows more quickly and accurately. Therefore, the topicis of theoretical and practical significance for the development of mesoscopic trafficflow model.The main achievements of this thesis are as follows:1. In the first time applying maximum information principle in synergetics topresent mesoscopic traffic flow models.The proposed model can use the macroscopic speed to derive speed distribution,which is an attempt of deriving mesoscopic state from macroscopic state. Theproposed model provides a method of studying the relation between information andtraffic flow state. The example given in the third chapter shows this relation. If thedenstiy is larger than critical density, the maximum information of speed distributionincreases with the density. If the density is less than critical density, the maximuminformation remains almost constant.2. Integrating CTM and Delitala-Tosin model to propose a mesoscopic trafficflow model under spatial heterogeneous traffic flow condition.The thesis first derives the general formulas for spatial homogeneous traffic flowconditions, and then examines how the Delitala-Tosin model controls speeddistribution evolution. For analyzing the effects of local density and distance oninteraction, the thesis introduces the inverse square to improve the Delitala-Tosinmodel. Then, the thesis integrates the CTM and Delitala-Tosin model under spatial homogeneous traffic flow condition to propose a new mesoscopic traffic flow model.In this model, the partial difference equation transforms to ordinary differenceequations, so that we can use more stable and effective methods method to solvethem.3. Proposing mesoscooic traffic flow model considering desired speedThe thesis introduces desired speed to formulate a new transition probability andpresents a mesoscopic traffic flow model. The thesis uses the hybrid programmingtechnique of VBA and Matlab to compile a program for this model. The results helpus to understand traffic flow in the following aspect. People often regard the denstiyas a determining factor for traffic flow average speed. The results show that thedrivers′desired speed is also a deterministic factor of average speed.4. Presenting weighted interaction function considering potential influence.Almost all mesoscopic traffic flow models focus on highway traffic flow. Thereare little studies regarding city traffic flow. In city traffic, there are more factorsdisturbing traffic flow operations, for example signal control, pedestrian crossing, busstations, etc.. Thus, we should consider the stimulus from actual influence region andpotential influence region. The thesis proposes a weighted interaction functionconsidering potential influence for better describing city traffic operation.5. Proposing an internal state model for signal timing optimization.The existing signal timing optimization models scarsely use of traffic flow model,which often adopt statistical methods to estimate various factors. The thesis firstextends internal state model from non-signalized control intersection to signalizedcontrol interseciton. Then the thesis applies cell transmission mechanism to formulatesignal timing model, which consider traffic flow dynamics. Finally, the thesiscompares the proposed model with Webster model and Gazis-Potts model.