Research And Practice On Simulation Of Dynamic Traffic Flow In Urban Roads

Traffic flow simulation is an important part of the intelligent transport system, which can dynamically and realistically simulates various traffic phenomenon, show the changes of time and space of traffic flow, as well as provide the test platform for traffic planning, traffic organization and management, material transport flow optimization and other aspects. However, there is a big difference between the simulation results and the actual operating conditions, it effects the effectiveness and practicability of traffic flow simulation. This paper aims at promoting the authenticity of traffic flow simulation, studies from the road network modeling, OD (Orignation Destination) prediction, vehicle routing, and proposes and implements corresponding improved method, then discusses modification method of the real data of traffic flow impacting on simulation model.Firstly, this paper discusses and completes the static physical model building of simulation system, mainly including the construction of urban road model, vehicle driving route setting, the construction of the vehicle model, collision detection between the models and so on. And we do research on how to get the city time-varying OD matrix, propose and achieve the forecast of trip distribution time-varying characteristics based on history data and gravity model, it can reduce workload of OD survey and objectively reflect the real distribution.In the aspect of dynamic traffic assignment, this article analyzes the existing road resistance function model, proposes the improved model which is more in line with the actual traffic situation and easily calculation. According to the driver’s travel behavior characteristics in actual situation, we proposes the method based on Multinomial Logit (MNL) model of vehicle dynamic route choice to realize the generation of traffic flow, which reflects the dynamic and random nature of actual traffic flow forming process better. According to the traffic demand and traffic network features, we defines the vehicle departure time model, effective road discrimination and calculation method of the effective path. In terms of simulation model calibration mechanism, according to the measured data of the traffic detector and the state of traffic simulation, kalman filter is used to predict and correct the real-time traffic simulation state.Finally, the paper analyses and realizes the overall structure of the traffic flow simulation prototype system, and in a particular road network, we realize the generation of vehicle, route choice of intersection and motion simulation on the road. The series of vehicle travel activities constitute the macroscopic traffic flow in urban road. From the simulation results we can see that the traffic flow distribution formed by driver’s dynamic route choice conforms to the characteristics of urban road traffic flow, it proves the feasibility and rationality of route choice model.