Multi-Agent-Based Simulation Model And System Of Cloverleaf

Recent years, along with a lot of new research findings in the area of artificial intelligence, the theory of Agent is maturing and widely applicable. In this thesis we introduce this theory into the traffic simulation of cloverleaf and provide a new method in research of the urban traffic simulation. In this system, we use the modeling method of multi-Agent to build the Road Agent, Vehicle Agent, Detector Agent, etc and also provide the intellective decision-making unit and communicate pipe for each Agent. The whole multi-Agent system runs based on the continuous communication among agents.The architecture of the cloverleaf is variable, the beeline, the round curve and the convolute curve is the basic elements for the plane curves,the longitudinal gradient reflects the change of the vertical section. We consider sufficiently these specific factors in building road Agent. The different parameters are set to describe the geometrical characteristic and traffic characteristic. The control models, such as enterin,changing lane and speeding limit, are added in the intellective decision-making unit to guarantee that the vehicle agents could run on the cloverleaf building by the connected road agents.Due to the affect of longitudinal gradient and curvature, the driving strategy of vehicle is quite different from that on the flat road. So in this system we build up the intellective decision-making unit based on the car-following and change lane models considering longitudinal gradient and curvature affection. We use different parameters to describe the different car styles,driver types and driving strategies truly.This system use the Multi-Agent modeling thinking and according to the goal of traffic simulation to building the traffic simulation of cloverleaf and the intellective decision-making warehouse for each Agent in the modeling thinking of multi Agent. An example simulation of a cloverleaf in Tianjin is run at last. The results of simulation are compared to the real detecting data, which can perfectly prove that the conclusion expounded in this paper is effective and accurate in simulate the cloverleaf.