A Microscopic Traffic Simulation System Based On MAS Theory

Traffic problem is an important problem all over the world, and the traffic simulation technology has become an important means to research and resolve traffic problem. Different from the traffic status aboard, the traffic flow in our country has many unique characters. And it cannot be simulated truly by most of traffic simulation software. In term of the latest research achievements in traffic and transportation engineering, this thesis develops an Urban Microscopic Traffic Simulation System to solve the traffic problem which is based on computer simulation technology and Multi-Agent System (MAS) theories.Firstly, according to the MAS theories and the feature of traffic system, the thesis uses the MAS technologies to model Urban Microcosmic Traffic Simulation System. In this system, road, crossing, vehicle, traffic light and traffic control center are regardes as intelligent agent, the thesis discusses the main architecture and property of the five simulation object. Considering the advantage of fuzzy control, the fuzzy logic is introduced into the decision of vehicle agent, which makes decision according to the external environment and its internal states during free driving, following ,overtaking and anti-collision passing crossing.Secondly, the thesis puts forward Fuzzy Self-Adaptive Control Strategy of Traffic Light Agent according to the deficiencies of present intelligent traffic signal control system. Through imitating the ways of traffic police managing the traffic, traffic light agent has the ability of intelligent phase switching and controlling in the light of the real traffic flow of the present crossing and the next crossing. This improves the intelligence of the traffic light agent layer, and it is more flexible for the control system to suit the variable traffic condition .Thirdly, the thesis explains the feasibility of using object-oriented method to realize the agent, and which is adopted in this research, furthermore, An Urban Microcosmic Traffic Simulation System is developed using Java. This system can draw road networks by GIS which is being simulated cover on the same part of the city according to the customer's demands.This system can implement the simulation of vehicle-driving. The system can offer all kinds of traffic information. The system also can implement all kinds of traffic control experiment by supplying programme interface. By doing so, this system can be more flexible, extendable, and ultimately reusable.Finally, the thesis implements traffic control experiment between the new traffic signal control strategy and the general fuzzy control strategy. The simulation result shows: the average vehicle speed can increase approximately 5%～10% compared with the general fuzzy control strategy and the timing control stategy.