Signal Control Of Oversaturated Intersection Group Based On The Multi-Agent

Intersection signal control is an important way to solve the traffic congestion. The traditional traffic control method that based on the accurate mathematical model can’t achieve the expected control effect in the practical application, and the computation is complexity. Multi-Agent technology as the hotspot research in the artificial intelligence field. Agent can control object independently, and they can coordinate control through the communication with each other, making the regional transportation system achieve greater intelligence. The thesis establishes the model of regional traffic control based on Multi-Agent, and introduced the adaptive algorithm into the intersection group agent, also using the fuzzy control theory to divise the traffic zone that considered five kinds of the impact factors at the same time. Finally it researches the signal timing of oversaturated intersection.The thesis proposes a distributed regional traffic control model based on Multi-Agent, this model includes subarea agent、area control center agent and master control center agent. This agent can coordination at all levels, it can also coordinating at the next higher levels. the paper brings the adaptive genetic algorithm into the intersection subarea agent, according to the sensing module perceived traffic information, and choosing the optimal way to control intersection through the response module and the knowledge module that storing large amount of difference traffic conditions.The simulation experiment shows that the combination of adaptive genetic algorithm and intersection subarea agent has better performance in intersection signal control and proves the feasibility of subarea agent substituting intersection agent.For the problem of Intersection group division about regional traffic in urban traffic network. The paper uses fuzzy control theory to solve the problem, and defines adjacent intersections coordination coefficient to indicate the degree of association of adjacent intersections, the coefficient considers five factors, they are the intersection distance, signal cycle, traffic flow, traffic flow impact factor and the queue length. This paper elaborated the influence degree of each influence, factor and calculating method.Finally, this thesis selects the traffic network to simulation, and makes intersection average delay time as evaluation index, then uses the adaptive genetic algorithm Intersection group agent to control the intersection group. Ultimately simulation results prove the effectiveness of this division method.An improved NSGA be used in the oversaturated intersection, this algorithmincreased density estimation and rapid on dominated sorting strategy based on the NSGA. The algorithm is verified by Matlab simulation, choosing the largest number of cars that through the intersection and minimum line length as the optimization goal.The experiment results prove the effect that the algorithm is used in the signal timing scheme of oversaturated intersection.