Study On Intelligence City Area Traffic Control

Due to the uncertainty, random and nonlinear of the traffic flow, it is hard to construct the precise mathematical model. Various intelligent methods provide new ideas and methods to solve the problem of constructing model for traffic, which have a good effect on the traffic control and counseling the traffic flow reasonably. This paper will use the intelligent methods to control the single intersection and regional traffic control. Totally, the main work of this paper includes the following three aspects:1. For the single intersection, take the algorithm realized by the two stage fuzzy neural network controller which combines the advantages of fuzzy control and neural network, and adjusts the phase sequence according to current traffic flow. According to the intersection situation determines the input parameters and judge the next step is green time module or the phase selection module. Each controller is realized by the fuzzy neural network. In the end, this two stage fuzzy neural network can reduce the average vehicle delay by the simulation on TSIS and MATLAB.2. For the road network area, the parameters need to optimize are green ratio, offset and cycle. Mainly use the Q learning algorithm. Add fuzzy control to Q learning algorithm for the deficiency of Q learning algorithm. First of all, judge whether the current state is concluded in the experiences, it is directly output the optimized parameters by the fuzzy neural network or into the Q learning algorithm and store the Q value by the BP neural network to improve the accuracy of learning rate and control algorithm. In the end, simulate on the TSIS and MATLAB, prove the effectiveness of the proposed algorithm.3. As the most popular mathematical software, MATLAB include many library functions, simplifies the complexity of programming. TSIS is the most popular traffic simulation software, but the development of language is simple, this paper make fully use of the advantage of the two software, validate the traffic control algorithms easily and effectively, resize the two times of TSIS development.