Study On Urban Traffic Signal Timing Optimization Control Schemes

Recently, with the rapid urbanization and ever-increasing vehicles, the traffic signal timing scheme long used for years has to give room to a more modern timing scheme which is more effective, flexible and intelligent for smooth traffic flow at the point of intersections than ever because it is too complicated to fit for the often-changing traffic conditions as for the fixed timing signal control. The study in this field has become one of the most important parts with regard to ITS (Intelligent Transportation System). Aiming at one certain isolated intersection in Shenyang, the corresponding traffic signal control model has been brought forward and the strategy of two fuzzy controllers in tune has been push forward too. These two controllers are responsible for the phase sequence and green light time length each phase respectively.Whereas fuzzy control has been playing a bigger and bigger role in traffic signal timing, many parameters, like the shape of membership functions and their distribution upon the whole universe of the discourse, plus the determination of universe width and fuzzy rules set, are always set up bluntly ahead of time. Worse yet, once done, all parameters will remain unchanged in the whole course. Hence, the setup and optimization of fuzzy controller parameters seem more and more urgent and necessary.At the beginning, the causes, means and objects with regard to parameter optimization of fuzzy controllers have all been studied and clarified. Then, based on historical traffic data, by means of ANFIS, fuzzy neural network toolkit, provided by Matlab software package, a particular traffic model has been set up. The comparison of membership functions between before optimization and after optimization has been made clear in terms of their shapes and distribution. Next, the GASA hybrid algorithm between genetic algorithm (GA) and simulated annealing (SA), has been set forth, which is able to take their advantages both. In the end, by virtue of variable universe principle and algorithm, evolved from the prototype of scalingfactors achieved by function models, the double-layer fuzzy inference structure with scaling factors achieved by fuzzy inference has been adopted here. The optimizations upon fuzzy controller parameters by three above-mentioned algorithms have been conducted and simulation results have shown that average delay time each vehicle would be reduced to some extent after optimization. Therefore, the smooth pass through at the traffic junction would be secured.