Application Of Genetic Algorithm In Traffic Control

With the development of economy and rise in level of urbanization, the traffic problems become more and more serious. How to manage and control the existing traffic effectively is a very important problem in traffic and transportation field. Along with the movement of Intelligent Transportation System, traffic signal control remains one of the most important research fields and development items. Traffic engineering practice indicates that building road properly is needed,at the same time, it is essential ways of solving traffic problems to regulate and control traffic flow through software properly and scientifically, to exert potential of current traffic networks fully,to make traffic flow move in an order, and to develop Intelligent Transportation System.。How to control traffic reasonably is an optimized question.This thesis discusses that how to use advanced science methods to control traffic signal. It is solved by Genetic-Algorithm (GA). Base on factual condition, it is made a primary improvement on algorithm which contains inequality constraints. The GA is combined with simulated annealing algorithm and adaptive fitness to solve this question. When the improved GA is applied on the model of an intersection, the results of the simulation show that the improved algorithm is effective. However, when the improved GA is applied on the model of two intersections, the inequality constraints and equality constraints are contained in it, so the GA is improved again. First, we create a great deal of species, and then make these species approach on feasible domain. If some species are out of the feasible domain, their fitness is reduced. Finally the best specie is found. The result of numerical experimentation shows the algorithm is better than the previous one.Whatever the scheme of the control is improved, the maximum passed capacity is finite. When traffic flow surpasses the maximum load of traffic network, the scheme of the signal control can do nothing. Therefore, for sake of this problem, a macroscopic traffic flows model of control is presented. The model of the net is built for discussing the maximum flow of the net. The new model is solved hardly by the previous improved GA, because not only one or two equality constraints but large amount of equality constraints in the model as the net enlarges. A new improved GA is proposed again. It is combined with the methods of queue, ecological niche and adaptive fitness. Ecological niche is used for creating more Pareto species, as decision-maker have more choices. Adaptive fitness is used for avoiding being premature. Comparing the new GA with the traditional algorithms and the previous improved GA, the simulation results of the new algorithm are better. It provides traffic with an important method for working easily. Computer simulations show that the results of the improved algorithm are effective and better than traditional algorithms.