Author(s)

H. Kazama, H. Mochizuki, S. Takahashi & H. Nakamura

Abstract

In order to minimize average travel time (ATT) through traffic signals, a traffic parameter offset and optimized offset pattern are discussed. For practical use, the search time of the optimized offset pattern should be as short as possible. To this end, a genetic algorithm is introduced. In the conventional method of searching for an optimal solution, a target fitness value is not given, making it extremely difficult to reduce the search time. To overcome this problem, here we introduce a new method that depends on the ATT distribution characteristics and that derives the target fitness value by using the target road parameters. In addition, the effectiveness of the method is illustrated with a case study. Keywords: traffic signal, genetic algorithm, offset, search time, average travel time. 1 Introduction One method of searching for an optimal solution is a genetic algorithm (GA). If a GA is given a target value of the optimal solution, it can be terminated upon reaching that solution; however, if the target value is unknown, a method such as treating the convergence conditions of the solution search as a basis for terminating the algorithm, or treating the best solution obtained during several trials as the \“found solution” can be adopted. In either case, however, if the target value is unknown, it inevitably takes a long time for the simulation compared with the case where the target value is known. We investigated optimization of systematic traffic signal control using a GA. In this problem too, no target value of the optimal solution is given, and the best solution obtained during a large number of trials is treated as the \“optimal

Keywords

traffic signal, genetic algorithm, offset, search time, average travel time