A Study of Initial Individual Generation in Search for the Optimal Traffic Signal Offset Using Genetic Algorithm

In this paper, the offset control of traffic signals is considered. An offset which is the target of signal control parameters in this study, is difficult to optimize owing to its variety of combinations. A traffic signal is controlled using three parameters, namely cycle, split and offset, and these parameters should be determined based on traffic conditions. The offset-optimalization is searched using a genetic algorithm (GA), which evaluates the average travel time calculated by the traffic simulator, and repeats generations to leave superiority individuals. The aim is to eliminate the search time, which has been a subject of previous research. When we propose an initial individual generation method for the GA and examined its effects, it is confirmed that it contributed to shortening the processing time.