Advances in Genetic Algorithm Optimization of Traffic Signals

Abstract

Recent advances in the optimization of fixed time traffic signals have demonstrated a move toward the use of genetic algorithm optimization with traffic network performance evaluated via stochastic microscopic simulation models. This paper examines methods for improved optimization. Factors examined included the number of replications of the stochastic traffic simulation performed, the use of common random numbers to reduce variability, modified versions of the genetic algorithm, and alternative genetic operators. Computing resources are found to be best utilized by using a single replication of the traffic simulation model with common random numbers for fitness evaluations. Application of the cross-generational elitist selection, heterogeneous recombination, and cataclysmic mutation search algorithm with real crossover and mutation operators is found to offer improved optimization efficiency over the standard genetic algorithm with binary genetic operators. Combining the improvements, delay reductions between 13 and 30% were obtained on the test networks relative to the standard genetic algorithm approach. A coding scheme allowing for complete optimization of signal phasing is proposed as well as an alternative delay measurement.