Using genetic algorithms and an indifference-zone ranking and selection procedure under common random numbers for simulation optimisation

Abstract

Genetic algorithms (GAs) are one of the many optimisation methodologies that have been used in conjunction with simulation modelling. The most critical step with a GA is the assignment of the selective probabilities to the alternatives. Selective probabilities are assigned based on the alternatives’ estimated performances which are obtained using simulation. An accurate estimate should be obtained to reduce the number of cases in which the search is oriented towards the wrong direction. Furthermores, it is important to obtain this estimate without many replications. This study proposes a simulation optimisation methodology that combines the GA and an indifference-zone (IZ) ranking and selection procedure under common random numbers (CRN). By using an IZ procedure, a statistical guarantee can be made about the direction in which the search should progress as well as a statistical guarantee about the results from the search. Furthermore, using CRN significantly reduces the required number of replications.