A Cooperatively Coevolutionary Genetic Algorithm to the Dynamic Facility Layout Problems

Abstract

A cooperative coecolutionary genetic algorithm (CCGA) to the dynamic facility layout problems (DFLP) is examined. The CCGA has two major features. Firstly, a multi-population approach is adopted. The entire solution is constructed by individuals picked up from each population which corresponds to layouts of a production period in DFLP. An individual's fitness value is calculated with the best individuals of the other populations so that the populations can co-evolve in a cooperative manner. Secondly, steady-state selection process is embedded in each population to promote the exploitation in the genetic search. Computer simulations on two benchmark problems are conducted. It is found that the CCGA outperforms the other GAs consistently.