Function optimization in nonstationary environment using steadystate genetic algorithms with aging of individuals

摘要

The authors explore the utility of the concept of aging ofindividuals in the context of steady state GAs for nonstationaryfunction optimization. Age of an individual is used as an additionalfactor in addition to the objective functional value in order todetermine its effective fitness value. Age of a newly generatedindividual is taken as zero, and in every iteration it is increased byone. Individuals undergoing genetic operations are selected based on theeffective fitness value, which changes dynamically. This helps tomaintain diversity in the population and is useful to trace changes inenvironment. Simulation results show some promise for the utility of thepresent technique for nonstationary function optimization