Influence of the population size on the genetic algorithm performance in case of cultivation process modelling

In this paper, an investigation of the influence of the population size on the genetic algorithm (GA) performance for a model parameter identification problem, is considered. The mathematical model of an E. coli fed-batch cultivation process is studied. The three model parameters - maximum specific growth rate (μ
_max
), saturation constant (k
_S
) and yield coefficient (Y
_S/X
) are estimated using different population sizes. Population sizes between 5 and 200 chromosomes in the population are tested with constant number of generations. In order to obtain meaningful information about the influence of the population size a considerable number of independent runs of the GA are performed. The observed results show that the optimal population size is 100 chromosomes for 200 generations. In this case accurate model parameters values are obtained in reasonable computational time. Further increase of the population size, above 100 chromosomes, does not improve the solution accuracy. Moreover, the computational time is increased significantly.