Survival selection methods for the Differential Evolution based on continuous generation model

Abstract

This paper presents several survival selection methods for a new Differentiation Evolution (DE) based on the continuous generation model. The standard DE employs the discrete generation model in which the current-generation population is replaced by the next-generation population at a time. On the other hand, only one population is used in the continuous generation model. Because a newborn excellent individual is added to the population and can be used immediately to generate offspring, it can be expected that the new DE based on the continuous generation model converges faster than the standard DE. Furthermore, it becomes easy to introduce various survival selection methods into the new DE. Therefore, five survival selection methods are contrived for the new DE. Finally, the effects of those survival selection methods are studied by using the analysis of variance (ANOVA).