Development of a Self Adaptive Genetic Algorithm

Genetic Algorithms have been used to solve difficult optimization problems in a number of fields. However, in order to solve a problem with GA, the user has to specify a number of parameters. This parameter tuning is a difficult task as different genetic operators are suitable in different application areas. This paper proposes a scheme for genetic algorithms where the genetic operators are changed randomly. The proposed approach aims to mimic the nature more closely. In this approach, inhomogeneous crossover and selection techniques are used. A gendered reproduction is also applied where the number of children is produced depending on the fertility rate. In addition, parents may adopt a new child. The age and death age are added to balance between exploration and exploitation of the search space. Using these simple approaches, the diversity of the population can be maintained efficiently. The experimental results of the proposed algorithm based on a mechanical design problem show promising results.