A directed mutation operator for real coded genetic algorithms

Abstract

Developing directed mutation methods has been an interesting research topic to improve the performance of genetic algorithms (GAs) for function optimization. This paper introduces a directed mutation (DM) operator for GAs to explore promising areas in the search space. In this DM method, the statistics information regarding the fitness and distribution of individuals over intervals of each dimension is calculated according to the current population and is used to guide the mutation of an individual toward the neighboring interval that has the best statistics result in each dimension. Experiments are carried out to compare the proposed DM technique with an existing directed variation on a set of benchmark test problems. The experimental results show that the proposed DM operator achieves a better performance than the directed variation on most test problems.