An Adaptive Mutation Operator for Global Optimization Problems

Abstract

Genetic algorithms are powerful search methods for global optimization problems. Different mutation operators have been suggested in GAs to produce the next generation; it is difficult to choose which mutation operator should be applied in the evolutionary process of GAs. The adaptive mutation approach integrating various mutation operators into a simple GAs; most suitable mutation operator is adaptively selected based on the probability matching technique while solving the problem. In this novel algorithm, each individual has a group of learning strategy along with different behaviour in the search phase. This paper evaluates the performance of our suggested scheme on different problem settings. The experimental result shows that our suggested mechanism is able to select automatically the appropriate mutation operator for different problems. Several mutation operators can obviously enhance the performance of GAs.