Reliability-inspired species conservation for multimodal functions

Abstract

Species Mutation technique, inspired from reliability design, is combined with a species conserving genetic algorithm (SCGA) to solve multimodal optimization problems. In reliability design, in order to improve system reliability and safety requirements, a redundancy component is added to a most efficient subsystem, in which the ratio of the increase of reliability to the increase of resources is maximal. This idea is introduced into the SCGA and a proper species is selected for mutation in order to maximally improve the SCGA efficiency. After each generation, the SCGA will determine species from populations and find the solutions from species. Species mutation mutates a selected species. Some numerical tests show that this simple technique can greatly decrease objective evaluations for the SCGA in searching solutions of the test problems.