A Genetic Algorithm Application with Multiple and Prioritized Objective Functions

Abstract

Since genetic algorithm was invented, it has been applied to solve lots of difficulties. One kind of the frequently encountered issues by it is problems having plural and conflicting objectives to be achieved simultaneously (i.e. multiobjective problems). However, it is unlikely to find an answer which is optimal or perfect to all conflicting objectives. The feasible way is searching compromises between these objectives and these compromises are called Pareto-optimal answers. There already are many techniques that can be used to find Pareto-optimal answers for multiobjective problems through genetic algorithm. The goal of us is addressing comprehensive service composition and it totally has four independent, conflicting, and prioritized objectives. The problem cannot be tackled by traditional way because the relationship between these objectives. To deal with such a special problem by genetic algorithm, we proposed a genetic algorithm design having layered and prioritized relationship between objective functions. Employing this design to overcome comprehensive service composition and, ultimately, we hope that the layered and prioritized functions design can be applied to other similar problems.