The Dual - Archive scheme which maintains the diversity of the solutions in the objective space and decision variable space

In this paper, Dual-Archive scheme (DA scheme) for Multi objective Genetic Algorithms is proposed. The DA scheme is the mechanism to maintain the diversity of the solutions of Multi objective Genetic Algorithms in both objective space and design variable space.When decision makers choose the solution from the Pareto solutions, they use not only the objective value information but also the design variable value information. Therefore, it is very important to maintain the diversity of solutions not only in the objective space but also in the design variable space. Since DA scheme can be applied to general Multi objective Genetic Algorithms, it is easy to use. DA scheme has two archives: one of them maintains the diversity of solutions in objective space and the other maintains the diversity in design variable space. The effectiveness of DA scheme is examined through the test functions where DA scheme is applied to SPEA2 and NSGA-II. The results showed that SPEA2 with DA scheme has the same searching ability as SPEA2 and SPEA2 with DA scheme found the solutions that have higher diversity in the design variable space compared with those of SPEA2. The tendency of the results of NSGA-II is almost the same. From these results, DA scheme is very effective scheme to derive the Pareto solutions that have the diversity not only in the objective space but also in the design variable space.