| Hi-index | 0.00 |
It could be concluded that all multi-objective evolutionary algorithms draw their strength from two aspects: convergence and diversity. In order to achieve these goals, This paper proposes a hybrid methods that combines GA with simplex search method for multi-objective optimization using preference order ranking. Preference order ranking is used as fitness assignment methodology to accelerate the performance of convergence, especially when the number of objectives is very large. The proposed algorithm also uses three subsets to evolve simultaneous and each subset is divided on the basis of different criterion. In every generation, We carry out simplex-based local search in the first two subsets to achieve faster convergence and better diversity, and the individuals in third subset execute ordinary genetic operator to avoid premature convergence. The proposed algorithm has been compared with other MOEAs in high dimensional problems. The experimental results indicate that our algorithm produces better convergence and diversity performance.