Abstract: For the multi-objective optimization in real life，there are many preferences of decision makers. This paper proposes a new preference: the weight preference of objective function which derives and scores from its importance according to policy makers by Delphi method. This preference, expressive and facilitating, combines with M2M algorithm into a hybrid algorithm for solving multi-objective optimization problems. The experiments show that various preferences lead to different results in working out the multi-objective optimization problem, which coincides with the actual practice. Apart from its theoretical contribution, this preference also has its realistic significance.

Key words: weight preference; multi-objective evolutionary algorithm; optimization