Multi-objective Evolutionary Algorithm Based On Weight Preference

In solving multi-objective optimization problems, we generally adopts the multi-objecti ve evolutionary algorithms, evolutionary algorithm is a random search algorithm of simulated biological natural selection and evolution, because it is suitable for solving the highly complicated nonlinear problem,so its application is very wide. Using evolutionary algorithm to solve multi-objective problems, we often work out a set of pareto optimal solutions, these solutions have no good or bad, so for policy makers, if they want to choose the satisfactory optimal solution,it’s the certain difficulty, then the key of breaking the balance is introducing the preference. The purpose of preference information is to make decision makers through its understanding of specific problems, involved in the optimization process, and then makes the algorithm search focuses on decision space and the ideal target space area, so as to make the algorithm more efficient.In this paper,we do a simple summary to the existing preference Multi-objective evolutionary algorithms, researchers have preferences in the study of ways including reference point, reference direction, preference area, fuzzy preference and random preference and so on, this article summarized the three preferences form in detail of the preferences area, random and fuzzy preference. the researchers made a specific research results.A preference for each way, is given in this paper, this paper summarizes the preference approach, different researchers have put forward and adopted by the multi-objective evolutionary algorithms, and several of prefer ways have carried on the analysis of the advantages and disadvantages, it is advantageous to the later scholars study.For the multi-objective optimization problems in real life, there are many preferences on decision makers.This paper proposes a new way of preference:the weight of the objective function preference.Under the Delphi method,policy makers score the importance of the objective function,this method can better reflect the preference of decision makers, and very simple. And then combined with M2M algorithm, the sub-region strategy has two very desirable properties with regard to multi-objective optimization.this paper formed a kind of hybrid algorithm for solving multi-objective optimization problems.Numerical experiment results show that under different preferences, the result of the multi-objective optimization problem is also different, that in conformity with the actual circumstances.