Research On Multi-objective Evolutionary Algorithms With Local Preference And Dynamics

The main content of this paper is to study the multi-objective optimization algorithm with preference and the dynamic multi-objective evolutionary algorithm,and numerical experiments are carried out on the proposed algorithms.This paper firstly gives some basic definitions of the multi-objective optimization problem,briefly introduces three kinds of commonly used multi-objective evolutionary algorithms,then gives the multiobjective evolutionary algorithm based on decomposition in detail,expounds the working principle of the decomposition method,and takes the two-objective problem as an example to analyze the geometric meaning of the scalar function used.Multi-objective optimization problems with user preferences are aimed at finding one or more local areas in the Pareto frontier that users are most interested in.Reference points and preference vectors are widely used to solve these problems.In this paper,based on the decomposition method,a multi-objective decomposition algorithm with preference is proposed,which takes the optimal preference solution as the center in the target space and uses the local principal component analysis method to determine the preference region.Principal component analysis(PCA)can reduce the computational complexity of the problem,facilitate the fast determination of the preference region,and at the same time,precisely adjust the weight vector in the preference region,and guide the population to concentrate in the preference region.In the dynamic multi-objective optimization problem,the key to solve the problem is to respond to the environment change quickly.Kalman filter as a linear filter is often used to evaluate the state of motion at the next moment.In this paper,a dynamic multiobjective evolutionary algorithm based on cooperative kalman filter prediction is proposed,and the variation rule of the population is added into the kalman filter prediction strategy,that is,the change of the optimal solution set under the adjacent environment is considered simultaneously.When the environment changes,the new algorithm makes full use of the similarity and difference between the optimal solution sets in the dynamic environment,which makes it easier to respond to the changes in the environment.,so as to quickly obtain the optimal solution set in the new environment.The experimental results show that the new algorithm is effective in eight dynamic multi-objective problems.