PROJECT TITLE :

A Multiobjective Evolutionary Algorithm Based on Decision Variable Analyses for Multiobjective Optimization Problems With Large-Scale Variables

ABSTRACT:

State-of-the-art multiobjective evolutionary algorithms (MOEAs) treat all the decision variables as a full to optimize performance. Inspired by the cooperative coevolution and linkage learning methods in the sector of single objective optimization, it's fascinating to decompose a tough high-dimensional downside into a collection of easier and low-dimensional subproblems that are easier to resolve. But, with no previous data about the objective function, it is not clear the way to decompose the target function. Moreover, it's troublesome to use such a decomposition methodology to resolve multiobjective optimization problems (MOPs) as a result of their objective functions are commonly conflicting with one another. That is to mention, changing call variables can generate incomparable solutions. This paper introduces interdependence variable analysis and control variable analysis to accommodate the on top of two difficulties. Thereby, an MOEA primarily based on call variable analyses (DVAs) is proposed in this paper. Control variable analysis is employed to recognize the conflicts among objective functions. A lot of specifically, that variables affect the diversity of generated solutions and that variables play an important role within the convergence of population. Primarily based on learned variable linkages, interdependence variable analysis decomposes call variables into a group of low-dimensional subcomponents. The empirical studies show that DVA will improve the answer quality on most difficult MOPs. The code and supplementary material of the proposed algorithm are obtainable at http://net.xidian.edu.cn/fliu/paper.html.