A Constrained Multi-Objective Evolutionary Algorithmbased on Fitness Landscape Indicator

Constrained multi-objective optimization problems (CMOPs) with constraintsin both the decision and objective space are shown to be great challenges tobe solved. Considering that different requirements of different problems onresource allocation of exploration and exploitation, this paper proposes anindicator to determine the selection of offspring generation mechanism. Theindicator regards the fitness landscape and evolutionary generation to meetdifferent problems. The proposed indicator is demonstrated to reflect thedifficulties of exploration and exploitation of different problems. By embeddingthe indicator of the selection of two differential evolution operators,we propose a constrained multi-objective optimization algorithm for solvingCMOPs. Numerical experiments on three test suites comparing with sixexisting algorithms show the proposed algorithm can effectively deal withdifferent types of CMOPs, especially in CMOPs with constraints in both thedecision and objective spaces.