A decomposition-based many-objective evolutionary algorithm with optional performance indicators

Evolutionary algorithms (EAs) have shown excellent performance for solving optimization problems with multiple objectives as they can get a set of compromising solutions on a single run. However, when the number of objectives increases, an efficient selection is significant to find a good set of solutions. In this paper, a decomposition-based many-objective evolutionary algorithm with optional performance indicators is proposed, in which the decomposition strategy is utilized to convert a many-objective optimization problem into a set of single-objective optimization problems, and the criterion to select a solution for the next generation along each reference is randomly set to convergence or diversity performance. The performance of the proposed method is evaluated on two sets of benchmark problems, and the experimental results showed the efficiency of the proposed method compared with seven state-of-the-art MaOEAs.