Abstract
In evolutionary many-objective optimization, an effective environmental selection strategy is crucial to the performance of evolutionary algorithms, especially when the number of objectives is large. In this paper, we propose a new method for environmental selection, where three indicators are calculated and nondominated sorted for individual selection to be passed to the next generation. Of the three indicators, two aim to select individuals close to the center and edge part of the front in the objective space, and third one focuses on promoting the diversity of the population in the decision space to find promising solutions. The proposed algorithm is evaluated on the commonly used DTLZ test suite for many-objective optimization. Our comparative experimental results show that the proposed method is competitive compared to the state-of-the-art, especially on convergence performance in solving many-objective problems.