Non-dominated sorting methods for multi-objective optimization: Review and numerical comparison

In multi-objective evolutionary algorithms (MOEAs), non-domina-ted sorting is one of the critical steps to locate efficient solutions. A large percentage computational cost MOEAs on non-dominated for it involves numerous comparisons. By now, there are more than ten different algorithms, but their numerical performance comparing with each other not clear yet. It necessary investigate advantage and disadvantage these consequently give suggestions specific users algorithm designers. Therefore, a comprehensively study presented in this paper. Firstly, we design population generator. This generator can generate populations features, such as size, number Pareto fronts points front. Then were tested using generated certain structures, results compared respect comparisons time consumption. Furthermore, In order compare MOEAs, embed them into MOEA, dynamic genetic (DSGA), use variations DSGA solve some benchmarks. Results show that dominance degree outperforms methods, fast non-dominance performs worst equally.