Various selection approaches on evolutionary multiobjective optimization

The study on selection approaches on multiobjective evolutionary algorithms (MOEAs) is scarce. This paper mostly investigates various selection approaches on MOEAs. Firstly, the paper discusses how to construct an appropriate fitness function in multiobjective optimization problem(MOP), then, selection approaches are classified as five categories through systematically analyzing various MOEAs. For each selection approach, we not only analyze its operation mechanism but also point out its advantage, weakness and application. At last, the paper proves the convergence of MOEAs with certain features, and the process of proof has shown that it is reasonable to regard P_known achieved from the final results of MOEAs as P_true or the approximated Pareto optimal set. Although the study is not profound, it will be helpful to design more efficient MOEAs.