I. Introduction
In engineering applications, there exist some multiobjective optimization problems (MOPs) that require the simultaneous optimization of multiple (often conflicting) objectives. As multiobjective evolutionary algorithms (MOEAs) can find a set of Pareto-optimal solutions in a single run, they become an effective and popular tool for tackling MOPs [1]. Based on the selection criteria, most MOEAs can be classified into three main categories [2]: 1) Pareto-based MOEAs [3], [4]; 2) decomposition-based MOEAs [5], [6]; and 3) indicator-based MOEAs [7], [8]. These traditional MOEAs generally assume a sufficient number of function evaluations, so that the population can converge. However, for some MOPs modeled from practical applications, e.g., finite element analysis [9], computational fluid dynamics [10], and computational electromagnetics [11], the function evaluations require computationally expensive simulations, which consume a considerable amount of time or material resources. These problems are often called expensive MOPs (EMOPs) [12].