MOEA/D with Gradient-Enhanced Kriging for Expensive Multiobjective Optimization

Fei Liu, Qingfu Zhang, Zhonghua Han

Abstract: Expensive multiobjective optimization problem poses a big challenge. In many real-world engineering design problems, the time-consumed function evaluation is done by solving partial differential equations. The partial derivatives of a candidate solution can be calculated as a byproduct. Naturally, these problems can be solved more efficiently if gradient information is used. This paper proposes such a method, called MOEA/D-GEK, which combines MOEA/D and gradient-enhanced Kriging to solve expensive multiobjective problem. The gradient information is used for the construction of the Kriging model. Experimental studies on a set of test instances and a real-world aerodynamic design problem show high efficiency and effectiveness of our proposed method.