A-Grid-Based-Evolutionary-Algorithm-for-Many-Objective-Opt啦啦啦啦啦啦imization.pdfVIP

IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION, VOL. 17, NO. 5, OCTOBER 2013 721 A Grid-Based Evolutionary Algorithm for Many-Objective Optimization Shengxiang Yang, Member, IEEE, Miqing Li, Xiaohui Liu, and Jinhua Zheng Abstract —Balancing convergence and diversity plays a key I. Introduction role in evolutionary multiobjective optimization (EMO). Most current EMO algorithms perform well on problems with two N THE REAL world, it is not uncommon to face a problem or three objectives, but encounter difficulties in their scalability Iwith several objectives to be met simultaneously, which to many-objective optimization. This paper proposes a grid- is known as a multiobjective optimization problem (MOP). based evolutionary algorithm (GrEA) to solve many-objective Due to the conflicting nature of objectives, there is usually no optimization problems. Our aim is to exploit the potential of the grid-based approach to strengthen the selection pressure toward single optimal solution but rather a set of alternative solutions, the optimal direction while maintaining an extensive and uniform called the Pareto set, for MOPs. Evolutionary algorithms (EAs) distribution among solutions. To this end, two concepts—grid have been recognized to be well suited for MOPs due to dominance and grid difference—are introduced to determine the their population-based property of achieving an approxima- mutual relationship of individuals in a grid environment. Three tion of the Pareto set in a single run. Over the past few grid-based