Set-based many-objective optimization guided by a preferred region

详细信息    查看全文

• 作者：Dunwei Gong^a ; ^{dwgong@vip.163.com}Author Vitae ; Fenglin Sun^a ; ^{fenglinsun0509@163.com}Author Vitae ; Jing Sun^b ; ^{jing8880@sina.com}Author Vitae ; Xiaoyan Sun^a ; ^{xysun78@126.com}Author Vitae
• 关键词：Many-objective optimization ; Set-based evolution ; Achievement scalarizing function ; Preferred region ; Pareto dominance on sets
• 刊名：Neurocomputing
• 出版年：2017
• 出版时间：8 March 2017
• 年：2017
• 卷：228
• 期：Complete
• 页码：241-255
• 全文大小：969 K
• 卷排序：228

文摘

Set-based evolutionary optimization based on performance indicators is one of effective methods to solve many-objective optimization problems. However, preference information of a high-dimensional objective space has not yet been fully used to guide the evolution of a population. In this paper, we propose a set-based many-objective evolutionary algorithm guided by a preferred region. In the set-based evolution, the preferred region of a high-dimensional objective space is dynamically determined, a selection strategy on sets by combining the Pareto dominance on sets with the above preferred region is designed, and the crossover operators on sets guided by the above preferred region are developed to produce a Pareto front with superior performances. The proposed method is applied to four benchmark many-objective optimization problems and a real-world engineering design optimization problem, and the experimental results empirically demonstrate its effectiveness.