基于混合选择的多目标进化算法及其在优化设计问题中的应用  

Hybrid selection based multi-objective evolutionary algorithm and its application in optimization design problem

作　　者：王万良[1] 李伟琨[1] 臧泽林 赵燕伟[1,2] WANG Wanliang;LI Weikun;ZANG Zelin;ZHAO Yanwei(Institute of Computer Vision,Zhejiang University of Technology,Hangzhou 310023,China;Key Lab of Special Purpose Equipment and Advanced Manufacturing Technology,Ministry of Education/Zhejiang Province,Zhejiang University of Technology,Hangzhou 310014,China)

机构地区：[1]浙江工业大学计算机视觉研究所,浙江杭州310023 [2]浙江工业大学特种设备制造与先进加工技术教育部/浙江省重点实验室,浙江杭州310014

出　　处：《计算机集成制造系统》2020年第7期1802-1813,共12页Computer Integrated Manufacturing Systems

基　　金：国家自然科学基金资助项目(61379123,61572438)。

摘　　要：为解决实际工程中复杂的多目标优化设计问题,提出一种基于混合选择的多目标进化算法(HSMEA)。该算法首先采用融合角度与距离的动态选择策略对个体进行划分,随后采用基于两种不同机制的混合选择的策略对解进行进一步筛选,从而使最终选择的目标解在具有良好的收敛性的同时最大程度地保留解集的多样性。算法与4个多目标优化算法在一系列测试函数上的结果表明,算法具有良好的多目标优化问题处理能力。此外,通过在实际工程优化设计问题的对比实验与分析,验证了所提算法在处理实际工程优化设计问题上具有良好的性能与潜力。To solve the complex multi-objective optimization design problem in actual engineering problem, a Multi-Objective Evolutionary Algorithm based on Hybrid Selection(HSMEA) was proposed. A dynamic strategy which combined angle and distance was implemented to assist the algorithm to cluster the individuals. Thereafter, the hybrid selection mechanism that included two different principles was conducted to further select the solutions. Thus, solutions were obtained with the better balance of convergence and diversity. The proposed method was compared with other four state-of-the-art multi-objective evolutionary algorithms on a number of test problems. The experimental results showed that the competitiveness and effectiveness of the proposed algorithm in addressing the multi-objective problems. Additionally, HSMEA was implemented in the engineering design optimization cases, which verified the well performance and the potential of the proposed algorithm in real-world optical problems.

关 键 词：多目标优化 混合选择 进化算法 工程设计 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]