基于改进MOEA/D的多目标置换流水车间调度问题  

Multi-objective permutation flowshop scheduling problem based on improved MOEA/D

作　　者：李林林[1] 刘东梅 王显鹏[1,3] LI Linlin;LIU Dongmei;WANG Xianpeng(Key Laboratory of Data Analytics and Optimization for Smart Industry, Ministry of Education, Northeastern University, Shenyang 110819,China;Department of Computer, Shenyang School of Traditional Chinese Medicine, Shenyang 110300, China;Liaoning Provincial Engineering Laboratory of Operations Analytics and Optimization for Smart Industry, Northeastern University, Shenyang 110819, China)

机构地区：[1]东北大学智能工业数据解析与优化教育部重点实验室,辽宁沈阳110819 [2]沈阳中医药学校计算机系,辽宁沈阳110300 [3]东北大学辽宁省智能工业数据解析与优化工程实验室,辽宁沈阳110819

出　　处：《计算机集成制造系统》2021年第7期1929-1940,共12页Computer Integrated Manufacturing Systems

基　　金：国家重点研发计划资助项目(2018YFB1700404);国家自然科学基金资助项目(61573086,71790614,71621061);教育部111创新引智基地资助项目(B16009)。

摘　　要：针对工期和总流水时间的两目标置换流水车间调度问题,提出一种改进的基于分解的多目标进化算法(MOEA/D)。为了改进非支配解集的质量,提高算法效率,在MOEA/D中嵌入分组和统计学习机制提出一种两阶段局部搜索策略改进外部存档。利用基于距离的替换策略更新种群,提高种群的多样性,保证了分组机制的有效性。基于Taillard标准测试问题的实验结果表明,所提出的改进MOEA/D算法明显优于传统MOEA/D、NSGA-Ⅱ、MEDA/D-MK等算法。To solve the bi-objective permutation flowshop scheduling problem with makespan and total flow time minimized,an improved Multi-objective Evolutionary Algorithm based on Decomposition(MOEA/D)algorithm was proposed.To improve the quality of non-dominant solutions and enhance the efficiency of proposed algorithm,the grouping and statistical learning mechanism was embedded in a two-stage local search strategy,which was used to improve the external archive.The population was updated by adopting the perpendicular distance-based replacement strategy,and the effectiveness of the grouping mechanism was ensured.Computational results on the Taillard benchmark problems showed that the proposed improved MOEA/D was significantly superior to the traditional MOEA/D,NSGA-Ⅱand MEDA/D-MK algorithms.

关 键 词：置换流水车间调度 多目标 分解 进化算法 统计学习 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]