多级生产批量规划问题的柔性惯量反捕食粒子群算法  

Anti-predatory particle-swarm optimization with flexible inertial weight for unconstrained multilevel lot-sizing problems

作　　者：韩毅[1] 蔡建湖[1] 周根贵[1] 李延来[2] 唐加福[2] 

机构地区：[1]浙江工业大学经贸管理学院,浙江杭州310023 [2]东北大学流程工业综合自动化教育部重点实验室,辽宁沈阳110819

出　　处：《控制理论与应用》2010年第10期1300-1306,共7页Control Theory & Applications

基　　金：国家自然科学基金资助项目(70625001;70721001;70671095;70971017);浙江省科技计划软科学研究资助项目(2009C35007);浙江省自然科学基金资助项目(Y1100854);浙江省社科规划课题资助项目(10CGGL21YBQ)

摘　　要：多级生产批量规划(MLLS)是原料需求计划(MRP)中主生产计划(MPS)的关键决策问题,具有广泛的工业应用;已被证明是NP-hard类型的组合优化问题.反捕食粒子群算法(APSO)是最近提出的一种与粒子群算法(PSO)密切相关的亚启发式算法.本文提出带柔性惯性权重的反捕食粒子群算法(WAPSO)对具有指定装配结构而无约束的MLLS问题进行了求解.本算法对12个小规模benchmark数据集和1个随机产生的较大规模数据进行了测试.测试结果与遗传算法(GA)和Wagner-Whitin(WW)动态规划算法的结果进行了比较.结果表明了WAPSO算法的有效性和适用性.Multilevel lot-sizing（MLLS） is a crucial problem in decision-making for the master production scheduling（MPP） of the material requirement plan（MRP）, which is with broad industrial applications and has been considered the NP-hard combinatory optimization problem. Anti-predatory particle-swarm optimization（APSO）, which is closely related to particle-swarm optimization（PSO）, is a recently emerged meta-heuristics. An anti-predatory particle-swarm optimiztion with flexible inertial weight（WAPSO） is proposed to solve the unconstrained MLLS problem in a given assembly structure. A set of 12 small-sized benchmark data and a randomly generated medium size data are adopted to test the proposed algorithm. The experimental results are compared with those of genetic algorithm（GA） and Wagner-Whitin（WW） dynamic programming algorithm, the results show that WAPSO algorithm is an effective and suitable tool for solving the unconstrained MLLS problem in a given assembly structure.

关 键 词：多级生产批量规划 反捕食粒子群算法 亚启发式算法 惯性权重 装配结构 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]