当前位置: X-MOL 学术J. Sched. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Less is more: variable neighborhood search for integrated production and assembly in smart manufacturing
Journal of Scheduling ( IF 2 ) Pub Date : 2019-09-07 , DOI: 10.1007/s10951-019-00619-5
Shaojun Lu , Jun Pei , Xinbao Liu , Xiaofei Qian , Nenad Mladenovic , Panos M. Pardalos

This paper investigates an integrated production and assembly scheduling problem with the practical manufacturing features of serial batching and the effects of deteriorating and learning. The problem is divided into two stages. During the production stage, there are several semi-product manufacturers who first produce ordered product components in batches, and then these processed components are sent to an assembly manufacturer. During the assembly stage, the assembly manufacturer will further process them on multiple assembly machines, where the product components are assembled into final products. Through mathematical induction, we characterize the structures of the optimal decision rules for the scheduling problem during the production stage, and a scheme is developed to solve this scheduling problem optimally based on the structural properties. Some useful lemmas are proposed for the scheduling problem during the assembly stage, and a heuristic algorithm is developed to eliminate the inappropriate schedules and enhance the solution quality. We then prove that the investigated problem is NP-hard. Motivated by this complexity result, we present a less-is-more-approach-based variable neighborhood search heuristic to obtain the approximately optimal solution for the problem. The computational experiments indicate that our designed LIMA-VNS (less is more approach–variable neighborhood search) has an advantage over other metaheuristics in terms of converge speed, solution quality, and robustness, especially for large-scale problems.

中文翻译:

少即是多:智能制造中集成生产和装配的可变邻域搜索

本文研究了一个集成的生产和装配调度问题,该问题具有串行批处理的实际制造特征以及恶化和学习的影响。问题分为两个阶段。在生产阶段,有几个半成品制造商首先批量生产订购的产品部件,然后将这些加工好的部件送到组装制造商。在组装阶段,组装制造商将在多台组装机器上进一步加工它们,在那里将产品组件组装成最终产品。通过数学归纳法,我们表征了生产阶段调度问题的最优决策规则的结构,并根据结构特性开发了一种方案来优化解决这个调度问题。针对装配阶段的调度问题提出了一些有用的引理,并开发了一种启发式算法来消除不适当的调度并提高解决方案的质量。然后我们证明所研究的问题是 NP-hard。受此复杂性结果的启发,我们提出了一种基于少即是多方法的变量邻域搜索启发式方法,以获得问题的近似最优解。计算实验表明,我们设计的 LIMA-VNS(少即是多——变量邻域搜索)在收敛速度、求解质量和鲁棒性方面优于其他元启发式算法,尤其是对于大规模问题。针对装配阶段的调度问题提出了一些有用的引理,并开发了一种启发式算法来消除不适当的调度并提高解决方案的质量。然后我们证明所研究的问题是 NP-hard。受此复杂性结果的启发,我们提出了一种基于少即是多方法的变量邻域搜索启发式方法,以获得问题的近似最优解。计算实验表明,我们设计的 LIMA-VNS(少即是多——变量邻域搜索)在收敛速度、求解质量和鲁棒性方面优于其他元启发式算法,尤其是对于大规模问题。针对装配阶段的调度问题提出了一些有用的引理,并开发了一种启发式算法来消除不适当的调度并提高解决方案的质量。然后我们证明所研究的问题是 NP-hard。受此复杂性结果的启发,我们提出了一种基于少即是多方法的变量邻域搜索启发式方法,以获得问题的近似最优解。计算实验表明,我们设计的 LIMA-VNS(少即是多——变量邻域搜索)在收敛速度、求解质量和鲁棒性方面优于其他元启发式算法,尤其是对于大规模问题。然后我们证明所研究的问题是 NP-hard。受此复杂性结果的启发,我们提出了一种基于少即是多方法的变量邻域搜索启发式方法,以获得问题的近似最优解。计算实验表明,我们设计的 LIMA-VNS(少即是多——变量邻域搜索)在收敛速度、求解质量和鲁棒性方面优于其他元启发式算法,尤其是对于大规模问题。然后我们证明所研究的问题是 NP-hard。受此复杂性结果的启发,我们提出了一种基于少即是多方法的变量邻域搜索启发式方法,以获得问题的近似最优解。计算实验表明,我们设计的 LIMA-VNS(少即是多——变量邻域搜索)在收敛速度、求解质量和鲁棒性方面优于其他元启发式算法,尤其是对于大规模问题。
更新日期:2019-09-07
down
wechat
bug