A multi-objective flow shop scheduling with resource-dependent processing times: trade-off between makespan and cost of resources
A multi-objective flow shop scheduling with resource-dependent processing times: trade-off between makespan and cost of resources
复制标题
DOI：
10.1080/00207543.2010.523724
复制
发表时间：
2011-01
影响因子：
9.2
通讯作者：
H. Mokhtari;I. Abadi;Ali Cheraghalikhani
中科院分区：
文献类型：
--
作者：
H. Mokhtari;I. Abadi;Ali Cheraghalikhani
在传统的流水作业调度问题(FSSP)中，加工时间被假设为预先已知的固定参数，而在许多实际环境中，加工时间可以通过消耗额外资源来控制。本文提出了置换FSSP问题的资源相关加工时间(RDPT)，其中一个作业的加工时间取决于分配给该作业的额外资源量。为了在最大完工时间和所需资源量之间进行权衡，考虑了两个冲突目标函数：最小化最大完工时间和最小化总资源成本。为了解决本文所考虑的问题，提出了一种分解方法，力求通过两个子问题来处理原始模型：(I)排序问题；(Ii)资源分配问题。将有效坐标方向搜索和变邻域搜索相结合的混合离散差分进化算法(HDDE)用于求解两个子问题。此外，使用统计过程来调整所提出的HDDE和VNS算法的重要参数。这一过程基于逐步回归(SR)技术。通过计算研究了该混合算法的有效性，结果表明该算法相对于其他算法具有较好的性能。
In traditional flow shop scheduling problems (FSSPs), the processing times are assumed to be pre-known and fixed parameters while in many practical environments, the processing times can be controlled by consumption of extra resources. In this paper, we propose resource-dependent processing times (RDPT) for permutation FSSP in which the processing time of a job depends on the amount of additional resources assigned to that job. To make a trade-off between makespan and required amount of resources, two conflict objective functions are considered: the minimisation of maximum completion time and the minimisation of total cost of resources. In order to solve the problem considered in this paper, a decomposition approach is suggested that strives to deal with the original model via two sub-problems: (i) sequencing problem; and (ii) resource allocation problem. A hybrid discrete differential evolution (HDDE) algorithm with an effective coordinate directions search and a variable neighbourhood search (VNS) are combined to solve two sub-problems. Furthermore, a statistical procedure is employed to adjust the significant parameters of the proposed HDDE and VNS algorithms. This procedure is based on the stepwise regression (SR) technique. The effectiveness of the suggested hybrid algorithm is investigated through a computational study and obtained results show the good performance of our approach with regard to the other algorithms.