摘要
模具设计任务具有随机返工的典型特征,决策者无法做出最优决策,使用马尔可夫决策过程对具有返工过程的任务动态调度问题建立数学模型,由于返工需要更多的时间,导致无法在任务完工期内按时完工,研究旨在实现任务总拖期期望成本最小化为目标,在此基础上应用多种排序规则组合优先筛选行动集解决动态规划维数灾难问题,对比单一的规则表明,该方法在求解问题上具有的优越性。
The die ＆ mould design task had the typical characteristics of random rework. The task might not be completed on time due to reworking, and the decision maker could not make the optimal decision. The Markov decision process was used to establish a mathe- matical model for the dynamic task scheduling problem with rework process. The research aimed to minimize the expected cost of the task total delay. On this basis, a combination of multiple sorting rules was applied, and the action set was preferentially selected to solve the ＂curse of dimensionality＂ of the dynamic programming. The results showed that the proposed method was superior to the single nile in solving the problem.
作者
陈武一
王小明
CHEN Wu-yi;WANG Xiao-ming(Key Laboratory of Computer Integrated Manufacturing System of Guangdong Province,Guangdong University of Technology,Guangzhou,Guangdong 510006,China)
出处
《模具工业》
2018年第11期9-14,共6页
Die & Mould Industry
关键词
动态调度
马尔科夫决策过程
多规则组合
模具设计
决策时刻
dynamic scheduling
Markov decision process
combination of multiple rules
die ＆ mould design
decision time
作者简介
陈武一（1993-），男（汉族），湖南株洲人，硕士研究生，主要从事动态任务调度方面的研究。
