• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>IEEE Symposium Series on Computational Intelligence >Evaluating decomposition strategies to enable scalable scheduling for a real-world multi-line steel scheduling problem
【24h】

Evaluating decomposition strategies to enable scalable scheduling for a real-world multi-line steel scheduling problem

机译：评估分解策略以实现可扩展的调度，以解决实际的多线钢调度问题

获取原文
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Steel scheduling is recognised as one of the most difficult real-world scheduling problems. It is characterised by a wide range of operational constraints, variable dependencies and multiple objectives. This paper uses a divide and conquer method to reduce the combinatorial complexity of a real-world multi-line steel scheduling problem. The problem is first decomposed into sub-problems which are solved individually in parallel using parallel branch and bound, then sub-problems are combined to form a solution to the original problem. Three decomposition strategies are compared, specifically: a manual heuristic domain knowledge (DOM) intensive strategy, K-means++ (KM) clustering and Self-organising maps (SOM). Experimental results show that using SOM for decomposition is a promising approach. This paper demonstrates that despite being a highly complex and constrained problem, it is possible to use divide and conquer to achieve potentially good scalability characteristics without significant detriment to the solution quality.
机译：钢调度被认为是最困难的现实调度问题之一。它的特点是广泛的操作约束，可变的依赖关系和多个目标。本文采用分而治之的方法来减少现实世界中多线钢调度问题的组合复杂性。首先将问题分解为子问题，然后使用并行分支和边界并行解决这些子问题，然后将子问题组合起来以形成原始问题的解决方案。比较了三种分解策略，特别是：手动启发式领域知识（DOM）密集策略，K-means ++（KM）聚类和自组织映射（SOM）。实验结果表明，使用SOM进行分解是一种很有前途的方法。本文证明，尽管存在高度复杂和受约束的问题，但仍可以使用分而治之来实现潜在的良好可伸缩性，而不会显着损害解决方案的质量。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号