Performance Evaluation of Bidding-Based Multi-Agent Scheduling Algorithms for Manufacturing Systems
Performance Evaluation of Bidding-Based Multi-Agent Scheduling Algorithms for Manufacturing Systems
复制标题
制造系统基于竞价的多智能体调度算法的性能评估
DOI：
--
复制
发表时间：
2014
期刊：
影响因子：
--
通讯作者：
A. Giret
中科院分区：
文献类型：
--
作者：
A. Gordillo;A. Giret
近年来，人工智能技术已被应用于制造系统中的许多问题。在具体的制造调度领域，已经发表了许多研究，试图科普复杂的制造环境。最常用的方法之一是基于（多）代理的调度。然而，尽管报告了这一领域的大量研究，但在非常常见和关键的执行情况下，没有关于这类方法业绩计量的资源或科学研究。本文主要研究基于多智能体系统（MAS）的任务分配算法，特别是在制造业中的应用。我们的目标是提供一个机制来衡量基于代理的调度方法的制造系统在关键的关键情况下，如：动态环境，重新调度，和优先级变化的性能。有了这种机制，它将有可能模拟关键的情况下，并强调系统，以衡量一个给定的基于代理的调度方法的性能。该机制是一个开拓性的方法，基于投标的制造调度MAS方法的性能评价。所提出的方法和评估方法可用于在不同的制造车间运行测试，因为它是独立的车间配置。此外，本文提出的评估结果显示的关键因素和情况下，最影响市场的MAS方法的制造调度。
Artificial Intelligence techniques have being applied to many problems in manufacturing systems in recent years. In the specific field of manufacturing scheduling many studies have been published trying to cope with the complexity of the manufacturing environment. One of the most utilized approaches is (multi) agent-based scheduling. Nevertheless, despite the large list of studies reported in this field, there is no resource or scientific study on the performance measure of this type of approach under very common and critical execution situations. This paper focuses on multi-agent systems (MAS) based algorithms for task allocation, particularly in manufacturing applications. The goal is to provide a mechanism to measure the performance of agent-based scheduling approaches for manufacturing systems under key critical situations such as: dynamic environment, rescheduling, and priority change. With this mechanism it will be possible to simulate critical situations and to stress the system in order to measure the performance of a given agent-based scheduling method. The proposed mechanism is a pioneering approach for performance evaluation of bidding-based MAS approaches for manufacturing scheduling. The proposed method and evaluation methodology can be used to run tests in different manufacturing floors since it is independent of the workshop configuration. Moreover, the evaluation results presented in this paper show the key factors and scenarios that most affect the market-like MAS approaches for manufacturing scheduling.