Dynamic Selection of Dispatching Rules: a Neural Network Approach

摘要：

Dispatching rules are widely used in production systems to schedule the production in real-time. Numerous dispatching rules exist. Unfortunately no Dispatching Rule (DR) is globally better than any other. Their efficiency depends on the characteristics of the system, operational conditions and the production objectives. A new approach based on neural networks (NN) is proposed here to select dynamically in real time the most suited DR, based on the current system state and the workshop operating conditions. A new decision about which rule to select is made each time a resource becomes available. Contrarily to learning approaches presented in the literature, used to select scheduling heuristics, no training set is needed. The NN parameters are determined through simulation optimization. Such an approach presents a great potential to address control problems. Its benefits are illustrated through the example of a simplified flow shop. It is shown that the NN can automatically select the most efficient DR without any a priori knowledge.

关键词： Simulation Neural Network dynamic scheduling learning flowshop

作者: Wiem MOUELHI Henri PIERREVAL Mame Bigué GNINGUE

作者单位: LIMOS,IFMA,Clermont-Ferrand,France Cerene Laboratory,Le Havre University,France

会议类型: 国际会议

会议名称: 工业工程与系统管理2007年国际会议(International Conference on Industrial Engineering and Systems Management)(IESM 2007)

会议地点: 北京

会议语种:英文

在线出版日期: 2007-05-30（万方平台首次上网日期，不代表论文的发表时间）