Hybrid hierarchical scheduling and control systems in manufacturing

Some experiments on the integration of algorithmic techniques with knowledge-based ones are discussed. Two case studies are presented: an FMS cell and a press shop. It was found that the algorithmic procedures developed for production scheduling resulted in limiting the ability to cope with the complexity of the real manufacturing world. The scheduling problem, seen as a constraint satisfaction problem, can be approached with rule-based techniques. Nevertheless, algorithmic techniques are found to be valuable for their efficiency and ability to deal with aggregated data. This ability is fundamental for an efficient implementation of hierarchical control systems in general and in the manufacturing context in particular. This suggests that the integration of rule-based techniques with algorithmic ones can increase the efficiency of searching in the space of possible solutions. The ability to deal with aggregated data can have little value when detailed real-time operation scheduling is needed. In this case, simple dispatching rules are often used, and sophisticated operations research methods are not used. In such a dynamic situation, a purely-rule based approach may be more suitable.