KNOWLEDGE-BASED DECISION-MAKING FOR JOB SHOP SCHEDULING

In this paper an approach to on-line job shop scheduling is described in which a knowledge-based system for analysis and decision-making and a dynamic scheduler for simulating dispatching strategies into a workplan, are combined. How relevant events originating from the shopfloor or higher hierarchical levels of the factory can be analysed by a knowledge-based system is investigated to determine if the current workplan or schedule should be modified. Secondly, in finding a new sub-optimal schedule the knowledge-based system selects priority rules to be used by the scheduler and evaluates the resulting schedule to explicit criteria which are represented in rules and therefore easily modifiable to the manufacturing environment. Some of the methods used by the current prototype system are described and some results of the knowledge acquisition are presented. The knowledge-based system will be part of a production control system that is under development in ESPRIT Project 809. The control system will be used in a real manufacturing cell.