An Integrated Optimization and Knowledge - Based Job Shop Scheduling System

Abstract The complexity of the job shop scheduling problem has led some researchers away from the difficulties associated with optimization approaches. Instead, they have chosen to pursue knowledge-based techniques more capable of handling dynamic changes such as machine breakdowns and new jobs. This paper presents a synergetic combination of optimization and knowledge-based methods in a system of distributed architecture. The high level knowledge-based component schedules the operations of a job through a contract negotiation process with relevant work centers. A scheduling algorithm based on Lagrangian relaxation is adapted to schedule individual work centers. Work centers “bid” on operations and the knowledge-based components “ward” contracts for their completion. Methods to prevent distributed components from violating the decisions of others are considered. The tradeoff between obtaining a good schedule and meeting originally promised job delivery dates is also examined. The net result is an effective, informative, and flexible job shop scheduling system.