Abstract
Abstract
This paper describes an investigation of the effects of dynamic job routing and job sequencing decisions on the performance of a distributed control system and its adaptability against disturbances. This experimental work was carried out to compare the performance of alternative control strategies in various manufacturing environments and to investigate the relationship between the ‘control’ and ‘controlled’ systems. The experimental test-bed presented in this paper consists of an agent-based control system (implemented in C++) and a discrete-event simulation model. Using this test-bed, various control strategies were tested on a benchmark manufacturing system by varying production volumes (to model the production system with looser/tighter schedules) and disturbance frequencies. It was found that hybrid strategies that combine reactive agent mechanisms (and allocation strategies such as the contract net) with appropriate job sequencing heuristics provide the best performance, particularly when job congestion increases on a shop-floor.