This paper introduces an intelligent flexible workstation controller integrating Petri Net models, heuristic search methods and error recovery capability. The activities of the physical workstation are modeled via a re-configurable Petri Net formalism that allows the incorporation of recovery tasks into its logic. The workstation controller follows a production plan generated off-line during its normal operation and appropriately reacts when the original plan is disturbed due to the presence of errors. The off-line production plan is generated via a heuristic Beam A{sup}* Search (BAS) algorithm presented in previous work. The implications of the error recovery tasks from the perspective of control are also discussed. When errors occur and the system cannot follow the original schedule, the controller attempts to reach the original production plan in a later stage in such a way that the effect of the disruption is minimized. This is accomplished with a modified version of the BAS algorithm adapted for on-line scheduling. This modified BAS algorithm generates a partial plan of activities that "matches-up" with the original production plan. Simulation tests were conducted in order to study the behavior of the algorithm under different scenarios. The results are and analyzed discussed. We conclude that our approach can be a valuable tool for real time control of flexible manufacturing systems.
展开▼
