Wyszukaj
Zmień rozmiar czcionki
Wersje kontrastowe
| Prelegent: | prof. MengChu Zhou, New Jersey Institute of Technology Newark, NJ 07102, USA, e-mail: zhou@njit.edu |
| Temat: | Deadlock-free Control of Automated Manufacturing Systems: A Petri Net Approach |
| Streszczenie: | In automated manufacturing systems (AMS), deadlocks disrupt production and leave resources idle, thereby lowering efficiency. Deadlock avoidance policies outperform deadlock prevention and recovery strategies in terms of system throughput and resource utilization. For systems of sequential systems with shared process resources (S4PR), which is a subclass of Petri nets (PNs) often used to model AMS, this work introduces a one-step look-ahead supervisory control approach to avoid deadlocks. First, deadlocks are identified through structural analysis. The states in S4PR are classified as legal or illegal, with illegal states being those that cannot return to the initial one. By backward traversing a reachability tree from obtained deadlocks, legal states, called dangerous ones, that can reach illegal ones in just one step are detected. The above two processes are done offline during the system design phase. Subsequently, an online supervisor control strategy is designed to prevent illegal states. While an AMS is running, a supervisor determines the AMS’s state by observing its behavior in real-time and compares the state with dangerous ones to guide the operation of the system, thereby ensuring that the system is live. In this supervisory control policy, for some observed behavior, the supervisor can directly make a correct decision without determining the state of the system. To further decrease the decision time of the supervisor, a set of minimal dangerous states is proposed to reduce the number of states that the supervisor needs to compare. Applying this policy results in a deadlock-free AMS that retains all legal states without altering the model structure and without requiring state-space traversal during system operation. This study enhances the application of PNs in AMS development to meet industrial needs. Further information on file. |
