Optimal Behavior Research For A Kind Of Deadlock Supervisor In Flexible Manufacturing Systems

The competition for limited resources can produce deadlocks in flexible manufacturing systems (FMS). Petri net is an effective tool to model, analyze, and control deadlocks in FMS. Most deadlock prevention methods in recent years add some supervisors which consist of new contorl places and related arcs to the target Petri net. Ezpeleta's method is to add supervisor to evory strict minimal siphon (SMS) which causes deadlocks such that no siphon can be emptied. Since the number of siphons grows in general exponentially with respect to the size of a Petri net, its disadvantage lies in that it often adds too many additional places and arcs to the net, thereby making the resulting net more complex than the originally built one. Deadlock prevention policy based on elementary siphon theory greatly decreases the number of the contorl places and related arcs. However, it is a conservative method, which decreases the resulting nets'permissive behavior by preseting the output arcs of the control places and influences the system's working efficiency finally. This thesis explores a new way to minimize the number of the new addition supervisors while optimize their control efficiency. It can keep the supervisor's structure simple and make the net hold more permissive behaviors. By analysing several different FMS, it shows that our method's advantage is clearer while the net is larger.