Optimization Of Design Of Deadlock Control Supervisors For Flexible Manufacturing System Based On Nonpure Petri Net

A flexible manufacturing system(FMS)is a computerized control system consisting of a computer numerical control machine and a material transfer system that can efficiently produce small or medium batches of products.In the different production processes of an FMS,competition for resources often results in deadlocks.Stalls and discontinuities of the whole or part of the system,long downtimes and low resource utilization caused by deadlocks are in some cases not just a simple matter of reduced productivity,but can even cause the whole system to crash,with serious economic consequences.There are many tools for studying the deadlock problem,of which Petri net is a well-established one.And there have been quite a few results in deadlock problem using Petri net analysis.Nonpure Petri net structures are more powerful in modeling and control of a system than pure Petri net structures.There are two contributions proposed in this thesis.One is to propose a vector covering approach for t-enabled good markings to reduce the computational complexity for the design of non-pure Petri net controllers,and the other part is to propose a new design method for non-pure Petri net supervisors to simplify the supervisor structures while guaranteeing the maximal permissiveness of the system.1.Covering approach for t-enabled good markings.The vector covering approach is an effective method to reduce the computational effort in the non-pure Petri net supervisor design method.In this thesis,a novel vector covering approach is designed for the t-enabled good markngs to further reduce the computational effort in supervisor design.2.A new method for designing non-pure Petri net supervisors.There are various methods for deadlock prevention.For a non-pure Petri net supervisor,our idea is to disable some specific transitions at some legal markings,and then prevent other legal markings from entering into the deadlock zone by firing those transitions.The supervisor is designed to prohibit as many transitions as possible to ensure that each control place does not only implement its own combination of marking and transition but also as many other other similar combinations of marking and transition as possible do,so that the system can achieve the maximum behavioral permission.Then,a set covering approach is applied to select the designed control places so that the non-pure Petri net supervisor composed of these control places has the minimum number of control places while guaranteeing the maximum behavioral permission.