Decomposition Control Based On The Shared Resources Of Petri Net

The occurrence of deadlocks in an automatic manufacturing system may reduces the productivity of the system and even make some irreversible results.Therefore,the research on the deadlock problem is necessary.Petri net can be effectively used to model and analysis deadlock for automatic manufacturing system.Hence it is widely used.The computational complexity is a problem that needs to be faced when designing a deadlock control strategy.This thesis proposes a Petri net decomposition control method based on shared resources,which can effectively solve the problem of computational complexity.First,decompose a larger net model into two subnets by sharing resources,and design a liveness-enforcing Petri net supervisor for each subnet by using existing deadlock prevention strategies.Then merge the two controlled subnets and analyze the cause of the deadlocks in the merged net.Finally,design a liveness-enforcing Petri net supervisor for the merged net.Since the method only needs to calculate the reachability states of the subnets,thus the amount of calculation can be reduced.The main works in this thesis are:1.Discover the causing of deadlocks of the merged net when each subnet is controlled by the existing deadlock prevention strategies.By analyzing all the states of the two controlled subnets,we know that some states will disabled the shared transitions in the merged net.In view of this situation,a method to avoid deadlock is given.2.Only if the token number in shared places are consistent,these states can appear in the merged net.When subnets are merged,they are paired according to the shared places to avoid unnecessary calculations.In addition,the vector covering method can effectively reduce the number of special states that need to be considered.It can also reduce the number of added control places and reduce calculation complexity.3.Through in-depth analysis of the combination of the special states and non-special states of the two controlled subnets,the control strategy is further improved by replacing some inapproprite control places.The improved control strategy can obtain a liveness-enforcing Petri net supervisor with more permissive behavior.4.Multi-level decomposition is used to decompose a net model into several sufficiently small subnets.Since the size of the reachability graph is exponentially related to the size of the Petri net,the size of the reachability graph of each subnet is much smaller than that of the entire net.Thus,it can reduce the amount of calculation.