Design Of Robust Optimization Controller For Petri Nets In Automated Manufacturing Systems

There have been a variety of deadlock control strategies proposed for automated manufac-turing systems(AMSs)based on Petri nets(PNs).These strategies often assume that the resources in AMSs are reliable.However,unpredictable resource failures may occur in real-world systems.In such a case,the existing deadlock control strategies become ineffective.In order to solve this problem,this thesis proposes a robust deadlock control strategy based on the reachability graph for generalized systems of simple sequential processes with resources(GS~3PRs)and manufacturing-oriented PNs(M-nets).The robust controllers obtained by this strategy can force the AMSs work normally even if unreliable resources break down or not.For a GS~3PR with unreliable resources,this thesis borrows PNs to model the resource failures,adds workpiece/resource separation recovery subnet to the places where resource failures may occur.The obtained PN is called an unreliable GS~3PR(U-GS~3PR).A work-piece/resource separation recovery subnet is used to model the separation of failure resource units and the corresponding workpiece,the recovery processes of resources and if there are enough available units of the unreliable resources,the workpiece is returned to the original system for further processing.The reachable markings are divided into robust legal mark-ings and forbidden markings according to the relationships among the number of broken resources,enabled transitions and the controllability of transitions.The robust controllers of U-GS~3PRs can be designed by-invariants.Note that the robust legal reachability spaces computed in some U-GS~3PRs may be non-convex.Hence,for such a system,a robust con-troller cannot be synthesized by the conjunctions of linear constraints.Interval inhibitor arcs are used to design the robust controllers of U-GS~3PRs.Subsequently,for M-nets with multiple parallel processing processes competing for limit-ed resources,in view of the structural characteristics of M-nets with unreliable resources,this thesis defines the workpiece/resource separation and return reliable resource recovery subnet and adds it to the places where resource failures may occur.The obtained PNs are called unreliable M-nets(U-M-nets).After classifying all reachable states,the robust con-trollers of U-M-nets can be obtained based on-invariants and interval inhibitor arcs.Due to the addition of recovery subnets,the U-M-nets structure becomes complex,leading to an increase in reachable states,and high computational complexity of the algorithm for design-ing robust controllers.In order to solve this problem,this thesis simplifies the structure of the U-M-nets,and uses the above-mentioned methods to design the robust controllers of the simplified net.The results demonstrate that the simplified method achieves the same control effect as before the simplification by calculating fewer reachable states.This thesis takes GS~3PRs and M-nets as examples to deal with the resource failures issue and provides a solution by designing robust liveness-enforcing controllers for AMSs based on the reachability graphs.Finally,the correctness of the proposed algorithms are verified by examples and the results show that the above methods can make AMSs work normally no matter whether unreliable resources break down or not.