Maximal Permissiveness Analysis For Petri Net Systems With Uncontrollable Unobservable Transitions And Unreliable Resources

In automated manufacturing systems,the problem of deadlock and system robustness have always concerned researchers.In the automated manufacturing system modeled by Petri nets,there are two kinds of deadlock analysis: structure analysis and reachability graph analysis.Structure analysis is mainly through adding a controller,so that the strict minimum siphon is not empty to achieve deadlock control.However most of the existing siphon-based control strategies are often not maximally permissive.The global reachability graph analysis technology based on Petri net model has great advantages in considering the behavior permissiveness of the whole system because it can obtain the global information of the whole system,but it has the problem of state explosion.In a typical automated manufacturing system,uncontrollable events and unobservable events are inevitable due to a series of reasons such as machine aging,observer performance and equipment properties,etc.Therefore,based on the Petri net model of automated manufacturing systems,it is necessary to consider the maximally permissive behavior analysis strategy for systems with uncontrollable and unobservable transitions so as to pave the way for the design of controller.In addition,resource failure is a very common phenomenon in automated manufacturing system,such as sensor failure,workpiece damage,electrical failure and signal loss.Once the unreliable resource fails,the process that needs to use the failed resource in the subsequent processing path of the process will not be able to proceed,causing other processes that do not use the failed resource to be blocked,and even leading to the stagnation of the whole system.Therefore,a robust control strategy is needed to ensure the operation efficiency of the system.Based on the current deficiencies on the maximally permissive behavior of automated manufacturing systems,this thesis is dedicated to solving the following problems:1.The current research on permissiveness of controlled systems mostly considers that the transitions of the system are fully controllable.Some proposed control strategies consider the impact of uncontrollable transitions whereas they are not maximally permissive.This thesis considers the influence of uncontrollable transitions on the permissiveness of the automated manufacturing system.Based on the partition of the system's global reachability graph,an algorithm is proposed,which starts from the forbidden marking and deduces the others backward.Due to the influence of uncontrollable transitions,the system's permissive marking will inevitably move to a bad marking or deadlock marking,which will cause the original permissive marking to become a bad marking,and the permissive behavior will be reduced.In the end,the maximum permissive behavior analysis strategy of the automated manufacturing system is realized when there are uncontrollable transitions.2.In the actual automated manufacturing system,because the observer cannot observe the occurrence of the event,it leads to unobservable events.This is a real existence in actual production and life,thus deserving further investigation.Therefore,in the Petri net model of the system,based on the analysis of all reachable markings in the global reachability graph,and considering the impact of unobservable transitions on the permission of system behavior,a system's maximal permissiveness analysis algorithm is proposed.The main innovation of this algorithm is to re-plan the global reachability graph when considering unobservable changes to analyze the maximal permissiveness of the system.Furthermore,this thesis considers the influence of uncontrollable and unobservable transitions at the same time,and proposes the maximal permissive analysis strategy of automated manufacturing system.3.At present,robust and deadlock-free control of automated manufacturing systems based on unreliable resources has been widely studied,but the control algorithms proposed by most scholars are not optimal in terms of permissiveness,i.e.,they cannot satisfy the maximally permissive behavior.This thesis proposes a marking robustness detection algorithm based on the mathematical programming algorithm,which formalizes the definition of unreliable resource failures,and can quickly diagnose the robustness of a given marking.Based on this analysis,one can consider the overall analysis of the maximally permissive behavior of robust and deadlock-free control.