Robustness Analysis and Control for Automated Manufacturing Systems With Unobservable Events Using Petri Nets
针对自动化制造系统中资源故障导致的部分可观测问题，提出用Petri网分析系统状态鲁棒性，并设计控制器确保连续生产。
Resource failures are commonly viewed in automated manufacturing systems (AMSs), such as control errors and mechanical failures. Therefore, a robustness analysis and control policy are highly desirable. Different methods are proposed by researchers and practitioners. However, they are incapable of performing robustness analysis for partially observable AMSs since few works consider the existence of unobservable events. In this article, we solve the robustness analysis and control problem for partially observed AMSs in the paradigm of Petri nets (PNs). A marking is denoted as a robust (resp., nonrobust) one if from it the system can (resp., cannot) proceed continuously even though there exist resource failures. We show that the robustness of markings under partial observation can be determined using a compact reachability graph (RG), namely, a remarkably reduced RG. Then, we define robustness observability, and propose a simple method to check the robustness observability with respect to a given set of robust markings. Finally, we design robust controllers under partial observation so as to ensure continuous production of partially observable AMSs against resource failures.