Robust Control of Automated Manufacturing Systems With Multiple Unreliable Resources Using Extended Structural Objects

Resource failures can be encountered in a practical automated manufacturing system (AMS), and thus, its robust supervisory control is an important issue to be addressed. This work uses Petri nets to model AMS with multiunit and unit-type resource acquisitions and releases. Such a system contains multiple unreliable resources. It is known that perfect activity circuits (PACs) are important structural objects in Petri nets. We define a new one called the maximally extended PAC (MPAC) that can be used to characterize deadlocks as PAC does. The number of MPACs is much smaller than that of PACs. The stalled production of a part of an AMS is caused by the occurrence of a blocking marking (BM) state, where a resource circuit-wait pattern is formulated with respect to MPAC. We use an integer linear program (ILP) to compute a problematic MPAC, and a monitor is designed to prevent all BM states caused by it. Then, a robust controller is synthesized in an iterative manner. Such a controller can be used to guarantee the continuous production of all part types even if some resources fail. Finally, the effectiveness and efficiency of the proposed method are demonstrated via experiments and comparative analysis.