Robust control of Backward Conflict Free Systems with Resources using Petri nets.

There are a plenty of deadlock-free supervisory control algorithms to address deadlock problems via resource allocation in automated manufacturing systems (AMSs). In most of such work, prerequisites are assumed that allocated resources never fail. Apparently, these designs do not conform to the reality completely. An AMS is composed of many numerically-controlled machines interacting with each other. It is even easier for resources to fail with increasingly complex system structures, e.g., disassembly/assembly operations and flexible routes. In our previous work, we develop a robust deadlock-free supervisory control policy to ensure robust deadlock-freeness for systems with simple structures. In this paper, a more complex system, i.e., Backward Conflict Free Nets with Resources (BCF-R), is proposed. Through dynamic analysis and real-time judgement, our robust supervisor automatically allocates resources so that processes stopped by unexpected resource failures do not block the production of parts not necessarily requiring failed resources at any time. The approach is achieved in a distributed and dynamic way and requires no state enumeration. In the meanwhile, our supervisor can improve the system performance in terms of a higher concurrency.