Fault-tolerant supervisory control with permanent faults

In our earlier work, we introduced a discrete-event system-based fault-tolerance approach designed to handle intermittent faults. This approach is different from the typical fault-tolerant methodology as the approach does not rely on detecting faults and switching to a new supervisor; it requires a supervisor to be designed that works correctly under normal and fault conditions. This is a passive approach that relies upon inherent redundancy in the system being controlled. This is also a foundation method that should allow a wide variety of existing fault approaches to be modelled but still allow controllability and nonblocking properties to be verified. Permanent faults could be modelled in this framework, but the current method was onerous. In this paper, we introduce a new modelling approach for permanent faults that is easy to use, as well as a set of new permanent fault-tolerant definitions. They are designed to capture several types of permanent fault scenarios (generic situations such as at most one fault occurs) and to ensure that our system remains controllable and nonblocking in each scenario. New definitions and scenarios were required as the previous ones were incompatible with the new permanent fault modelling approach. Finally, we present algorithms to verify these properties, followed by complexity analyses and correctness proofs of the algorithms. An example is then provided to illustrate our approach.