A Framework for Control-Reconfiguration Following Fault-Detection in Discrete Event Systems

Abstract

We introduce a framework for control-reconfiguration in discrete event systems that performs fault-detection and then reconfigures the control to ensure that certain desired specifications are met. The controlled system has separate specifications prior to a fault versus after a fault (to allow for any changes in the specification owing to the occurrence of a fault). Prior to the occurrence of a fault, a certain controller is applied so as to ensure a given pre-fault specification. The same controller is continued to be used even after a fault and until its detection. Once a fault is detected, the control is reconfigured. All along the post-fault duration (prior to as well as after reconfiguration), the controlled system is required to satisfy a give post-fault specification. We mathematically formulate the above control problem and provide a condition for reconfigurability. We also establish the condition for the limiting case where control-reconfiguration is not required. Also, we introduce the notion of degree of disambiguability of faulty traces which when increased, increases the ability to meet the control requirements of pre- and post- fault specifications.