Abstract

As systems grow more complex to cater to demanding operational requirements, they tend to suffer from increasing component failures. It is important to minimize the effect of these failures on the overall performance of these systems. In this thesis, fault recovery using discrete event systems theory is studied. It is assumed that the plant can be modeled as a finite state automaton, and that is prone to failures. For this study all events are assumed observable and the extension to the case of partial observation is left for future research. The problem of the synthesis of fault recovery procedures is studied. In particular, the cases are studied in which the plant may return to normal operation. This could be either because the failures are intermittent or because the plant has the capacity to repair or reset. Both of the above cases are studied in this thesis. It turns out that the problem is an instance of the problem of robust nonblocking supervisory control for countably infinite number of plants. The objective of the thesis is to obtain maximally permissive solution for the above problem. It is shown that the desired supervisor can be obtained as the maximally permissive solution of a robust control problem involving a bounded number of plants. Furthermore, an iterative procedure is provided to solve the original problem involving an infinite number of plants. The procedure is guaranteed to converge in a bounded number of steps. Several examples are provided to illustrate the proposed procedures