Approximate Evaluation of Reliability and Related Quantities via Perturbation Techniques
Abstract
The evaluation of the reliability, stability, and performance of fault tolerant control systems (FTCS) is considered. New sufficient conditions for stochastic stability of FTCS with standard Markovian component failure behavior and Markovian failure detection decision behavior are derived. By specializing these results to the class of linear time-invariant (LTI) FTCS with linear state feedback control laws that are reconfigured by switching the feedback gain matrix according to the identified failure configuration, the stability results are strengthened to necessary and sufficient conditions for stochastic stability of a special type (exponential in mean square) that implies a very strong sense of stability (a.s. in probability). An approximate feedback control design technique for LTI FTCS is then proposed and demonstrated on a simple numerical case. In addition, previous results on semi-Markov analysis of FTCS reliability are used to derive a numerical method for establishing approximately optimal failure detection test thresholds for sequential failure detection tests. This method, though approximate, is shown to yield thresholds that provide a considerable increase in system reliability relative to those provided by a method based on a rigorously derived reliability approximation for one numerical example.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 1990
- Accession Number
- ADA240049
Entities
People
- Bruce K. Walker
- Ramaswamy Srichander
Organizations
- University of Cincinnati