Reliability and fault detection techniques of large scale systems a synthesis of the EEC report

Abstract

This report provides a synthesis of the works on reliability, maintenance, fault detection and fault-tolerant controller design carried out by the members of the teams involved in the EEC real time control project. In the reliability area, a repairable standby system of assemblies with a hierarchical structure was studied and optimized, via the generalized minimum principle, by selecting appropriate maintenance control laws. A model of a repairable standby system with repair facilities subject to breakdowns was also studied. In the area of fault detection, an up-to-date survey of the available techniques was made, and a method was developed for detecting sensor failures in large scale systems using decentralized observers designed by the overlapping decomposition technique of Siljak. A similar observer-based failure detection scheme for distributed-parameter systems was also developed and applied to gas transportation and distribution systems. Finally, a fault-tolerant controller scheme was proposed, and tested both by simulation and by building an experimental set-up using Z-80 microprocessors. For completeness, a discussion of some related results concerning the stability and performance of hierarchical controllers, under structural perturbations, is also included.