Fault-Tolerant Control of Dynamical Large Scale Systems

A fault-tolerant design procedure for large scale engineered systems is introduced. The proposed methodology is based on a hybrid approach that capitalizes upon beneficial attributes of conventional systems and control-theoretic techniques, as well as methods of artificial intelligence. The paper focuses on a fault propagation algorithm that assesses quickly the impact of a failure on other healthy components and subsystems. Qualitative reasoning techniques are employed to determine the state transitions. An example from a major subsystem of the space station is used to illustrate the algorithmic developments.