Improved robustness and sensor fault tolerance via a generalized internal model-based fault-tolerant controller

Abstract

In this paper the synthesis of an implicit fault-tolerant control system that is capable of handling multiple sensor faults in addition to modelling uncertainties is presented. The implicit approach allows sensor fault tolerance without the need for accurate information from the fault diagnosis element. Difficulties encountered when solving for a fault-tolerant control problem commonly related to the reliability and accuracy of its fault diagnosis element are thus avoided. Based on extensions to the generalized internal model-based controller (GIMC), the new development proposed in this paper includes the introduction of an Hinfin loop-shaping controller in the structure to facilitate best control performance augmented by standard H∞ controllers as a fault-compensating block to facilitate fault tolerance towards multiple sensor fault conditions and robustness towards modelling uncertainties. Crucial stability and robustness analysis/proofs of the proposed design are conveniently obtained. Sensor faults are modelled as functions within a bounded family of signals and an internal model of this family is embedded into the fault compensating H∞ controller which is cascaded to the H∞ loop-shaping controller. The reliability of operation towards modelling uncertainties is also shown to be guaranteed. Its potential for providing multiple sensor fault tolerance in addition to robustness towards modelling uncertainties is illustrated with a flight control system example; results are significant, as observed in a comparison of control performance with standard H∞ and μ controllers.