Nonlinear decoupling approach to fault isolation in linear systems

Abstract

The problem of fault isolation in linear systems is considered. Based on a novel fault representation (multiplicative-like) and on a nonlinear decoupling transformation similar to one used for nonlinear systems, it is shown how the fault isolability can be improved. A further advantage of this approach is that it allows the application of observer-based methods to multiplicative fault representations. The price to be paid is the requirement of a nonlinear decoupling transformation as well as a nonlinear residual generation. Existence conditions as well as a design procedure are provided. An example is used to show the application of the above ideas.