Event-Triggered Fault Detection and Isolation of Discrete-Time Systems Based on Geometric Technique

Abstract

This brief considers the problem of fault detection and isolation using the geometric approach for discrete-time systems. The description of unobservable subspace is presented. The residuals are generated by the filter via employing the geometric technique so that each residual is influenced by a specific fault and uncoupled from others. Sufficient conditions are established such that the $ {\mathcal {H}}_{\infty }$ norm of the transfer function is less than a pre-given value. In order to reduce computational cost, an event-triggered mechanism is presented to determine whether the current data should be released or not. Finally, numerical demonstration is given to verify the usefulness of the proposed new design technique.