A measurement analysis approach to online fault detection and isolation for linear discrete-time uncertain systems

This work considers a robust fault detection and isolation (FDI) problem for linear discrete-time systems subject to faults, bounded additive disturbances and norm-bounded uncertainties. We propose a receding horizon estimation procedure (which is a dual scheme to Model Predictive Control (MPC)) to solve FDI problems in which the upper and lower bounds on the faults are computed by using a system model, together with input/output measurements over a finite estimation horizon. Then a fault is regarded as detected and isolated if its obtained upper and lower bounds at some sampling instant are both larger than zero or smaller than zero. Linear Matrix Inequality (LMI) optimization techniques are used to obtain the bounds. Moreover, a subsequent-state-estimation technique, together with an estimation horizon update procedure are given to allow the on-line fault detection and isolation process to be operated in an iterative procedure. Finally, the approach is verified using a numerical example.