Observer-based fault detection and diagnosis scheme using optimal FIR filters

This paper proposes an observer-based fault detection and diagnosis scheme for a continuous-time dynamic system with multiple sensors. The optimal finite impulse response (FIR) filter is adopted as an observer. The proposed scheme consists of the main optimal FIR filter and a bank of additional optimal FIR filters whose number is that of observations. We show that the proposed scheme can maintain the online computational burden at a lower level. The fault detection is performed by a test variable which is formulated by the estimation difference. It is shown that the exact diagnosis of location and time of fault occurrence is achieved within short delay time for any set of threshold values. The proposed scheme can also be applied even multiple simultaneous sensor faults as well as single one. The numerical examples are presented to illustrate the capability of the proposed scheme on a vertical takeoff and landing aircraft system