Time‐varying threshold‐based fault detection for a class of uncertain non‐linear systems in strict‐feedback form

Abstract

This study presents a time-varying threshold-based fault detection (FD) method for a class of single-input single-output non-linear systems in strict-feedback form with unmeasurable states, unmatched non-linear fault functions and uncertainties. First, a residual generator with prescribed performance is designed in a framework of recursive technique. Then, the tight time-varying thresholds are computed by using the prescribed performance bound (PPB) and the FD scheme is presented. Furthermore, the fault detectability is analysed to characterise the sensitivity to the faults. It is shown that the residual signal satisfies the PPB and all the estimation errors are uniformly bounded before fault occurrence. Compared with the method without considering the PPB, the proposed time-varying thresholds improve the FD performance with fewer false alarms caused by overshoot during the transient process. Finally, a comparative simulation study illustrates the superiority of the proposed method.