A Deadbeat Estimator-Based Fault Isolation Scheme for Nonlinear Systems

This paper deals with a fault isolation scheme based on a deadbeat estimation methodology for abrupt faults occurring in nonlinear uncertain dynamic systems. In principle, the deadbeat estimator allows the parameter estimates to converge within an arbitrarily small finite-time for a class of nonlinear system with full-state measurements. The corresponding adaptive isolation thresholds are designed based on an a priori known bound on the system uncertainty. A fault isolation decision is made if the residual associated with the matched isolation estimator remains below its corresponding adaptive threshold, whereas at least one of the components of the residuals associated with all the other estimators exceeds its threshold at some finite time. Fault isolability conditions are illustrated and simulation trials are given to assess the expected improvement of the fault isolation methodology in terms of fault isolation time.