Model based fault detection and diagnosis of LTI systems using likelihood ratio tests.

摘要

This study aims to develop a new framework to detect, isolate and estimate the magnitude of additive faults that occur in linear time invariant (LTI) systems. This study starts with introduction of a new framework to deal with additive step type faults to improve the shortcomings attributed to detection of time of occurrence of the fault (TOF) in existing methods. In the next step, a Bayesian approach is used to decouple detection and isolation phases from the estimation of fault magnitude. The final part of this study is dedicated to investigation of more realistic ramp and truncated ramp type faults.