Observer-based fault detection and isolation: Robustness and applications

This paper studies the observer-based fault detection and isolation problem with an emphasis on robustness and applications. After introducing some basic definitions, the problem of model-based fault detection and isolation is introduced. This is followed by a summary of the basic ideas behind the use of observers in generating diagnostic residual signals. The robustness issues are then defined and ideas for improving the robustness properties are outlined. This provides an opportunity to give some of the fundamental ideas behind eigenstructure assignment and the unknown input observer approaches for enhancing robustness. An important focus is on the use of disturbance principles for robust fault diagnosis. The goal of the paper is to show how robustness techniques must be used for real applications. Copyright (C) 1997 Elsevier Science Ltd.