Robust fault diagnosis using the model-based approach

Focuses on the state (or output) estimation approach to FDI. In particular, the interest is in the use of robust observers or robust state estimators which have the designed capability of discriminating between effects due to unknown inputs, disturbances and actual faults. There is always a difference between actual process parameters and those of the nominal model used in the observers (or Kalman filters). The effect of parameter variations obscures the effect of fault detection in that it acts like noise as a source of probable false alarms. The objective of this work is to provide an introduction to the subject of fault detection and isolation in dynamic systems so that the reader may know where to turn for definitions and for reference material. The authors provide an extensive historical review of the subject. Finally, they provide the basic mechanism of an FDI scheme based on the use of the so-called General Observer Scheme (GOS) for fault detection. Results are given to show the potential of the method when applied to a (simulated) nonlinear aircraft system