Observer Based Fault Detection and Identification in Differential Algebraic Equations

Fault detection and identification (FDI) are important tasks in most modern industrial and mechanical systems and processes. Many of these systems are most naturally modeled by differential algebraic equations. One approach to FDI is based on the use of observers and filters to detect and identify faults. The method presented here uses the least squares completion to compute an ODE that contains the solution of the DAE and applies the observer directly to this ODE. Robustness with respect to disturbances is also addressed by a frequency filtering technique.Copyright © 2013 by ASME