The paper by Christopher Edwards and Chee Pin Tan represents an interesting contribution to the important area of the fault diagnosis of dynamic systems. The work under discussion compares a diagnosis method based on sliding mode observers for fault reconstruction [1] with the linear technique relying on unknown input observers (UIOs) [2].Bot h the schemes described in the mentioned references exploit their robustness properties to overcome uncertainty acting on the system under diagnosis.In particular, Tan and Edwards in Ref.[1] suggested an observer design method, based on the solution of a linear matrix inequality (LMI), which seeks to minimize the L2 gain between the uncertainty and the fault reconstruction signal.In terms of the UIO literature, the work of Saif and Guan has been chosen because the problem it tackles is the closest to the one associated with the sliding mode scheme and because the assumptions in Ref.[2] are similar to those from Ref.[1].On the basis of this hypothesis and using also the results achieved from other related works by the same authors, the links between the two approaches are recalled and investigated, even if the details of the methods are quite different.These methods are finally compared by exploiting a nonlinear dynamic model of a crane system affected by simulated faults.
Discussion on "A comparison of sliding mode and unknown input observers for fault reconstruction" by Christopher Edwards and Chee Pin Tan
SIMANI, Silvio;BONFE', Marcello
2006
Abstract
The paper by Christopher Edwards and Chee Pin Tan represents an interesting contribution to the important area of the fault diagnosis of dynamic systems. The work under discussion compares a diagnosis method based on sliding mode observers for fault reconstruction [1] with the linear technique relying on unknown input observers (UIOs) [2].Bot h the schemes described in the mentioned references exploit their robustness properties to overcome uncertainty acting on the system under diagnosis.In particular, Tan and Edwards in Ref.[1] suggested an observer design method, based on the solution of a linear matrix inequality (LMI), which seeks to minimize the L2 gain between the uncertainty and the fault reconstruction signal.In terms of the UIO literature, the work of Saif and Guan has been chosen because the problem it tackles is the closest to the one associated with the sliding mode scheme and because the assumptions in Ref.[2] are similar to those from Ref.[1].On the basis of this hypothesis and using also the results achieved from other related works by the same authors, the links between the two approaches are recalled and investigated, even if the details of the methods are quite different.These methods are finally compared by exploiting a nonlinear dynamic model of a crane system affected by simulated faults.I documenti in SFERA sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
