Using an information point of view, we discuss deterministic versus st
ochastic tools for residual generation and evaluation for fault detect
ion and isolation (FDI) in linear time-invariant (LTI) state-space sys
tems. In both types of approaches to off-line FDI, residual generation
can be viewed as the design of a linear transformation of a Gaussian
vector (the finite-window input-adjusted observations). Several statis
tical isolation methods are revisited, using both a linear-transform f
ormulation and the information content of the corresponding residuals.
We formally state several multiple-fault cases, with or without causa
lity assumptions, and discuss an optimality criterion for the most gen
eral one. New information criteria are proposed for investigating the
residual optimization problem. (C) 1997 Elsevier Science Ltd.