It is assumed that the system matrices of a stationary linear dynamical system were parametrized by a set of unknown parameters. The question considered here is, when can such a set of unknown parameters be identified from the observed data? Conditions for the local identifiability of a parametrization are derived in three situations: (1) when input/output observations are made, (2) when there exists an unknown feedback matrix in the system and (3) when the system is assumed to be driven by white noise and only output observations are made. Also a sufficient condition for global identifiability is derived.
Parameter identifiability of linear dynamical systems
1974-04-01
Conference paper
No indication
English
Near Identifiability of Dynamical Systems
NTRS | 1987
|Parameter identifiability of the IAWQ Model No. 1
British Library Conference Proceedings | 1996
|