Title :
On the design of a stable adaptive filter
Abstract :
A problem of designing a stable adaptive filter (AF) for generating a state estimate in high dimensional systems is addressed. The procedure is essentially based on imposing additional constraints on the location of the eigenvalues of the filter´s transition matrix and on minimizing the prediction error. It is shown that under mild conditions, detectability is a necessary and sufficient condition for stability of the AF. Different simple structures for the gain matrix and suitable parametrizations for the parameters to be adjusted in the filter gain are proposed in order to satisfy the imposed constraints. Numerical results are presented which show the effectiveness of the proposed approach
Keywords :
adaptive filters; eigenvalues and eigenfunctions; filtering theory; matrix algebra; multidimensional systems; prediction theory; stability; state estimation; constraints; detectability; eigenvalues; gain matrix; high dimensional systems; necessary condition; prediction error; stability; stable adaptive filter; state estimation; sufficient condition; transition matrix; Adaptive filters; Adaptive systems; Data assimilation; Eigenvalues and eigenfunctions; Equations; Filtering; Numerical models; Stability; State estimation; Sufficient conditions;
Conference_Titel :
Decision and Control, 1996., Proceedings of the 35th IEEE Conference on
Conference_Location :
Kobe
Print_ISBN :
0-7803-3590-2
DOI :
10.1109/CDC.1996.573721
