On adaptive threshold intervals for stopping recursive least squares in the space of parameters

Article type: Research Article

Authors: Pupeikis, Rimantas

Affiliations: Institute of Mathematics and Informatics, Akademijos St. 4, 2600 Vilnius, Lithuania

Abstract: In the papers (Kaminskas, 1972; Kaminskas and Nemura, 1975; Yin, 1989) the stopping rules of recursive least squares (RLS) are worked out using the ellipsoidal confidence region for the respective parameter vector of a linear dynamic system. The aim of the given paper is the development of the technique for calculating threshold intervals of respective criterions, used in a stopping rule, which are presented in Kaminskas (1972). In this connection adaptive threshold intervals based on the Cramer-Rao lower bound are proposed here. The results of numerical simulation by IBM PC/AT are given.

Keywords: dynamic system, recursive algorithm, threshold, least squares

DOI: 10.3233/INF-1995-6305

Journal: Informatica, vol. 6, no. 3, pp. 299-312, 1995