Kybernetika 28 no. 3, 169-180, 1992

On least squares estimation in continuous time linear stochastic systems

Tyrone E. Duncan, Petr Mandl and Bożenna Pasik-Duncan

Abstract:

The sufficient conditions for the convergence of a family of least squares estimates of some unknown parameters are given. The unknown parameters appear affinely in the linear transformations of the state and the control in a linear stochastic system. If the noise in the stochastic system is colored then the family of least squares estimates does not converge to the value and the bias is given explicitly.

Classification:

60G35, 60H10, 93E12, 62M09, 93E10, 93E24