A. A. Lomov
Sobolev Institute of Mathematics
Siberian Branch of the Russian Academy of Sciences
4 Acad. Koptyug avenue,630090 Novosibirsk, Russia
The parameter estimation problem for linear stochastic systems with errors in variables is considered. The methods of the multivariate orthogonal regression type are compared and classified by the ability to take into account an apriory knowledge of linear relations between variables. The way to compare nonlinear estimators by their linear approximations in limiting cases of large samples and small errors is proposed. The application of multivariate orthogonal regression methods to the identification of uncontrollable systems arising in trend separation problem is discussed.