Title :
On the use of compensated total least squares in system identification
Author :
Vandersteen, Gerd
Abstract :
A bias-compensated total least squares estimator is proposed for models which are linear-in-the-parameters and nonlinear in the perturbed input and output signals. The estimates are obtained using a noniterative and numerically stable method. Strong consistency of both the generalized total least squares and the compensated total least squares estimates are proven. The theoretical properties are verified through simulations.
Keywords :
Additive noise; Automatic control; Control systems; Least squares approximation; Least squares methods; Polynomials; State-space methods; System identification; Generalized total least squares;
Journal_Title :
Automatic Control, IEEE Transactions on
DOI :
10.1109/TAC.1998.720868
