Convergent Identification Algorithms.
Technical rept. 1 Jun-1 Dec 73,
COLORADO STATE UNIV FORT COLLINS DEPT OF ELECTRICAL ENGINEERING
Pagination or Media Count:
A least-squares algorithm is derived for memoryless system identification. Then a stochastic approximation algorithm is developed for identifying mixed auto regressive moving average ARMA processes. Since the correct auto regressive AR model is in general of infinite order, errors appear in an otherwise consistent estimation procedure. Upper bounds of these errors are developed for the ARMA parameters and for the Kalman-Bucy filter based on these identified parameters. Finally, an adaptive array estimation algorithm is developed for the case of correlated signal and noise fields and shown to converge in mean-square. Modified author abstract
- Statistics and Probability