Title :
Asymptotic gain and search direction for recursive identification algorithms
Author_Institution :
Link??ping University, Link??ping, Sweden
Abstract :
Recursive identification algorithms contain a number of "tuning parameters" to be chosen by the user. Two important such choices are the search direction (the direction in which the estimates are updated) and the gain sequence (the step length). In this paper a family of recursive (prediction error) identification algorithms is considered. The asymptotic distribution of the obtained estimates is derived. It is shown that a gain sequence decaying as 1/t and the Gauss-Newton search direction yields optimal asymptotic accuracy (meeting the Cram??r-Rao theoretical lower bound). It is also shown that these are essentially the only asymptotic choices of direction and gains that give this optimal accuracy.
Keywords :
Algorithm design and analysis; Convergence; Gain measurement; Least squares methods; Newton method; Predictive models; Recursive estimation; Variable speed drives; Vectors;
Conference_Titel :
Decision and Control including the Symposium on Adaptive Processes, 1980 19th IEEE Conference on
Conference_Location :
Albuquerque, NM, USA
DOI :
10.1109/CDC.1980.271947