Parameter information content of measurable signals in direct adaptive control

Abstract

It is postulated that available parameter information is a functional of all past history of measurable signals. From this point of view, direct model reference adaptive controllers become functional differential equations. A modified model reference adaptive control structure and functional adaptation algorithm are presented. It is shown that the system is well-defined, stable, and achieves asymptotic model transfer function matching under weaker excitation conditions than those required by earlier model reference adaptive systems. It is claimed that the information content perspective developed in this paper has implications in robust adaptive control theory.