Robust adaptive regulation with minimal prior knowledge

Abstract

A robust adaptive regulator is designed under minimal prior knowledge, namely, the order of a nominal plant model (in a well defined sense). The involved robustness means that the closed-loop stability is guaranteed in the presence of some class of time-varying parameters and unmodeled dynamics. The main design feature of the proposed adaptive regulator consists in its self-excitation capability together with an appropriate identification-stabilization time splitting. Unlike in the available literature concerning robust adaptive control, the involved self-excitation is established independently of the parameter estimator properties. The use of the robustness-oriented modifications as deadzone and parameter projection or contraction, which are aimed at achieving the parameter estimates boundedness, is no longer required beforehand, thereby reducing the underlying prior knowledge. >