Least-squares model-reference adaptive control with high-order parameter tuners

Ramon R. Costa

DOI: https://doi.org/10.1016/j.automatica.2024.111544

IF: 6.4

2024-02-20

Automatica

Abstract:Recently, a Lyapunov design of least-squares model-reference adaptive control (LS-MRAC) has been proposed. Unlike other least-squares based algorithms, the update law is driven only by the tracking error and, in this way, it does not require normalization or parameter projection to assure stability of the closed-loop system. Moreover, the algorithm has striking improved tracking error transient and superior parameter convergence properties when compared with a conventional MRAC. Also recently, a new design employing high-order parameter tuners has been proposed. In this design, the parameter derivatives are quite easily and systematically obtained. A key point of the procedure is the Lyapunov function used for the analysis that explicitly considers the derivatives of the parameter. The implementation of the algorithm requires finding a symmetric positive definite matrix such that a Schur complement is negative definite. This is a fundamental property to assure its stability. The solution to this problem leads to a novel simple proof for the equivalence between the Routh–Hurwitz stability criterion and the Lyapunov's method. In this work both design methodologies are combined, that is, the LS-MRAC algorithm is redesigned with high-order tuners.

automation & control systems,engineering, electrical & electronic