An LFT approach to parameter estimation

Authors:
Kenneth Hsu;Tyrone Vincent;Greg Wolodkin;Sundeep Rangan;Kameshwar Poolla
Affiliations:
Department of Mechanical Engineering, University of California, Berkeley, CA 94720, USA;Division of Engineering, Colorado School of Mines, Golden, CO 80401, USA;The Mathworks Inc., Natick, MA 01760, USA;Qualcomm Technologies, Bedminster, NJ 07921-2608, USA;Department of Mechanical Engineering, University of California, Berkeley, CA 94720, USA
Venue:
Automatica (Journal of IFAC)
Year:
2008

Citing 4
Cited 0

System identification: theory for the user

System identification: theory for the user
The complex structured singular value

Automatica (Journal of IFAC) - Special issue on robust control
Robust and optimal control

Robust and optimal control
An analysis of the parametrization by data driven local coordinates for multivariable linear systems

Automatica (Journal of IFAC)

Quantified Score

Hi-index	22.14

Visualization

Abstract

In this paper we consider a unified framework for parameter estimation problems. Under this framework, the unknown parameters appear in a linear fractional transformation (LFT). A key advantage of the LFT problem formulation is that it allows us to efficiently compute gradients, Hessians, and Gauss-Newton directions for general parameter estimation problems without resorting to inefficient finite-difference approximations. The generality of this approach also allows us to consider issues such as identifiability, persistence of excitation, and convergence for a large class of model structures under a single unified framework.