Advanced search
Start date

A new approach to the estimation of multivariable linear models


The identification of multivariable linear systems using only input and output data is a subject of great interest in various areas of knowledge. The parameterization of dynamic models for systems with multiple inputs and multiple outputs is considered to be one of the most important steps in the identification procedure of this class of systems. This research project aims at developing and evaluating a new identification method for linear systems represented in the state space. Such approach is based on a canonical representation, which allows the estimation of the model parameters through linear least-squares methods. Because of the properties of this approach, it is avoided issues that often severely degrade the quality of the estimates, such as: the excess of parameters, the obtainment of extraneous unstable modes and the necessity of nonlinear estimation algorithms. With this, it is expected to develop a numerically robust identification algorithm that provides accurate models for a variety of applications. For evaluation purposes it is intended to compare the proposed technique with algorithms based on the prediction error method and on the subspace approach. Two test platforms are planned: a twin rotor system and a control moment gyroscope. (AU)

Articles published in Agência FAPESP Newsletter about the research grant:
Articles published in other media outlets (0 total):
More itemsLess items

Scientific publications
(References retrieved automatically from Web of Science and SciELO through information on FAPESP grants and their corresponding numbers as mentioned in the publications by the authors)
ROMANO, RODRIGO ALVITE; PAIT, FELIPE. Matchable-Observable Linear Models and Direct Filter Tuning: An Approach to Multivariable Identification. IEEE Transactions on Automatic Control, v. 62, n. 5, p. 2180-2193, . (12/03719-3)

Please report errors in scientific publications list by writing to: