An Efficient Algorithm for a Bounded Errors-in-Variables Model

From MaRDI portal





The paper is devoted to the problem of parameter estimation in presence of bounded data uncertainties. The considered version of the problem incorporates a priori bounds on the size of the perturbations. It has a ``closed form solution that is obtained by solving an ``indefinite regularized least-square problem with a regression parameter that is determined from the positive root of the secular equation. The solution method proposed in the paper proceeds by first providing a geometric interpretation of the new optimization problem, followed by an algebraic derivation that establishes that the optimal solution can in fact be obtained by solving a related ``indefinite regularized problem. The regression parameter of the regularized step is further shown to be obtained from the positive root of a secular equation. It is shown that several extensions are possible. For example, weighted versions with uncertainties in the weight matrices are useful in several applications, as well as cases, in addition to more discussion on estimation and control problems with bounded uncertainties, can be found.











This page was built for publication: An Efficient Algorithm for a Bounded Errors-in-Variables Model

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4702232)