Nuclear Norm Subspace Identification Of Continuous Time State-Space Models

Abstract

Subspace identification techniques derive approximate models rather than models that are optimal with respect to a goodness of fit criterion. To obtain low rank models, a nuclear norm minimization method for estimating the system matrices of linear time invariant continuous time state-space models in the presence of measurement noise is proposed. In the proposed approach, Generalized Poisson Moment Functional (GPMF) method is used to circumvent the time-derivative problem which is inherent in continuous time models. To make the proposed algorithm consistent, instrumental variables (Hankel matrix of past inputs) are considered. The accuracy of the proposed method is demonstrated with the help of numerical simulations on a variety of systems.