Non-Biased Recursive Subspace Identification Algorithms

Abstract

In this paper two recursive algorithms are presented, based on NB-N4SIDsubspace identification algorithm. The main idea was to show that subspace identificationmethods can be represented as sequences of least squares problems and implementedthrough sequences of modified Householder algorithms. Therefore, it is possible todevelop iterative algorithms with most of the advantages of this kind of methods, andstill improve the numerical efficiency, in order to deal with real-time applications andminimize the computational burden.