Recursive least squares estimation with rank two updates

Abstract

This paper presents new recursive least squares (RLS) algorithms with enhanced performance, achieved via a combination of exponential forgetting and windowing techniques. The proposed algorithms with rank two updates are systematically aligned with established RLS algorithms with rank one updates to ensure unification and clarity. Newly identified properties of the recursive algorithms, associated with the convergence of both the inverse of the information matrix and the parameter estimates which are presented in this paper, offer great potential for further enhancement of the estimation performance. The proposed algorithms demonstrate significant improvements in the estimation of the grid events in the presence of substantial harmonic emissions.