New Recursive Least Square Algorithms without using the initial information
Authors: | Quan Zhonghua, Seoul National University, Korea, Republic of Han Soohee, Seoul National University, Korea, Republic of Kwon Wook Hyun, Seoul National University, Korea, Republic of |
---|
Topic: | 1.1 Modelling, Identification & Signal Processing |
---|
Session: | Recursive Estimation Methods |
---|
Keywords: | Estimation, initial information, RLS, state estimator, numerical stability |
---|
Abstract
In this paper, new types of recursive least square (RLS)algorithms, without using the initial information of a parameteror a state to be estimated, are proposed. The proposed RLSalgorithm is first obtained for a generic linear model and is thenextended to a state estimator for a stochastic state-space model.Compared with the existing algorithms, the proposed RLS algorithmsare simpler and more numerically stable. It is shown, bysimulation studies, that the proposed RLS algorithms have betternumerical stability for digital computation than existingalgorithms.