This paper proposes a new recursive least squares adaptive algorithm, called the variable memory length (VML) algorithm. The new algorithm is robust in system identification problems in which the input power can be significantly reduced during operation. Most RLS-type algorithms tend to increase the error in the estimated weight vector in such situations. The VML algorithm keeps the mean square deviation of the weight unchanged during the absence of signal power. It should encounter application in systems such as automotive suspension fault detection and system identification using speech signals. In both cases, considerable periods of low input power during operation are common.
A recursive least squares algorithm robust to low-power excitation
2004
4 Seiten, 11 Quellen
Conference paper
English
Robust recursive least squares algorithm for automotive suspension identification
SAE Technical Papers | 2005
|Recursive Least-Squares Filtering
AIAA | 2015
|Recursive Least-Squares Filtering
AIAA | 2009
|Recursive Least-Squares Filtering
AIAA | 2005
|