Title of article
The residual-based ESG algorithm and its performance analysis
Author/Authors
Xiao، نويسنده , , Yongsong and Wang، نويسنده , , Dongqing and Ding، نويسنده , , Feng، نويسنده ,
Issue Information
روزنامه با شماره پیاپی سال 2010
Pages
12
From page
426
To page
437
Abstract
The performance of the residual-based extended stochastic gradient (ESG) algorithms for identifying CARMA models with disturbances is analyzed under weaker conditions on statistical properties of the noise. The paper derives the conditions under which the parameter estimation errors converge to zero. Three examples are given to show the advantages of the proposed algorithm.
Keywords
Stochastic approximation , Recursive identification , Parameter estimation , Convergence properties , Stochastic gradient
Journal title
Journal of the Franklin Institute
Serial Year
2010
Journal title
Journal of the Franklin Institute
Record number
1543516
Link To Document