Non-linear system identification using wavelet multiresolution models

From MaRDI portal





The authors propose a new methodology for identifying nonlinear NARMAX models from noise corrupted data based on semi-orthogonal wavelet multiresolution approximations. They introduce an adaptive model sequencing strategy to infer model complexity from the data, while reducing computational costs. Using this strategy in conjunction with an iterative orthogonal-forward-regression routine coupled with model validity tests, they identify sparse but accurate wavelet series representations of nonlinear processes. Finally, they use experimental data from two real systems to illustrate the effectiveness of the proposed identification procedure.



Cites work


Cited in
(26)








This page was built for publication: Non-linear system identification using wavelet multiresolution models

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3151500)