Logo image
Nonlinear system identification using the recursive weighting LASSO method
Thesis

Nonlinear system identification using the recursive weighting LASSO method

Colbin Eric Erdahl
University of Iowa
Master of Science (MS), University of Iowa
Autumn 2007
pdf
t2007_e658-o4.99 MB
Delayed release, Embargo ends: 01/01/2099

Abstract

In the area of nonlinear control, system identification is often the first step. Parametric nonlinear system identification is often separated into the two problems of regressor variable selection and parameter estimation. The least absolute shrinkage and selection operator (LASSO) method attempts to combine regressor selection and parameter estimation into a single problem by taking the coefficients of the regressors to exactly zero if they do not appear in the true system. This thesis proposes a recursive weighting LASSO method, which is shown to take these coefficients to zero under certain conditions. The recursive weighting LASSO is compared to the least-squares method, the LASSO method, and the adaptive weighting LASSO method, and found to perform better in many cases.

Details

Metrics

1 File views/ downloads
4 Record Views
Logo image