PLS algorithm for radial basis function networks

Author

Wang, Yin ; Rong, Gang ; Wang, Shuqing

Author_Institution

Nat. Key Lab. of Ind. Control Technol., Zhejiang Univ., Hangzhou, China

Volume

4

fYear

1998

fDate

16-18 Dec 1998

Firstpage

4748

Abstract

The PLS (partial least squares) algorithm is introduced into the radial basis function (RBF) networks to construct MIMO nonlinear models. The PLS algorithm projects the correlated basis functions and the outputs down to a number of principal factors to construct parsimonious models. Example of nonlinear modeling is used to demonstrate better generalization and noise tolerable performance of the proposed algorithm than the full RBF network models. Finally, it is used as a product quality predictor for industrial distillation columns

Keywords

MIMO systems; correlation methods; least squares approximations; modelling; nonlinear systems; radial basis function networks; MIMO nonlinear models; PLS algorithm; RBF networks; correlated basis functions; industrial distillation columns; nonlinear modeling; parsimonious model construction; partial least squares algorithm; product quality predictor; radial basis function networks; Distillation equipment; Industrial control; Laboratories; Least squares methods; Linear regression; MIMO; Predictive models; Radial basis function networks; Training data; Vectors;

fLanguage

English

Publisher

ieee

Conference_Titel

Decision and Control, 1998. Proceedings of the 37th IEEE Conference on

Conference_Location

Tampa, FL

ISSN

0191-2216

Print_ISBN

0-7803-4394-8

Type

conf

DOI

10.1109/CDC.1998.762085

Filename

762085