A neural network approach of input-output linearization of affine nonlinear systems
Resource
American Control Conference, 1994
Journal
American Control Conference
Pages
-
Date Issued
1994-07
Date
1994-07
Author(s)
Lin, Wei-Song
Shue, Hong-Yue
Wang, Chi-Hsiang
DOI
N/A
Abstract
For practical reasons, in the technique of feedback linearization, the requirements of mathematical modeling and access of internal states of complicated nonlinear systems should be removed. This paper demonstrates that, simply using output feedback, the input-output linearization of affine nonlinear systems with zero dynamics being exponentially stable can be accomplished by using multilayer neural network to estimate the instantaneous values of the nonlinear terms appearing in the feedback linearizing control law. Neither mathematical model nor internal state of the nonlinear system is required. The configuration for training the multilayer neural network as a device of the input-output linearizing controller is established. An example of affine nonlinear system is studied by computer simulations for various cases linearizing control.
SDGs
Type
journal article
File(s)![Thumbnail Image]()
Loading...
Name
00735106.pdf
Size
529.26 KB
Format
Adobe PDF
Checksum
(MD5):8009aaad3409db37fd0b1eed48abfa5c