Identification of an Experimental Yo-yo Motion Control System by Evolutionary B-spline Neural Network

Content Provider	Semantic Scholar
Author	Coelho, Leandro Dos Santos
Copyright Year	2008
Abstract	In an attempt to accurately model nonlinear syste ms, a wide variety of techniques have been develope d, such as the Volterra series, Wiener models, Hammerstein models, and others. Such approaches have had limite d success in industry, due primarily to their complexity. Recent ly, artificial neural networks have generated consi derable interest as alternative nonlinear modeling tool. B-spline neura l network (BSNN), a type of basis function neural n etwork, is trained by gradient-based methods, which may fall i nto local minimum during the learning procedure. To overcome the problems encountered by the conventional learni ng methods, differential evolution (DE)  an evolutionary computation methodology  can provide a stochastic search to adjust the cont rol points of a BSNN is proposed. The potentialities of DE are its simple structure, easy u e, convergence property, quality of solution and robustness. In this paper, we propose a DE approach to train a BSNN. Th e numerical results presented here indicate that th e DE is effective in building a good BSNN model for nonline ar identification of an experimental nonlinear yo-y motion control system.
File Format	PDF HTM / HTML
Alternate Webpage(s)	http://www.abcm.org.br/symposium-series/SSM_Vol3/Section_VII_Emerging_Technologies_and_Applications/SSM3_VII_04.pdf
Language	English
Access Restriction	Open
Content Type	Text
Resource Type	Article