A Neuro-Fuzzy Network for Generating Fuzzy Rule-based Nonlinear Dynamic Systems

Título: A Neuro-Fuzzy Network for Generating Fuzzy Rule-based Nonlinear Dynamic Systems

Autores: Ferreira, Antonio Luiz S.; Nascimento, Edson

Resumo: A fuzzy neural network (FNN) architecture for generating fuzzy rule database from sparse knowledge about the system dynamic is proposed and its performance in solving the truck backer-upper nonlinear control problem is analyzed. Since the FNN maps fuzzy input to fuzzy outputs vectors deriving a knowledge kernel for a given specific problem, considering the actual experience of experts in the domain, it may becomes an important tool for handling fuzzy mapping in many practical applications. Further improvements to allow fast convergence of the FNN learning phase, controlling the learning and momentum parameters through an adaptive fuzzy logic controller (FLC), is described. Simulations results are included to confirm that this approach may be of great practical interest since the actual results are very encouraging.

Palavras-chave:

Páginas: 6

Código DOI: 10.21528/CBRN2001-078

Artigo em pdf: 5cbrn_078.pdf

Arquivo BibTex: 5cbrn_078.bib