Affiliations: Department of Industrial Engineering, Amirkabir
University of Technology, Tehran, Iran, P.O. Box 15875-4413. E-mail:
zarandi@cic.aut.ac.ir, smhadian@cic.aut.ac.ir | Department of Mechanical and Industrial Engineering,
University of Toronto, Toronto Ont., Canada. E-mail:
turksen@mie.utoronto.ca
Abstract: In this paper, we propose an indirect method to fuzzy modeling which
implements a clustering algorithm to build a linguistic fuzzy controller. Based
on output data clustering and projection onto the input spaces, the number of
clusters is determined and rules are generated automatically. A new methodology
based on output sensitivity is developed for input variable selection. Then,
implementing an Adapted Neural Network for the selection of membership
functions optimizes all membership function parameters. The unbounded
parameters of fuzzy operators and the inference methods of FATI (First
Aggregate, Then Infer) and FITA (First Infer, Then Aggregate) are optimized
through a simple and efficient tuning strategy.
