Hybrid learning for interval type-2 fuzzy logic systems based on orthogonal least-squares and back-propagation methods

  • Authors:

  • Gerardo M. Méndez;M. de los Angeles Hernandez

  • Affiliations:

  • Department of Electrical and Electronic Engineering, Instituto Tecnológico de Nuevo León, Calle Septima 822, Col. La Herradura, 67140 Cd. Guadalupe, N.L., Mexico;Department of Economics and Administration Sciences, Instituto Tecnológico de Nuevo León, Cd. Guadalupe, N.L., Mexico

  • Venue:
  • Information Sciences: an International Journal
  • Year:
  • 2009

Quantified Score

Hi-index 0.07

Visualization

Abstract

This paper presents a novel learning methodology based on a hybrid algorithm for interval type-2 fuzzy logic systems. Since only the back-propagation method has been proposed in the literature for the tuning of both the antecedent and the consequent parameters of type-2 fuzzy logic systems, a hybrid learning algorithm has been developed. The hybrid method uses a recursive orthogonal least-squares method for tuning the consequent parameters and the back-propagation method for tuning the antecedent parameters. Systems were tested for three types of inputs: (a) interval singleton, (b) interval type-1 non-singleton, and (c) interval type-2 non-singleton. Experiments were carried out on the application of hybrid interval type-2 fuzzy logic systems for prediction of the scale breaker entry temperature in a real hot strip mill for three different types of coil. The results proved the feasibility of the systems developed here for scale breaker entry temperature prediction. Comparison with type-1 fuzzy logic systems shows that hybrid learning interval type-2 fuzzy logic systems provide improved performance under the conditions tested.