Intégration de la méthode de Taguchi et d'un modèle inverse d'un système d'interférence neuro-flou adaptatif à multiples entrées, multiples sorties pour la conception optimale d'un condenseur à eau.

Integration of Taguchi's method and multiple-input, multiple-output ANFIS inverse model for the optimal design of a water-cooled condenser.

Auteurs : HUANG C. N., YU C. C.

Type d'article : Article

Résumé

The design of water-cooled condensers is a typical nonlinear and multiple-input, multiple-output (MIMO) problem; to achieve optimal performance, Taguchi's method and inverse-model technology were integrated, and a simulation platform developed by using the COMSOL Multiphysics software was used. First, the minimum number of experiments required to represent the full-factorial design and the most critical factors affecting product performance were determined by using Taguchi's method. Next, the adaptive neuro-fuzzy inference system (ANFIS) was extended to an MIMO–ANFIS architecture and used to train an inverse model for realizing the desired design. By using a well-trained model that can determine the inverse relationship of each input–output set, the manufacturing parameters for the optimal performance were obtained. A simulation study on a water-cooled condenser is presented to demonstrate the effectiveness of the proposed method.

Détails

  • Titre original : Integration of Taguchi's method and multiple-input, multiple-output ANFIS inverse model for the optimal design of a water-cooled condenser.
  • Identifiant de la fiche : 30017615
  • Langues : Anglais
  • Source : Applied Thermal Engineering - vol. 98
  • Date d'édition : 05/04/2016
  • DOI : http://dx.doi.org/10.1016/j.applthermaleng.2015.11.112

Liens


Voir d'autres articles du même numéro (61)
Voir la source