Skip to Main content Skip to Navigation

Méthodologie de conception de contrôleurs intelligents par l'approche génétique. Application à un bioprocédé

Abstract : In this work, the problem of design of fuzzy controllers is studied. In a first part, we present a state of art on the techniques used, namely the genetic algorithms and its various alternatives, the neural networks, fuzzy logic and their hybridizations. Taking support on this state of art, we propose a first design method of the fuzzy controllers of Mamdani by simple genetic algorithms. Thereafter, this method is improved by the use of the hierarchical genetic algorithms. These algorithms allow, by the means of the structure of their chromosomes, a better optimization of the controller parameters while eliminating the incoherent rules which can arise, as well as for the first method, at the end of the optimization process. The last method suggested relates to the synthesis of the fuzzy controllers of Sugeno. It is based on a hybrid procedure of training which proceeds in two stages. During the first stage, the fuzzy controller is represented in the form of a network of multi-layer neural networks, whose parameters are optimized by the algorithm of retro propagation. In the second phase, the parameters obtained at the end of the first phase are extracted and optimized by the NSGA-II according to a coding arranged hierarchically. These methods are applied for control of an alcoholic fermentation process in continuous mode.
Document type :
Complete list of metadatas
Contributor : Arlette Evrard <>
Submitted on : Monday, June 8, 2009 - 11:03:03 AM
Last modification on : Friday, January 10, 2020 - 9:08:09 PM
Long-term archiving on: : Thursday, June 10, 2010 - 10:18:48 PM


  • HAL Id : tel-00392473, version 1


Ouabib Guenounou. Méthodologie de conception de contrôleurs intelligents par l'approche génétique. Application à un bioprocédé. Automatique / Robotique. Université Paul Sabatier - Toulouse III, 2009. Français. ⟨tel-00392473⟩



Record views


Files downloads