DEFECT CLASSIFICATION IN NONLINEAR SYSTEMS USING SELF-ORGANIZING MAPS

The present work describes a new approach to the detection and diagnosis of functional defects within a nonlinear system highly unstable. The highly unstable nonlinear system is a distillation column of methylcyclohexane from a toluene / methylcyclohexane, which mass composition was defined to 23% i...

Full description

Saved in:
Bibliographic Details
Published inJournal of nature science and sustainable technology Vol. 9; no. 1; p. 45
Main Authors Boudebbouz, B, Manssouri, I, Ghanou, Y, Manssouri, T
Format Journal Article
LanguageEnglish
Published Hauppauge Nova Science Publishers, Inc 01.01.2015
Subjects
Clustering
Fault diagnosis
Neural networks
Nonlinear systems
Online AccessGet full text

Cover

More Information
Summary:The present work describes a new approach to the detection and diagnosis of functional defects within a nonlinear system highly unstable. The highly unstable nonlinear system is a distillation column of methylcyclohexane from a toluene / methylcyclohexane, which mass composition was defined to 23% in methylcyclohexane. The latter allows the partial separation of the more volatile compound methylcyclohexane content in the liquid mixture. The originality of this work lies in the application of Kohonen network Self Organizing Map as a classification tool in unsupervised learning of normal and abnormal modes of a nonlinear system highly unstable. The performances of the classifier based on Kohonen neural networks (Self Organizing Map) are considerably superior to other classic statistical methods.
ISSN:1933-0324