New nlpca model combining neuronal principal curves and rbf neural network for process monitoring

M. Bouakkaz, Mohamed-Faouzi Harkat, D. Messadeg

Research output: Contribution to journalArticle

Abstract

This paper describes a new Nonlinear Principal Component Analysis (NLPCA) model for process monitoring. The proposed model consists of two cascade three-layer neural networks for mapping and demapping, respectively, the mapping model is identified by using a Radial Basis Function (RBF) neural networks and the demapping model is performed by using an Input Training neural networks (IT-Net), in order to unify the optimization problem (estimation of nonlinear principal components and training of the neural networks), the nonlinear principal components, which represent the desired outputs of the first network, are obtained by the IT-Net (neuronal principal curves). In this paper, the algorithm of proposed (NLPCA) model is developed and applied to sensor fault detection and isolation of the Tennessee Eastman Process (TECP). For fault isolation, nonlinear fault reconstruction approach is considered. The obtained results verify the effectiveness of the presented model for fault detection, and demonstrate two advantages of the nonlinear fault reconstruction approach, on one hand, to isolate the faulty sensors and, on the other hand, to estimate the fault amplitude.

Original languageEnglish
Pages (from-to)492-502
Number of pages11
JournalMediterranean Journal of Measurement and Control
Volume11
Issue number4
Publication statusPublished - 1 Oct 2015
Externally publishedYes

    Fingerprint

Keywords

  • Fault detection and isolation
  • Fault reconstruction
  • Input training neural network
  • Nonlinear principal component analysis
  • Process monitoring
  • RBF-Neural network

ASJC Scopus subject areas

  • Control and Systems Engineering
  • Instrumentation

Cite this