Fault classification in structures with incomplete measured data using autoassociative neural networks and genetic algorithm

Tshilidzi Marwala, S. Chakraverty

Research output: Contribution to journalArticlepeer-review

64 Citations (Scopus)

Abstract

A method for fault classification in mechanical systems in the presence of missing data entries is introduced. The method is based on autoassociative neural networks where the network is trained to recall the input data through some nonlinear neural network mapping. From the trained network an error equation with missing inputs as design variables is constructed. Genetic algorithm is used to solve for the missing input values. The proposed method is tested on a fault classification problem in a population of cylindrical shells. It is found that the proposed method is able to estimate single-missing-entries to the accuracy of 93% and two-missing-entries to the accuracy of 91%. The estimated values were then used in the classification of faults and the fault classification accuracy of 94% was observed for single-missing-entry cases and 91% for two-missing-entry cases while the full database set is able to give classification accuracy of 96%.

Original languageEnglish
Pages (from-to)542-548
Number of pages7
JournalCurrent Science
Volume90
Issue number4
Publication statusPublished - 25 Feb 2006
Externally publishedYes

Keywords

  • Fault classification
  • Genetic algorithm
  • Neural networks

ASJC Scopus subject areas

  • Multidisciplinary

Fingerprint

Dive into the research topics of 'Fault classification in structures with incomplete measured data using autoassociative neural networks and genetic algorithm'. Together they form a unique fingerprint.

Cite this