Abstract
Control chart pattern recognition is one of the most important tools in the identification of process problems in modern manufacturing industries. Abnormal patterns including systematic, cyclic, drift, and shift could be involved with certain assignable causes. Conventional control charts could not inherently recognize these patterns. In this paper, multi-resolution wavelet analysis is used to extract features. A self-organizing map then generates cluster vectors with wavelet coefficients. Using these features, a back-propagation network classifies unnatural patterns. The performance evaluation result is better than those of other competitive methods.
Original language | English |
---|---|
Pages (from-to) | 863-866 |
Number of pages | 4 |
Journal | Advanced Science Letters |
Volume | 13 |
DOIs | |
Publication status | Published - 2012 Jun |
Keywords
- Artificial neural network
- Control charts
- Feature extraction
- Multi-class classification
- Pattern recognition
- Wavelet analysis
ASJC Scopus subject areas
- Computer Science(all)
- Health(social science)
- Mathematics(all)
- Education
- Environmental Science(all)
- Engineering(all)
- Energy(all)