Classifying disc defects in optical disc drives by using time-series clustering

• Published in: 2004 American Control Conference Proceedings; Volume 4 of 6 Vol. 4; pp. 3100 - 3105 vol.4
• Main Authors: van Helvoirt, J., Leenknegt, G.A.L., Steinbuch, M., Goossens, H.J.
• Format: Conference Proceeding Journal Article
• Language: English
• Published: Piscataway NJ IEEE 01.01.2004; Evanston IL American Automatic Control Council
• Subjects: Applied sciences; Asia; Computer science; control theory; systems; Control theory. Systems; Digital TV; Drives; Exact sciences and technology; Fault detection; Fingerprint recognition; Mechanical engineering; Optical beams; Optical control; Optical distortion; Surface contamination; Hierarchical classification; Defect; Time series
• ISBN: 9780780383357; 0780383354
• ISSN: 0743-1619
• DOI: 10.23919/ACC.2004.1384385

Optical disc drives are subject to various disturbances and faults. A special type of fault is the so-called disc defect. In this paper we present an approach for disc defect classification. It is based on hierarchical clustering of measured signals that are affected by disc defects. The time-series are mapped into a feature space after which the feature vectors are clustered in a hierarchical fashion. Finally, signals are fitted onto the clusters to obtain single representations for each fault class. The resulting class descriptions can then be used for (on-line) classification of new disc defects. The approach is evaluated by applying it to a set of test data.