Modern manufacturing plants rely heavily on the use of automation. Automated facilities use sensors to measure material state and react to data patterns, which correspond to physical events. Many patterns can be predefined either by careful analysis or from domain experts. Instances of these patterns can then be discovered through techniques such as pattern recognition. This approach will fail to detect events that have not been predefined, however, potentially causing expensive production errors. A solution to this dilemma, novelty detection, allows for the identification of interesting data patterns embedded in otherwise normal data. This paper describes several novelty detection methods for time series data that have been proposed in the literature.
Last modified: Tue 2006.07.25 at 09:25 NDT by Dennis Peters