Novelty detection in jet engines
Tarassenko L., Nairac A., Townsend N., Cowley P.
The principle of novelty detection in jet engines was discussed and the vibration spectra recorded from the jet engines was analyzed. The real advantage of novelty detection, which have not previously been seen would also be highlighted. The best description of normality in the statistical sense, was the unconditional probability density function p(x) of the training data, where x were considered a vector of features or parameters extracted from the data. The analysis of the vibration spectra recorded from jet engines during their passs-off tests was made with the intention of using both Gaussian mixture models and Parzen windows to construct statistical models of normality.