Dennis M. King

Static and dynamic novelty detection methods for jet engine health monitoring.

Novelty detection requires models of normality to be learnt from training data known to be normal. The first model considered in this paper is a static model trained to detect novel events associated with changes in the vibration spectra recorded from a jet engine. We describe how the distribution of energy across the harmonics of a rotating shaft can be learnt by a support vector machine model of normality. The second model is a dynamic model partially learnt from data using an expectation–maximization-based method. This model uses a Kalman filter to fuse performance data in order to characterize normal engine behaviour. Deviations from normal operation are detected using the normalized innovations squared from the Kalman filter.

The use of novelty detection techniques for monitoring high-integrity plant

Total care schemes are now a common feature in the sales of power generation and propulsion plant. To mitigate risk of financial penalties and maximise profit, many suppliers will rely on health usage and condition monitoring techniques. Intelligent condition monitoring is a relatively new concept in this field and introduces prognostic capability. One key obstacle in this approach is the implementation of some form of rule-base that encapsulates possible fault conditions. The difficulty here is that a given fault scenario will not necessarily manifest itself in the same manner twice and will require complex rule-sets to describe possible variations in the development of the fault. In addition, due to the robustness of current high-integrity plant, example fault conditions are very rare and hence difficult to model using data driven approaches. Seeding faults during development is one approach often used, however, this can never be entirely representative of an in-service failure in addition to being a costly exercise. This paper describes the practical implementation of novelty detection schemes that aim to overcome the limitations described above.