Vibration based mechanical diagnostics, such as used in health and usage monitoring systems (HUMS), can successfully identify anomalous components. For shafts, where damage can be directly measured as an exceedance in shaft order 1, 2 or 3 (a vibration vector 1, 2 or 3 times the RPM of the shaft), it is simply a matter of taking the spectrum of the vibration data, and noting if the magnitude of the shaft order frequency exceeds some threshold. In some cases, the component manufacturer has published limits. In other cases, statistics from a population of components are calculated, and a threshold is set such that the probability of the component being normal, when exceeding the threshold, is small. This event or fault detection is diagnostic in nature. To increase the utility of the HUMS system, the ability to predict the remaining useful life (RUL) on the component would greatly improve the operational readiness, reduce the logistic foot print and consequentially, reduce maintenance cost. The current diagnostics capability gives some marginal prognostic capacity (i.e. if the measured component health is deteriorating, one would suggest that maintenance will be required "soon"). Further benefit would be gained if a reliable estimate of the component's RUL was similar to the logistic supply line time scale. That is, 50 to 100 operating hours of advance notice vs. 2 to 5 hours of notice currently available. The paper investigates the use of Hidden Semi-Markov Model (HSMM) to predict the RUL on a component. The intent is to investigate modeling techniques which will push the RUL prediction, reliably, to 100 hours. Real world data from 30 utility helicopters is used to train and validate the model.
Use of hidden semi-markov models in the prognostics of shaft failure
2006
6 Seiten, 9 Quellen
Conference paper
English
Use of Hidden Semi-Markov Models in the Prognostics of Shaft Failure
British Library Conference Proceedings | 2006
|Online Identification of Hidden Semi-Markov Models
British Library Conference Proceedings | 2003
|Large-Scale Event Detection Using Semi-Hidden Markov Models
British Library Conference Proceedings | 2003
|