The spectral signature of HUMS vibrations acquisitions are affected by the state of the monitored assets, but also environmental factors such as airspeed and engine torque. The former correlations is what makes vibration-based fault detection possible. The latter will manifest random changes in environmental factors as random noise on the acquisitions. This has a negative effect on vibration-based fault detection, as noisy signals increases uncertainty, making fault detection more difcult. This paper discusses the impact of changes in environmental conditions on HUMS vibration acquisitions, and methods to deal with this. Solutions discussed are traditional methods, such as limitations on the conditions where acquisition is permitted, as well as modelling to de-correlate vibration signatures and environmental conditions. Normalization by modelling is proposed as a method to correct each condition monitoring indicator given a number of contextual parameters. A method is also Outlined which permits correcting the raw acquisitions using an adaptive lter.
ASJC Scopus subject areas