Amplitude of probability density function (APDF) of vibration response as a robust tool for gearbox diagnosis

Abstract

A ‘Go' or ‘No Go' assessment is a safety requirement for quick and robust estimation of the condition of gearboxes used in helicopters and other critical machines. A range of vibration-based condition indicators (CIs) has been developed to meet this requirement. CIs are compared automatically with pre-set threshold values representing a healthy system, so that the health of the gearbox can be assessed and diagnosis made where necessary. The use of kurtosis of the residual signal of the measured vibration data, computed as part of the ‘FM4' method, is widespread, because it is accepted as a good and reliable indicator. However, it has been observed in some cases that FM4 may not show a continually increasing trend with the propagation of a fault. This behaviour may lead to improper assessment of the severity of the fault. Hence, a new CI, based on the deviation in the normal probability density function (PDF) of the measured vibration data, is suggested which demonstrates an increasing trend that is more robustly and monotonically correlated with the fault propagation.