Vibration-based health monitoring methodology for helicopter gearboxes


This paper proposes a vibration-based health monitoring methodology for gears. This methodology mainly consists of three principal steps. In the first step, the gear residual signal is computed by applying first a synchronous average with respect to its rotating period, then subtracting the meshing signal from the latter. In the second step, an optimal filtering is performed to enhance the modulation of the suspected gear. More specifically, a linear periodically time-varying filter is designed to optimize the cyclostationary criterion of the output (i.e. the filtered signal) associated with the suspected gear frequency. In the third step, the squared envelope is calculated through the Hilbert transform and the synchronous average is applied, once again, to the latter with respect to the wheel period in order to obtain the synchronous power. Incidentally, such pre-processing has for advantage to conserve the shape of the wheel modulation, thus offering a relevant indicator for fault detection and identification. The potentiality of the proposed approach is demonstrated on real vibration signals captured from a helicopter gearbox.