Use of Empirical Wavelet Transform for Detection of Compound Fault Based on Vibration Signals

A gearbox consists of gear pairs, bearings and shafts. So far, investigation of gearbox vibration signal is performed assuming either gear faults or bearing faults separately. Despite this, it can have both gear and bearing faults concurrently. Such an occurrence of a fault is treated as a compound fault. A vibration signal obtained for such compound faults would have multi-modulation effects. In FFT, the sidebands in the resulting vibration signal will be difficult to investigate due to overlapping and noise. Hence, an attempt is made to exhibit the capability of empirical wavelet transform to detect the faults separately. Simulation and experimental investigations are presented to show the compound fault detection approach.