Wavelet-Based Variance Analysis for Condition Monitoring of Gears

Gears faults are one of the most common issues in rotating machinery. If not detected on time, they may result in severe dysfunction of the kinematic chain or even in catastrophic breakdown of the whole mechanism. Vibration based methods dedicated to gears monitoring are widely studied and applied techniques, that in majority rely on comparison of some spectral features. In this paper, however, the method based on fractal theory and wavelet decomposition is studied for possible application to gears diagnostics. The discussed technique may serve as a tool for signal contraction and/or signals comparison according to frequency content. Presented results are based on experimental data recorded from gradually deteriorated gearbox and gear with mechanically introduced local fault.