Gear spall detection by non-stationary vibration signal analysis

This work seeks the diagnosis of gear local faults by non-stationary vibration signal analysis, such as continuos wavelet transform, Wigner-Ville distribution and cyclostationary analysis. The diagnostic capabilities of the foregoing techniques are discussed on the basis of simulated and experimental results, in particular tooth spalls of different sizes are artificially introduced along the gear tooth face via a drilling process. A vibration signal model including the effects of local faults in involute gears is presented. This model is applied to study the potential effectiveness of the considered processing techniques. After that, the vibration signals acquired from a gearbox in sound and faulty conditions are analyzed with the foregoing non-stationary signal processing techniques, in order to compare the effectiveness of the considered techniques for local fault diagnostics. © 2008 by the Katholieke Universiteit Leuven Department of Mechanical Engineering All rights reserved.