Abstract
The goal of this work is to apply higher-order statistical foundations on rotating machine monitoring by vibrations processing. Specifically the investigation was approached to detect unbalance failure and to estimate its magnitude. The machine under diagnosis was an experimental rig so that results obtained are all experimentals.
Higher-order moments and cumulants of horizontal and vertical vibrations were the statistical features computed and used to perform diagnosis. Higher-order statistical characteristics are less affected by
Gaussian noise than first and second-order statistics.
It was proved the effectiveness of the application when the system carries the specific failure (unbalance) detection out under noisy conditions. In particular, this work could be applied on other failure diagnosis.