Detection of Rolling Element Bearing Damage by Statistical Vibration Analysis

Abstract

A new method is presented for predicting rolling element bearing condition from measurements of bearing housing vibration. This method is based on a statistical parameter Kurtosis, that remains constant for an undamaged bearing irrespective of load and speed, yet changes with damage. The extent of damage can be assessed from the distribution of this statistical parameter in selected frequency ranges. An assessment of bearing condition can thus be made with minimum recourse to historical information. Most other damage detection techniques rely heavily on the trend analysis of data and so this new method may prove to be a significant advance in bearing fault detection technology, at least when viewed within the original objective to provide a simple and cheap technique. As with most other simple detection techniques, the precise nature of the fault cannot be defined and for such information it is necessary to use the more sophisticated diagnostic methods.