New Statistical Moments for Diagnostics of Rolling Element Bearings

Statistical moment analysis has proven to be a very effective technique for diagnosis of rolling element bearings. The fourth normalized central statistical moment, kurtosis, has been the major parameter in this method. In this paper it will be shown that the third normalized statistical moment can be as effective as kurtosis if the data is initially rectified. The advantage of this moment over the traditional kurtosis value is its lesser susceptibility to spurious vibrations, which is considered to be one of the shortcomings of higher statistical moments including kurtosis. The sensitivity of this moment to changes of load and speed is also less than kurtosis. The proposed method can also be applied to higher odd statistical moments.