A Novel Threshold for the Diagnostics of Rolling Element Bearing

The identification of the damage type in rolling element bearings is usually performed by means of suitable vibration signal analysis tools such as the most used and simplest method, Envelope Analysis through the corresponding Square Envelope Spectrum. The diagnostics and the monitoring of the bearing health are often performed by means of other approaches based on the evaluation of a damage index as the root mean square, the Kurtosis of the filtered signal, or more efficient indexes as the so-called Ratio of Cyclic Content. At any rate, in the case of real-time diagnostics, the definition of a threshold for the assessment of the bearing health is mandatory due to the presence in the vibration signal of additional sources and noises. In the paper, a threshold for the band-Kurtosis index that depends only on the sampling frequency and the bandwidth of the filter used for the demodulation of the vibration signal has been introduced. The effectiveness of the threshold has been proven by the experimental data of a damaged bearing.