The weak fault diagnosis and condition monitoring of rolling element bearing using minimum entropy deconvolution and envelop spectrum

Abstract

In vibration analysis, weak fault detect and diagnosis is of great importance. A method based on minimum entropy deconvolution and envelop spectrum analysis is proposed in this article. Minimum entropy deconvolution technique searches for an optimal set of filter coefficients to enhance the impulse making the filtered signal to contain clearer fault information. When there is a fault in bearing, it can be obviously reflected by the power spectral density estimation of the filtered signal even if the power spectral density estimation of original signal denies the information. The filtered signals are then analyzed by performing envelop spectrum analysis, where the bearing characteristic frequencies are quite clear for further diagnosis. The feasibility and validity of utilizing the minimum entropy deconvolution in weak fault diagnosis and condition monitoring is demonstrated by both simulation and experiments. The analysis of actual data from accelerated life test of rolling bearing shows that it can detect and diagnose fault 7 min earlier than RMS can reflect while the entire life takes 1962 min. This is considered to be significant in condition monitoring and early incipient fault diagnosis. The results of gearbox test also demonstrates the effective bearing fault diagnosis under strong noise and gear vibration interactions.