Combination of minimum entropy deconvolution method and Van Cittert algorithm for features extraction of bearings

Abstract

Rolling bearings functionality has a primary importance for the correct operation of the rotating machines. In this paper, a monitoring technique based on deconvolution approach is proposed to restore the impulsive shape from the measured vibration signal. This latter is obtained from a convolution of real impulse signal and transmission function. The proposed procedure consists of two major steps; firstly, using the minimum entropy deconvolution (MED) to obtain the inverse filter, secondly introducing the iterative deconvolution algorithm to go back to the initial problem that is mathematically described by the convolution process to restitute the impulsive signal. The proposed procedure is applied to bearing diagnosis, and its effectiveness is validated by simulated and experimental data acquired from operational bearings. Moreover, the monitoring obtained results are satisfactory.