Current Noise Cancellation for Bearing Fault Diagnosis Using Time Shifting

The stator current of a typical induction motor involves the supply fundamental and its harmonics existing before and after the presence of a bearing defect, and much of the information they contain is not related to the bearing defect. In this sense, they are basically noise to bearing fault diagnosis problem. This paper develops a current noise cancellation method using time shifting. Current residue is obtained by adding digital current signal to its own sample-delayed representation as an antinoise component. The amount of the sample delay is only dependent upon the supply frequency and the sampling rate. This amount is set to eliminate the supply fundamental and its odd multiple harmonics. After obtaining the current residue, through the spectral analysis of the envelope of the current residue, the characteristic fault frequencies for a defective bearing can be revealed. To show the superiority, the developed method is experimentally compared with three of the current-based methods in the literature. Experimental results for the defects in the outer raceway, the inner raceway, and the ball of a bearing verify the merits and effectiveness of the developed method.