Journals Proceedings

Abstract

Ball bearings are one of the most important components in the machine involving the rotary motion. Ball bearing failure can cause a significant amountof maintenance cost and serious safety problems. Hence detection of the fault at its early stage is important. In this paper, a novel approach for fault detection of ball bearing by combining time, frequency and time-frequency domains are used.Two layer multiclasssvmis used for the classification of fault into an outer race fault, inner race fault, ball fault and healthy bearing. A comparison between Envelope Wavelet Packet Transform, Wavelet Packet Transform and proposed method is carried out. The experimental observation shows that the proposed method is able to detect the faulty condition with high accuracy.