Bearing fault severity estimation using time-based descriptors for rotating electric machines

Abstract

This paper evaluates the bearing fault severity in rotating electric machines based on the time-based descriptors derived from machine vibration signal. Time-based descriptors are independent of bearing dimensions and machine dynamics. To establish the relationship between the time-based descriptor and bearing fault severity, a wide range of defective bearings are examined on an experimental set-up and the factors contributing to the value of time-based descriptor are determined. The application of this technique to the bearing faults such as rolling element, inner-race and outer-race fault, has confirmed that time-based descriptor successfully measures the severity of faults. The significance of time-based descriptors has been evaluated and tested by developing online bearing failure situation via shaft current. Since the time-based descriptors are independent of dimensions and characteristic frequencies of the bearing, the severity estimation is less sensitive to operating conditions for data acquisition. The laboratory investigations on 10-hp cage induction motor are presented.