Investigation on Effect of Various Parameter on Fault Diagnostics of Rolling Element Bearing

Abstract

Machinery fault diagnostics is an essential part of condition based maintenance activity in the industries. Among many techniques, the vibration signature analysis is a widely used technique to achieve the same. In the present work the problem of detection of fault in rolling element bearing is undertaken. The vibration signature of healthy and defective bearing are to be obtained from the developed experimental test rig for variable operating parameters like radial load, axial load and speed. The comparison of vibration signature profile for healthy and faulty bearing is to be done to facilitate the prediction of fault operating condition. Further, it is also necessary to find some statistical parameter such as RMS, Peak to peak, Kurtosis, Crest factor and Skewness for detection of faults in bearings. The artificial faults of localized nature are seeded on inner race and outer race of bearing however, the distributed defect is seeded only on the outer race in the present work. The faulty bearings are loaded externally through hydraulic cylinder in test rig. The signature of vibrations are analyzed for radial load as well as combine loading. Due to fault present on the inner race, the obtained vibration signature are combined with side bands. Because of side bands it is difficult to find the defect frequency accurately. In this work side bands are due to relative motion between the balls and rotating inner race. In order to remove this side band, band pass filter is used, for that required bandwidth is obtained with the help of kurtogram. Further, envelope analysis technique by Hilbert transform is used for removing the side bands. However, when the defect is present on outer race only Hilbert transform is required without plotting kurtogram. Another objective of this work is to compare various estimated statistical parameter from signature for the localized and distributed seeded faults.