Bearing fault analysis using variational mode decomposition

Abstract

Bearing health analysis plays a significant role in industry to improve reliability and performance of critical processes by alarming the faults at early stages. Conventional techniques do no guarantee to detect the faults at early stages because the low energy bearing frequencies get suppressed by stern noise and higher vibrations. The Fast Fourier Transform fails to analyse the transient and non-stationary signals directly. This paper performs the signal analysis on vibration data of ball bearing using Variational mode decomposition (VMD). Firstly, the intrinsic mode functions are extracted using VMD followed by Fast Fourier Transform, and finally the status of bearing is analyzed to be faulty or impeccable. This paper, stress on VMD rather than on EMD, due to its qualities in the detection of close tone vibration signatures and takes less computation time.