EMD and MCSA Improved via Hilbert Transform Analysis on Asynchronous Machines for Broken Bar Detection Using Vibration Analysis

Abstract

Induction motors are very important components of electromechanical energy conversion equipment, as they are robust and reliable machines. One of the main techniques for identifying incipient faults in these rotary machines is vibration-based condition monitoring. In this paper, the analysis is focused on squirrel-cage broken bars, a typical fault in induction motors. Using an experimental test rig in which this fault may be gradually introduced in a healthy motor, the paper shows how it changes the vibration pattern. This methodology, based on motor current signature analysis, originally developed for motor current analysis, was modified and applied to the mechanical vibration signals. It has shown itself effective for detecting faults when the motor load is 37.5% higher than the nominal value. Then this methodology was combined with Empirical Mode Decomposition to filter the signal and extract the intrinsic mode functions that contain fault’s characteristic-frequency components. This methodology successfully detected faults when the motor load was at 12.5%, that is the lower load tested, and with less sampling time. Besides that, the fault’s frequencies characteristics were found and demonstrated how important is the motor operation regime.