Bearing fault detection in induction motors using digital Taylor-Fourier transform

Induction machines are the most used electric machines in the industry, and bearing damage is their most common fault type. Although bearing damage can be localized or generalized, it can be caused by mechanical wear, manufacturing defects, incorrect installation, incorrect operation, or environmental factors. Bearings are a fundamental part of the rotating electrical machines; any abnormal conditions in their structure may cause, among others: overheating, progressive damage, vibrations, mechanic and electric stress or energy losses. Therefore, to avoid this type problems, reduce costs, and prevent a complete stop in the processes; actions of vital importance, such as prompt detection and classification of faults, must be fulfilled. Moreover, to accelerate repair processes it is necessary to implement an adequate schedule of maintenance or a punctual machine replacement. This upkeep goes under the current global trend in electrical systems which consists of automatizing systems based on intelligent algorithms and simplifying processes for users. This paper aims to propose motor current signature analysis through the digital Taylor-Fourier transform for detection and classification of bearing faults, aiming to apply the digital Taylor-Fourier filters in frequencies of interest, with the final purpose of the reconstruction of the filtered signal to obtain its frequency spectrum and, through statistical methods, identify in a precise way the bearing damage. The methodology was implemented in MATLAB and applied to three fault types for both loads and unload conditions: bearing ball damage, outer-race damage, and corrosion damage; each one of these for an operating frequency of 60Hz.