Analysis of acoustic signals and fault diagnosis in electric motors

The industrial use of electric motors is increasing day by day. The ability to detect faults in electric motors in advance is of vital importance both for the healthy and efficient operation of the motors and for the prevention of temporal and financial losses that may occur in the industrial activity where the motor is used. Fault diagnosis with the analysis of sound waves is a method that has been studied a lot lately and it is economical as well as giving results with high accuracy. By analyzing sound waves, it is possible to distinguish between a healty electric motor and a motor with an electrical or mechanical fault. In this study, it is aimed to process the audio signals related to certain faults of an electric motor and to classify them using machine learning algorithms. The MATLAB program has been preferred because it can convert sound waves into acoustic signals by directly transferring them to the program, transfer signals from time axis to frequency axis with the Fast Fourier Transform feature, examine the signals in detail with the spectrum analyzer feature, and provide ease of access and use of machine learning algorithms.