Motor Current Signature Analysis using Spectral Methods for Bearing Fault Diagnosis
Condition monitoring of induction motor is vital in the present scenario. However, 45-65% of induction machine faults are mechanical faults. Thus this work deals with the real time applications of motor current signature analysis (MCSA) for bearing fault diagnosis using spectral methods. In addition the proposed spectral analysis methods are compared tested by simulation also. The spectral methods implemented are Spectrogram, Short Time Fourier Transform (STFT) and Fast Fourier Transform (FFT) spectra of stator current for different bearing fault conditions. The work is implemented in three stages: Firstly, validating practical findings withtheoretical models for healthy and different fault conditions of the bearing. Second, visualizing the data with spectrograms by varying window sizes. Finally proposing adaptive window size selection for spectrogram and also enhancing the spectrogram by using Constant Q Transform(CQT). It was found that the proposed CQT spectrograms resultin the least temporal and spectral smearing and thus proven to be effective to diagnose the bearing faults.
Keywords - Motor Current Signature Analysis (MSCA), Localized Faults, Fast Fourier Transform (FFT), Short Time Fourier Transform (STFT) and Constant Q Transform (CQT).