Analysis, Evaluation And Comparison Of Different Signal Processing Techniques For Motor Fault Diagnosis

Abstract

Induction motors (IMs) are amongst the most important machineries in industries and play a very crucial role. Any unnoticed failure in IM causes an unplanned shutdown. Condition monitoring (CM) is needed for the early diagnosis of developing faults. A faulty IM is characterized by unique spectral patterns. Digital signal processing is used for the extraction of spectral components. Often used method is  fast Fourier Transform (FFT). Matter-of-factly, it is not always possible to achieve coherent sampling. The window is used to compensate for waveform discontinuities that cause spectral leakage. It results in effective detection of faults at an early stage. It also helps in eliminating false-positive and false-negative fault diagnosis. The fault signals are simulated, and FFT is computed using standard libraries in C#. This paper presents the basic level diagnosis of faults in IMs through the use of FFT. Besides, through in-depth analysis and experimental validation, this paper also discusses an advanced level diagnosis of faults, wherein more appropriate window functions are suggested for the detection of faults.