Analytical and Practical Perspectives on Fault Identification of Induction Motors

Despite being a widely adopted method, the short-time Fourier transform fails to identify fault characteristic frequencies (FCF) from rapidly changing vibration signals. The wavelet transform further has the underlying issues of energy leakage, long computation time, interference term, and border distortion. The Hilbert-Haung transform (HHT) analyzes the vibration signal using intrinsic mode functions (IMFs), which are acquired through empirical mode decomposition (EMD). However, decomposition may produce redundant and extraneous IMFs that hamper the result accuracy. Furthermore, the use of the Hilbert transform (HT) in HHT for demodulation intensifies computational complexity and introduces subjective error in the FCF calculation. In this paper, an accurate and efficient signal analysis approach based on EMD, Correlation Coefficient (CC), Teager-Kaiser Energy Operator (TKEO), and Spectral Analysis (SA) is proposed for the FCF identification of an induction motor. Significant and related EMD-based decomposed IMFs are segregated by CC to avoid computational complexity and irrelevant analysis. Moreover, the computation of the envelope by the TKEO only requires three consecutive measurements of the IMF signals, making the result almost instantaneous.