Transient detection of eccentricity-related components in induction motors through the Hilbert–Huang Transform

The identification and extraction of characteristic patterns are proposed in this work for the diagnosis and evaluation of mixed eccentricities in induction electrical machines with parallel stator branches. Whereas the classical diagnosis approaches, deeply spread in the industrial environment, are based on the Fourier analysis of the steady-state current, the basis of the proposed methodology consist of analysing the current demanded by the machine during the connection process (startup transient); the objective is to extract the characteristic evolution during the transient of some harmonic components created by the fault; this evolution is caused by the dependence of these components on the slip (s), a quantity varying during the startup transient from 1 to almost 0. For this feature extraction, the Hilbert–Huang Transform (HHT) is proposed. An analysis of the behaviour of this transform in comparison with another time-frequency approach used in other works, the Discrete Wavelet Transform (DWT), is also presented in the paper. The results show the usefulness of the methodology for the reliable diagnosis of the mixed eccentricity fault and for the correct discrimination against other types of failures.