Detection of broken rotor bar fault in induction machine fed by frequency converter

This paper proposes a new technique for detecting broken rotor bar faults in induction machines supplied by frequency converters. Traditional analysis methods, based on the fast Fourier transform (FFT), have some limitations, such as diagnosis under low load conditions, due to the spectral leakage effect, or the need of using long time samples for achieving sufficient resolution in the frequency domain. The method proposed in this paper, based on advanced signal processing techniques like the Hilbert transform (HT), addresses these problems and allows detecting broken bar even in the case of no-load conditions. The proposed method consists of the Hilbert transformation of the phase current and the spectral analysis via FFT of the modulus of this transformation. Experimental results from healthy and broken bar machines fed with a frequency converter are presented and compared. The new technique is also compared with classical FFT, and the chirp Z transform (CZT).