Rotor bar fault feature extraction of induction motor base on FFT and MUSIC

In this paper, the rotor broken-bar faults feature component are difficult to detect because it always hides behind the strong supply frequency component in the spectrum of the stator current .So a method is put forward that uses FFT algorithm to pre-estimate the frequencies of a signal then uses MUSIC (Multiple Signal Classification) algorithm to complete the frequency search. This method overcomes spectral leakage and fence effect of FFT and shortens peak search time of MUSIC. Simulation results show that fault characteristic components can be obtained accurately through the method presented even with small samples.