Detection of Nonadjacent Rotor Faults in Induction Motors via Spectral Subtraction and Autocorrelation of Stray Flux Signals

In this paper, statistical signal processing techniques are applied to electromotive force signals captured in external coil sensors for adjacent and nonadjacent broken bars detection in induction motors. An algorithm based on spectral subtraction analysis is applied for broken bar identification, independent of the relative position of the bar breakages. Moreover, power spectrum analyses enable the discrimination between healthy and faulty conditions. The results obtained with experimental data prove that the proposed approach provides good results for fault detectability. Moreover, the identification of the faults, and the signal correlation indicator to prove the results are also presented for different positions of the flux sensor.