The higher order spectrum analysis for fault pattern extraction of induction motors

This paper presents an application based on higher order spectra (HOS) for fault pattern extraction of induction motors. The focus was put on one of major HOS measures, the bispectrum in particular. The bispectrum is the third-order frequency domain measure, which contains information that conventional spectral analysis techniques cannot provide. A series of experiments for fault diagnosis of induction motors by vibration bispectrum analysis were carried out. The experimental results showed that the bispectrum of vibration could significantly suppress the noise and therefore highlight the relevant fault symptoms. It was also found that there were unique bispectrum patterns for different faults of induction motors.