Diagnosis method for rub-impact fault of rotating machinery based on EMD and Hilbert envelop spectrum

Aiming at the characteristics of the casing vibration acceleration signal of rotating machinery with rubbing fault, a diagnosis method based on EMD (empirical mode decomposition) and Hilbert envelop spectrum is put forward. When the rubbing fault occurs between the rotor and the stator, the vibration acceleration signals from the casing of the rooting machinery will present amplitude-modulated feature due to the periodical rub-impact between the rotor and the stator. However, it is relatively difficult to extract the rub-impact feature since the rub-impact signal in the process of transferring to the casing is always buried in the noise and background signals. In this paper, by means of the powerful capability of EMD and Hilbert envelop spectrum in processing nonlinear signal can accurately describe nonlinear behaviours, the fault signal is adaptively decomposed by EMD to acquire intrinsic mode functions (IMF) containing obvious feature of amplitude modulation. Then some of the sensitive IMF components are transformed by FFT and their Hilbert spectrum are analysed. The analysis results from the experimental signal show that the EMD provides an effective and adaptive method to obtain 'purified' casing vibration acceleration signal, which demonstrates the transient casing vibration more precisely and vividly, decreasing misinterpretation of the machine running condition. Therefore, the EMD approach can be applied to the single-point rubbing fault diagnosis of the rotor systems efficiently.