Rotor Asymmetries Faults Detection in Induction Machines under the Impacts of Low-Frequency Load Torque Oscillation

Low-frequency torque oscillations (LTOs) characteristic components emerge in the stator current spectrum of induction machines (IMs) as additive frequencies near the rotor asymmetry fault (RAF) indices especially in gearbox-based electromechanical system. The interactions between these two components make the fault detection process complicated and lead to false alarm. In this paper, a new technique for detection and separation of RAF from LTOs in IMs based on single phase stator current data is proposed. The method benefits from a novel pre-processing stage based on several sign functions. Hence, a two-axis rotating reference frame with a single phase of stator current of IMs with no prior knowledge of the rotational speed is introduced. The proposed method maps the static reference frame obtained through single stator current and its associated Hilbert transform to the proposed rotating reference frame which can separate the effects of LTOs from RAF, effectively. The validity of the proposed technique is tested through theoretical analysis, and experiments in both steady-state and transient conditions. In this regard, Synchro-squeezing Wavelet Transforms (SWT) is used for time-frequency analysis of faulty stator current in transient conditions. The obtained results confirm the effectiveness of the proposed approach to separate the RAF characteristic frequency from LTOs even in line-fed IMs applications.