Motor Speed Signature Analysis for Local Bearing Fault Detection with Noise Cancellation Based on Improved Drive Algorithm

Motor speed signature analysis provides a noninvasive method for bearing fault detection. However, for the vector-controlled ac motors, periodic speed ripples related to fundamental frequency f_e and its twice harmonic 2f_e, which are caused by current measurement errors, are difficult to be attenuated by motor inertia or bandwidth of speed loop under low-speed conditions. The unwanted components would reduce the signal-to-noise ratio of motor speed and increase the difficulty of bearing fault detection. To solve the problem, this paper proposes a new noise cancellation strategy, which applies the improved drive algorithm instead of conventional signal processing schemes to cancel out the noise component before the data acquisition. Specifically, resonance controllers are introduced and set in parallel with the existed proportional-integral controller to suppress the speed ripples. Moreover, the envelope spectrum analysis is carried out to detect fault characteristic. The effectiveness of the proposed method is validated through simulation and experimental tests. Besides, its superiority under low-speed conditions is also demonstrated, compared with the spectral kurtosis of speed signal and three current-based methods.