Improving the signal data acquisition in condition monitoring of electrical machines

Abstract

The occurrence of a broken bar in induction machines causes an internal magnetic imbalance, which is reflected in the stator current. This effect may be detected by estimating the spectral signature of the stator current, particularly the sidebands around the fundamental frequency. However, the fundamental 60-Hz amplitude is considerably greater than the sideband amplitude, which usually requires an analog 60-Hz notch filter in the data acquisition system. Analog filters are sensitive to temperature variations, which, in this case, may shift the filter resonance frequency and degrade the desired response. It is proposed in this paper that the analog notch filter may be replaced by a digital/analog cancelling technique based on the recursive discrete-time Fourier transform. The technique is shown to attenuate the fundamental 60-Hz amplitude considerably before A/D conversion, leaving only the sidebands in the current signal