Wavelet analysis for fault diagnosis of electrical machines using current signals

Fault diagnostic algorithms for electrical machines are various. The most popular approach - the Fourier transform - suffers several drawbacks. Therefore, the authors offer to replace the Fourier transform with the wavelet analysis. As a reference base, they have selected the Daubechies wavelet. The replacing diagnostic algorithm is described in two steps: intransferring the Fourier frequency characteristics to the waveletscale and subsequent finding the total energy and amount of the wavelet transform at each scale. The hypothetical approach is endorsed with experimental results, which show that when a fault of a certain characteristic occurs, the energy scale increases and the wavelet amount reduces.