Using wavelets to define and detect harmonic fingerprints in non-sinusoidal waveforms

Waveforms acquired from a power plant were analyzed with the Discrete Wavelet Transform (DWT). For the steady state case reference data were calculated (details and approximations vectors and 3-D tables which can be used to determine faster, in an original way, the parameters of a harmonic distortion). An original algorithm for the evaluation of parameters corresponding to harmonic disturbances was issued. It involves the preliminary determination of proper hash functions and “harmonic fingerprints”. For the 1-st analyzed signal, tests on 3000 harmonic disturbances with randomly generated parameters proved a “hit rate” of 96.5% and the mean execution time for the determination of harmonic distortion parameters was reduced by a factor of 8.37. Tests on other steady non-sinusoidal waveforms yielded similar results.