A. Domijan, A. Hari, and T. Lin (USA)
Power quality, Harmonics, Power quality monitoring, Power instrumentation, Wavelet transform.
The wavelet transform (WT), as a tool to analyze non-stationary power quality (PQ) events, has been widely explored by researchers and engineers in the electric power engineering field. However, performance of various wavelet filter banks suitable for PQ monitoring has not been thoroughly studied and investigated. In this paper, the authors present a criterion to evaluate various wavelet families and select the appropriate wavelet filter bank for PQ monitoring. This criterion is based on the correct localization of a specific signal component in the frequency domain and accurate measurement of its amplitude. For this purpose, frequency properties of the selected wavelet filter bank should be flat in their respective frequency bands and should decrease sharply near the cut off frequencies, i.e. the frequency properties should be as close to ideal low pass and high pass filters as possible. Also, the frequency properties should not overlap with each other in order to avoid energy leakage in the scalogram, which is a serious pitfall of Fast Fourier Transforms (FFT). By using software simulation, it has been proven that Coiflet 5 (coif5) wavelet filter bank has the best frequency properties among the wavelet families under consideration (Daubechies, Coiflet and Symlet wavelets). It satisfies the criterion and hence can be used for analyzing power disturbances. For the purpose of real time implementation of the WT via Mallat Algorithm, computational complexity has also been compared and discussed in this paper.
Important Links:
Go Back
