Improved wavelet-based techniques for power quality evaluation in three-phase systems

The paper deals with the analysis based on the Discrete Wavelet Transform (DWT) of stationary regimes in three-phase power systems with significant distortions. Firstly one presents the conclusion of preliminary evaluations used to select the most appropriate analysis technique: RMS values were evaluated using different DWT-based methods, analysis of errors, of memory consumptions and respectively of runtimes was done. Afterward data gathered from the excitation of the main generator from a power plant were submitted to a complex analysis. Three DWT-based methods were selected for comparative studies: one relies on the function dwt, provided by Matlab, the other two rely on original functions implementing the so-called “D4 Wavelet Transform” (“dwm” assumes that the analysed signal is periodic, whilst “dwi” performs interpolations, being applicable to non-periodic signals as well). The superiority of dwm and dwi is sustained by smaller errors (results yielded by the Fast Fourier Transform (FFT) were used as reference) and significantly smaller requirements for computational resources (memory and runtime).