Estudo e implementação de algoritmos para representação de sinais em distribuições tempo-frequência

Abstract

This work aims to implement time-frequency distributions in a structure developed to facilitate the design and implementation of new algorithms that assist in the analysis of signals containing power quality problems. Frequency time distributions are particularly useful for analyzing time-varying signals with non-stationary characteristics, as they have good temporal localization capability of the spectrum. Three signals containing power quality problems, transient oscillating signal; SAG, highlighted as a decrease in voltage over a short period of time (0.5 cycle at 1 minute and magnitude ranging from 0.1 to 0.9 pu) and SWELL, identified by increasing the effective value of voltage over a short period of time. Signals were analyzed with the Short-Term Fourier Transform, Continuous Wavelet Transform,Wigner-Ville Distribution, PseudoWigner-Ville Distribution, Smoothed Pseudo Wigner-Ville Distribution, and Choi-Williams Distribution, and compared by Error Square Mean Root method. The best result according to the RMSE criterion of root mean square error was found by applying the Wigner-Ville distribution to signals containing high frequency disturbances.