Análise e implementação de métodos aplicados aos sinais variantes no tempo presentes em sistemas elétricos de potência

Abstract

Traditionally, the harmonic analysis in power systems is performed in frequency domain and considering the signals in steady state. Moreover, the majority of power quality indices has been developed considering the periodicity of the voltage and current waveforms and as function of the Fourier Transform. However, power signals are becoming increasingly complex and the stationarity conditions are difficult to be achieved. This is due to the widespread use of electronic devices and nonlinear loads that produce harmonics, inter-harmonics and sub-harmonics and the penetration of distributed generation that can bring power quality issues such as voltage fluctuations, flickers and waveforms distortions. Methods such as Fourier Transform are not efficient in analyzing time-varying signals because they do not have dynamic response, lose the time information and suffer of frequency spillover. In this context, new signal processing techniques must be considered for a more accurate analysis of power signals. These techniques must provide information in time and frequency domain and in real time. The methods discussed in this work are the Slide Window Recursive Fourier Transform, Wavelet Packet Transform and Filter Bank. Methods to calculate parameters considering non-sinusoidal, unbalanced and time-varying signals are presented. It is shown the performance, improvement and applications of the time-varying decomposition methods using simulated, laboratory and industrial signals. These methods allow researchers and engineers to look into the transient phenomena through the observation of the behavior of time-varying harmonics for better understanding the time-varying nature of harmonics.