Um método de estimação de potência baseado na transformada de Stockwell.

Abstract

A power calculation method for non-sinusoidal conditions based on the Stockwell transform (ST) is proposed in this thesis. The proposed power components are founded

on an apparent power decomposition based on the impedance of each frequency com- ponent. As a consequence, the power components have characteristics associated with

impedance, resistance, and reactance. Additionally, the ST is applied to estimate, in the time-frequency domain, the voltage and current characteristics, which is a fundamental

task for the proposed method’s suitability for power estimation in non-stationary con- ditions. The ST performs a time-frequency analysis with absolute information on each

frequency component’s amplitude and phase angle, which is essential for power represen- tation. Furthermore, tests were carried out aiming to evaluate the ST performance, i.e.,

to indicate its potentialities and limitations for signals’ amplitude and phase estimation.

The proposed power definitions were compared with those established in Currents’ Phy- sical Components (CPC) power theory and in IEEE Std. 1459-2010 by using stationary

synthetic signals. Moreover, the proposed method was evaluated under non-stationary conditions, including synthetic, simulated, and actual signals. The results attest to the suitability of the proposed method for characterizing the electrical power in systems with non-sinusoidal signals with non-stationary characteristics. Additionally, a new method

for detecting transient disturbances in power systems was presented. The method is ba- sed on the proposed power components, which presented a typical behavior for various

disturbances, being quite appropriate for extracting patterns that allow the detection of disturbances in electrical systems.