Estimação da contribuição de múltiplas fontes harmônicas utilizando regressão não paramétrica

Abstract

This paper presents a new methodology for estimating the contribution of multiple harmonic-producing loads in the voltage harmonic distortion in an electrical system. This method creates statistical models that describe the behavior of the harmonic voltage of an electrical system in function of the harmonic current of the loads present in the system so that it is possible to evaluate the impact of harmonic-producing loads in the levels of harmonic voltage in the electrical system under consideration. The creation of these models is performed based on the statistical method called local polynomial kernel regression, which is a nonparametric regression method, whose main feature is to create models not conditional on a specific family of regression curves, that is, data are solely responsible for the shape of the model. Once the model is created, it is estimated the harmonic voltage caused by the loads in the period under consideration, and, finally, this voltage is evaluated in relation to the harmonic voltage measured.