Análise de impactos harmônicos em rede de média tensão utilizando redes neurais

Abstract

A methodology is presented to assess the impact caused by the generation of harmonics, located on a bus of interest in the medium voltage substation, in the electric grid of the energy concessionaire. The estimate is to carry out the construction of a model based on data collected, applying the technique of artificial neural network (ANN). The input of the neural model is built by harmonic currents, impacted by nonlinear loads that make up the system of the proposed methodology. The objective of this dissertation is to conduct a measurement campaign to collect data from the electrical system, to demonstrate a study on the quality of electrical energy, considering the recommendations of IEC 61000-3-6 (IEC / TR, 2008-02), IEEE Std. 519-2014, EN 50160-2004 and PRODIST (module 8,2017) and propose the development of an algorithm capable of evaluating measurements using the artificial neural network technique. The technique used in this project was the artificial neural network (ANN) technique, which made it possible to construct the model well with the validation of the results obtained from comparisons of the reference results via computer simulation with results obtained through the neural model. Therefore, the dissertation presented and applied in practice with case studies of the actions for analysis of harmonic impacts in electrical energy distribution systems through the construction of mathematical models using the neural network, obtaining good results, thus validating the applied technique.