Uma nova proposta para detecção de faltas de alta impedância em redes de distribuição de média tensão

Abstract

Electric power distribution systems are susceptible to a serious and worrying problem, called high impedance fault (HIF), coming from the contact of the conductor with a certain surface of high electrical resistance, with possible falling of the energized conductor. This contact occurs through several factors resulting from equipment malfunctions, as well as by objects thrown at the network, natural disasters and the like. Conventional overcurrent protection equipment is not capable of detecting this type of phenomenon, since current levels generated are insuficiente to sensitize them, offering a high degree of danger to people and animals that come to the site, in addition to possible fire hazards on property. Given this, many researches have been done to develop methodologies capable of detecting HIFs. However, the detection of this type of event is still an open problem in the literature, and there is no completely robust and economically viable method. Thus, this work proposes a new technique for the detection of HIFs by extracting the characteristics of the currents sampled in the energy substation. This technique has a Fuzzy inference system that evaluates these characteristics and makes the classification of HIFs. With the help of the EMTP-ATP® software, the set of data needed to develop the method was achieved by simulating a radial feeder coupled to an electric arc model capable of representing the real characteristics of this type of fault. Once the oscillographs of the currents are obtained, the detection is done through some algebraic calculations that identify the existence of abnormalities in the signals, which are then analyzed and classified through fuzzy logic. After this detection, an analysis of the signal envelope is made, so that cases of single-phase, twophase and three-phase short-circuits are differentiated from possible cases of HIF. The proposed algorithm was implemented in the MATLAB® environment. The results obtained were satisfactory since the proposed technique correctly classified the occurrence of high impedance faults in all analyzed cases, as well as to discriminate these events from other short-circuit cases, proving that the methodology proposed in this work is robust and feasible for real-time application.