Detecção de faltas de alta impedância em sistemas de distribuição primários de energia elétrica.

Abstract

In general, all electrical systems are susceptible to failure. However, there are faults that are not detected by the protection system, such as those caused by tree branches that hit the power grid or a cable break under an insulating surface such as asphalt. These faults are known as high impedance faults (HIF). HIF present themselves as a challenge and cannot be detected by traditional methods. In this work, a detailed study of detection techniques for high impedance faults presented in technical literature was carried out, identifying the conditions in which each improves detection behavior as well as possible deficiencies in each technique studied. The techniques existing in the literature can be classified as either passive or active. From the passive techniques present in the literature, five were chosen so that a comparative study could be carried out in a balanced way, starting with the same scenario and evaluating all the techniques under the same conditions. An active technique has also been studied from the standpoint of performance in terms of detectability and type of fault. Based on all these studies a change was proposed in the active technique. These modifications resulted in improved performance compared to the original. In order to complete this study, this work presents the possibility of combining the passive techniques for detecting HIF. The technique based on the observation of low order harmonics of negative sequence was combined with the technique based on wavelets. An artificial neural network was used to produce a final result, since it is necessary to combine two results to allow precise diagnosis of the fault condition. This combination allowed the detectors to distinguish the events in the system and correct detection of the faults in all cases.