Uma nova técnica para detecção e classificação de transitórios em sistemas elétricos de potência

Abstract

Often, the electrical power systems are exposed to transient disturbances that produce distortions in the current and voltage signals of the network, which may cause the interruption in the supply of energy by the action of protective relays. In this context, various methodologies for detection and classification of transients have already been proposed. However, none of them presented satisfactory performance for all operational and transitory conditions of the electrical power systems. Thus, this study proposes the enhancement of a technique for detection and classification of transitory events. The transient detection occurs when an acceptable limit value for normal operating conditions of the system is exceeded in one of the plans, corresponding to the current and voltage signals of the network, for at least one of the phases. From the detection of transient, it enters in the classification stage, which is made in accordance with the patterns recorded for each type of transiente event measured in relation to the behavior of the Euclidean distances in the already mentioned plans. To do this, it is necessary to use three levels, both for the current and voltage signals, which, when exceeded in accordance with the pattern observed for each waveform, characterize certain transitory. The classification is established when the patterns detected do not change by at least one cycle of the signal fundamental frequency. The proposed technique was implemented in software MATLAB® And tested for sampling rates of 16, 32, 64 and 96 samples per cycle, as well as for a wide range of transients, such as single-phase, phase-to-phase and, three-phase short-circuits, as well as events of load change, mono/bipolar oppenings and transformers energization. Finally, it should be emphasized the great robustness and ability to classification of the proposed methodology, which acted correctly in all analyzed signals.