Implementação de sistema para detecção e classificação automática de Sags em sistemas trifásicos

Abstract

The growing number of pollutant loads linked to the grid leads to research in the area of electric power quality. Such pollution is visible as deformities in the voltage waveform, among which the voltage sags stand out. This work aims to detect the moment of occurrence of the disturbance through Independent Component Analysis. With such information, the system segments the electrical signal so that it can be processed through higher order statistics. The information obtained by such a process is used for automatic classification of causes of the three-phase sag - faults, induction motors starting and transformer energizing - by means of a multiclass Support Vector Machine. The system was developed in LabVIEW and its performance was evaluated off-line and through an integrated data acquisition system, with a global accuracy of 96,7%.