Sistema automático de detecção e classificação de distúrbios elétricos em qualidade de energia elétrica

Abstract

This contribution outlines signal processing-based algorithms for the detection and classification of voltage disturbances in power system. Basically filtering technique is applied to decompose the voltage signal into two primitive components which are named fundamental and error ones, then higher-order statistics (HOS)-based feature are selected and applied to detect and classify disturbances. Bayes- and Neural Network-based techniques are designed for the detection and classification respectively. The system was simulated considering six classes of disturbances, achieving a global efficiency about 100% to such disturbances. The performance of the method is compared with other methods presented in the literature.