Classification of Multiple and Single Power Quality Disturbances Using a Decision Tree-Based Approach

Abstract

This paper presents a new approach for power quality (PQ) disturbance classification. In order to classify both single and multiple disturbances, the principle of divide-and-conquer is employed and a tree structure is constructed by means of simple classifiers: Perceptrons and a Bayesian classifier. Aiming at reducing its computational cost, only six parameters extracted from the filtered electrical signal are used by the final classifier. Such parameters are the second-order cumulants and the RMS value. Results show that the proposed approach can classify many types of PQ disturbances with good accuracy even for different values of signal-to-noise ratio and for real data.