Detection and classification of voltage disturbances using a Fuzzy-ARTMAP-wavelet network

Abstract

This paper presents a method for automatic detection and classification of voltage disturbances for problems related to power quality using signal processing techniques and intelligent systems. This support tool for decision making is composed of four modules. The first module continuously evaluates the system's operation state. The second module extracts the essential features from the three-phase voltage signal based on the discrete wavelet transform, multi resolution analysis and entropy norm concepts. The signal signature is processed via standardization and codification in the third module. The fourth module classifies the type of disorder using a Fuzzy-ARTMAP neural network. A total of 7023 power quality events, including voltage swell, voltage sag, outage, harmonics, swell with harmonics, sag with harmonics, oscillatory transient and flicker, were obtained through mathematical models and simulations using the ATP software. To demonstrate the performance of this method, an application is submitted considering a real electric energy distribution system composed of 134 buses with measurements performed on a 13.8 kV and 7.065 MVA feeder. The results indicate that the proposed method is efficient, robust and has high computing performance (low processing time), which allows, a priori, its application in real time. (C) 2011 Elsevier B.V. All rights reserved.