Metodologia de Treinamento de Redes Neurais Artificiais para Detecção de Ilhamento de Geradores Distribuídos Fotovoltaicos

Abstract

The anti-islanding protection of distributed generators is typically performed by conventional protection schemes that monitor the magnitude and frequency of voltage signals. However, one of the main issues to setting these protection schemes is to identify and differentiate the magnitude and frequency variations of an islanding event from other disturbances that may occur along the system, such as voltage sag or swell. By using an Artificial Neural Network (ANN) based algorithm, it is possible to recognize existent patterns on the distributed generator voltage waveform, which makes possible to obtain an accurate response about islanding events. However, the ANN training process involves important issues such as the definition of the ANN architecture, the data window length, the sampling rate, and the selection of a representative training set for the analyzed power grid. In this context, this paper discusses the fundamental aspects for training an ANN used for islanding detection of photovoltaic distributed generators.