Detecção de ilhamento em geração distribuída utilizando análise de componentes independentes

Abstract

Alternative sources of energy emerged by the necessity of solving environmental problems, dealing with rising energy prices and also with the costs of plants maintenance and construction. Such alternative sources, which are part to distributed generation, may have suffered from some problems that did not exist before in the electric power system. One of the problems generated by the use of distribuited generation is the occurrence of non programed island, caused by adverse operations in the power system. This event can, result in serious accidents to technicians that are working to solve the problem of lack of electricity, and generate problems to electrical system. In this context, the present work to detect unintentional islands uses ICA to perform such detection. By using higher-order statistics-based features, it was possible to detect islands as well as to classify events associated to generator distribuited with rates of 100% right hit for islanding.