Fusão de Estimativas Resiliente a Erros Grosseiros para a Estimação de Estados Híbrida em Sistemas Elétricos de Potência

Abstract

Recently, fusion architectures have presented great notoriety in the context of hybrid power system state estimation. By employing a two-stage model, those architectures allow the inclusion of new data sources such as synchronized phasor measurements, without requiring the exclusion of well-established estimators. According to the conventional fusion paradigm, in case gross measurements pass unnoticed through first stage estimators, final fusion results will inevitably be also contaminated, as there is no second layer of defense against spurious data. This paper presents an alternative formulation based on the Maximum Correntropy Criterion (MCC) to constitute a novel fusion framework. In the absence of gross errors, the fusion via MCC approach is completely equivalent to the classical one. In addition to that, however, the proposed strategy equips the fusion module with an extra functionality as an active layer of protection against inconsistent data, whenever gross errors are not properly filtered by the first- stage estimators. The proposed approach is evaluated through several experiments conducted on the 14-bus IEEE test system.