Islanding detection in DC ring microgrid using improved complete ensemble empirical mode decomposition with adaptive noise and multi-class AdaBoost algorithm

To meet the energy demand and increase the reliability, DC microgrids are mainly preferred in the present era. Particularly, photovoltaic-based DC microgrids are considered due to ease of use, availability and economic operation. These DC microgrids face protection issues mainly due to absence of several parameters such as reactive power and frequency. Because of this many algorithms applied in ac networks are not valid. This study introduces a novel protection mechanism of proposed DC ring microgrid for islanding and non-islanding disturbance detection. The extracted DC signals are processed with improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN) for accurate detection. An ICEEMDAN method gives the intrinsic mode functions (IMFs) from which the best IMF is chosen by sparse kurtosis (SK_I). The best IMF is passed through different indices like energy, mean, variance, etc., for retrieving the data. The acquired data are input to the multi-class AdaBoost approach for classification, which recognizes faults by modifying the distribution of data and iteratively adjusting the weight of each instance. The proposed system efficacy is tested using the MATLAB/Simulink platform under various operating situations such as load variation, irradiation and fault resistance changes. The proposed algorithm superiority is demonstrated by comparing it to existing approaches using confusion matrix (CM) parameters.