Inrush Current Identification Based on Multiscale Sample Entropy EMD-GAELM

The differential protection device of transformer is easy to be misoperated by inrush current. The excitation inrush and fault waveforms are analyzed. The method of inrush identification based on multi scale sample entropy EMD-GAELM is proposed. First, the empirical mode decomposition (EMD) and the multi-scale sample entropy are used to obtain the feature extraction of the current waveform information. Then, the powerful pattern classification ability of the genetic algorithm limit learning machine (GAELM) is combined to identify the fault type. The simulation results are compared with the inrush current of the ultimate learning machine transformer. It is shown that the inrush current identification method based on the multiscale sample entropy EMD-GAELM improves the accuracy of the diagnosis.