Study of steady-state feature extraction algorithm based on EMD

Abstract

The accuracy of appliance identification is mainly determined by the selection of features in Non-Intrusive Load Monitoring (NILM). The time domain features extracted based on the appliance steady-state depends highly on the stable voltage. This paper proposes a frequency domain features extraction method, which applies the empirical mode decomposition (EMD) principle. The combination of frequency domain features and time domain features can effectively reduce the impact of voltage fluctuations on the accuracy of electrical identification. Through experimental comparison, it is proved that this method can solve time domain features overlapping, with higher anti-interference ability and robustness, especially for the non-resistive electronic loads.