A New Adaptive Threshold Algorithm for Narrow-band Interference Suppression in Partial Discharge Based on Improved Empirical Mode Decomposition and MDL Criterion

According to the different characteristics of the spectra between the narrow-band and the PD signals, a new adaptive threshold algorithm using improved empirical mode decomposition (EMD) and Minimum Description Length (MDL) criterion is applied to extracting the PD signals from narrow-band noises. Firstly band-pass filter is designed to deal with the PD signals to reduce the amplitude of narrow-band interference in frequency. Then mid-signal after it is decomposed by EMD. The above method is called improved EMD. Finally the most appropriate coefficients are obtained by MDL criterion, and, therefore the corresponding coefficients are reconstructed to achieve de-noised PD. It not only retained the advantages of adaptive filtering by EMD but also features of adaptive threshold selection by MDL. Simulation shows that the algorithm has strong suppression capability of narrow-band interference and ultimately achieved optimal separation of PD signals and noise.