Time-varying waveform analysis for power transformer protection using frequency-shift empirical mode decomposition

The empirical mode decomposition (EMD) method is a novel time-frequency analysis tool to analyse non-stationary signals, which decomposes the signal concerned into intrinsic mode functions (IMFs) modulated in terms of both amplitude and frequency. The original version of the EMD algorithm, however, suffered from a difficulty in separating two individual components, the frequencies of which are within an octave. To improve the frequency resolution, a signal pre-processing method is used to shift the frequencies apart based on signal communication theory. This proposed frequency-shift EMD has been successfully applied to extract modal parameters of low-frequency oscillations for power systems. In this presentation, this method is used to analyse the transient signals (inrush current) in the field of power transformer protection. Compared with low-frequency oscillation signals, these signals include more frequency components with faster dynamics. Numerical simulations are conducted to verify the frequency-shift EMD. The results show that this method is adequate for time-varying waveform analysis for power transformer protection, and with the information provided by this frequency-shift EMD, more robust protection schemes for power transformers could be proposed.