Two Techniques to Enhance Empirical Mode Decomposition for Power Quality Applications

Joint time-frequency characterization techniques are gaining importance in power quality studies of time-varying waveform distortions. In this context, the performance of the recently introduced empirical mode decomposition (EMD) as a second order filter is examined in this paper. The original EMD technique is incapable of separating components whose frequencies lie within the same octave. Using appropriate masking signals, the EMD can be made more discriminating for closely spaced frequency components. However, limitations still remain, and this paper discusses some aspects thereof. A frequency shifting technique is suggested to further enhance the discriminating property of EMD. Synthetic data similar to signals found in power quality studies are employed for demonstration purposes.