DCCA Enhanced Forced Oscillation Frequency Detection Using Real-world PMU Data

This paper studies forced oscillation frequency detection using real-world Phasor Measurement Unit (PMU) data. The accurate identification of forced oscillations can help operators prevent power system failures and take appropriate remedial actions. To detect forced oscillation frequencies, we first decompose the PMU data into a series of intrinsic mode functions (IMFs) using the Improved Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (ICEEMDAN) technique, which can effectively de-noise the raw PMU data. Then, we choose the optimal mode for frequency detection by selecting the IMF most strongly correlated with the original signal based on detrended cross correlation analysis (DCCA), as real-world PMU data obtained from oscillation events are often non-stationary. Compared with the cross-correlation coefficient used in the existing studies, the DCCA coefficient can better analyze non-stationary data and thus find a better mode for frequency detection. Using the real-world PMU datasets for oscillation events from the ISO-NE grid, experimental results show that the proposed DCCA enhanced forced oscillation frequency detection can accurately detect the oscillation frequency.