Multiple scale identification of power system oscillations using an improved Hilbert-Huang transform

This paper presents a new algorithm for the identification of spectral properties of power system oscillations using an improved Hilbert-Huang transform. Firstly, a relevant vector machine is used as a preprocessor to extend the length of the signal at both ends. Secondly, the empirical mode decomposition method is utilized to decompose the power system oscillation signal into a set of intrinsic mode functions. Then the maximal overlap discrete wavelet transform is applied to each intrinsic mode function and the selection process is employed to select the optimal intrinsic mode function. Finally, oscillation parameters are identified using the Hilbert transform followed by an instantaneous frequency computation. The analysis of the wide area measurement system data from the massive breakup experienced by the western interconnection shows that the proposed method offers advantages in frequency resolution, and produces more physically meaningful Hilbert spectrum than the original Hilbert-Huang transform method, fast Fourier transform and wavelet transform.