Subsynchronous Oscillation Detection Using Synchrosqueezing Wavelet Transforms and K-means Clustering

For the non-stationary and nonlinear characteristics of the Subsynchronous Oscillation (SSO) of the power system, the existing detection methods are sensitive to noise, and it is difficult to obtain the characteristic and development trend of oscillation. In this paper, Synchrosqueezing Wavelet Transforms (SWT) combined with K-means Clustering is introduced into subsynchronous oscillation harmonic detection and analysis. The strong anti-mode aliasing ability and anti-noise ability of SWT are utilized to clearly and intuitively represent the oscillation modes contained in the signal under the background of noise. Frequency domain slicing is used to extract and reconstruct the modes that need attention from the power signals containing multiple oscillation modes for parameter identification. Combined with the feature of SWT that squeezes the wavelet coefficients to the center frequency, the K-means clustering method is used to accurately locate the center frequency of the oscillating mode contained in the signal before reconstruction, and the frequency slice interval of the signal reconstruction can be automatically selected. Finally, the simulation results verify the validity and accuracy of the proposed method.