Identification of dominant oscillation mode using complex singular value decomposition method

In the bulk power system, most of the operating processes involve variations in both time and space. Therefore, characterization of low frequency oscillation from the aspects of temporal and spatial has great theoretical and practical significance. In this paper, complex-singular value decomposition (C-SVD) method is presented to explore the temporal evolution and spatial distribution of the dominant oscillation mode in the ensemble measurement matrix. Firstly, each variable in the ensemble matrix is separated into the trending component and the fluctuating component. The trending component is extracted by the band-pass filter or the moving average detrend method according to different conditions meanwhile the fluctuating component is extended to complex function form by Hilbert Transform. Then, the most energetic proper orthogonal mode (POM) and its temporal dynamic characteristic as well as spatial energy distribution relation are analyzed by C-SVD. Next, the energy contribution factor (ECF) is used to indicate the energy contribution degree of each node and the energy oscillation interface (EOI) is employed to identify the oscillation boundary. Finally, the validity and feasibility of proposed method are tested by the synthetic signal and actual measured data from Wide Area Measurement System (WAMS) in Chinese Northeast 500 kV transmission network.

► Complex-singular value decomposition (C-SVD) is proposed to extract the dominant oscillation mode based on the ensemble measurement matrix. ► The relationships between C-SVD and complex empirical orthogonal function (C-EOF) are researched and compared. ► Energy contribution factor (ECF) and energy oscillation interface (EOI) are presented in order to indicate the oscillation features. ► The efficiency of proposed method is validated by numerical simulation and actual application.