Low frequency oscillation modal identification based on blind source separation considering uncertainty of modal parameters

Abstract

In this paper, a novel method is proposed for identifying modes of low frequency oscillation (LFO), which uses the ambient signals and considers uncertainty. In this method, blind source separation (BSS) is used to obtain a number of single-mode signals from the ambient signals. Then, based on random decrement technique (RDT) and Hilbert transform (HT), the free decay signals are extracted from the single-mode signals for modal parameter estimation. Besides, taking into account the random factors in practice, the proposed method involves the bootstrap method to evaluate the uncertainty of the estimations of the modal parameters. The performance of the method is studied with simulation signals generated by transfer functions, IEEE WSCC three-machine nine-bus system and IEEE New England 39-bus system, respectively. The results indicate that the proposed method can effectively estimate LFO parameters with high accuracy and the uncertainty of estimation can be reliably calculated. • Realizes LFO modal parameters estimation by using ambient signals. • Blind source separation model for low-frequency oscillation LFO signal separation. • Considers the uncertainty of modal parameters and has a good noise robustness. • Outperforms NExT-ERA when implemented in short data window.