Low Frequency Oscillation Analysis Taking Nonlinear And Non-autonomous Factors Into Consideration In Power System

Traditional low frequency oscillation analysis and control are based on the first-order approximation of nonlinear power system, and so the low frequency oscillation has been treated as the small signal behavior. However, the electromechanical oscillation excited by perturbation in power system often exhibits intensively nonlinear and non-autonomous dynamic behavior. Thus it's necessary to take nonlinear and non-autonomous factors into consideration for multi-swing stability analysis. Taking that the size of the perturbation is relative for different system conditions as precondition, this thesis discusses the effects of nonlinearity (the high-order terms in Taylor series expansion) on low frequency oscillation in power system.Firstly, the shortcoming of eigen-analysis for low frequency oscillation has been presented: the validity of linear modal analysis is restricted to a small neighborhood of the operating point. When the system is subjected to large disturbance, the nonlinearity of the system must be considered.Secondly, all methods being applied for nonlinear modal analysis have been divided into two categories: analytic method based on model and data mining based on curve. The principle, development and disadvantage of these methods have been reviewed. The primary obstacle for all analytic methods is also presented. It's highlighted that the nonlinear and non-autonomous property of the system could be manifested from the information extracted from curve.Furthermore, the effects of time-delay on multi-swing stability have been explored. The time-delay can be approximated by a Pade expression, and next we apply it in an OMIB system. The small time-delay how to influence the small signal stability has been analyzed. We have quantitatively study the influence of small time-delay on the stable region (limit cycle) by using the third order Normal Form transformation. The information extracted from curve validate the aforementioned quantitatively study.Lastly, the most dangerous interface for interarea low frequency oscillation could be positioned by applying curve stable margin provided by EEAC. Thus the dominant oscillating mode can be extracted from EEAC equivalent machine curve by using wavelet transformation. We clarify that the manner of dominant complementary cluster provided by EEAC could be taken as the guidance for online control of low frequency oscillation. Particularly, the most important thing that must be noticed is that the most dangerous interface transfers from one place to another due to the control activity.