Wideband Oscillation Identification Method in Power System Based on Relaxation Algorithm

The high proportion of power electronic equipment and the interaction between power grids lead to frequent wideband oscillation events. Accurate and rapid identification of these oscillation signals is crucial for the analysis and control of wideband oscillations. However, traditional methods, such as Fourier Transform (FT) and Prony Analysis, have limitations in identifying oscillations with wide frequency ranges and strong time-varying characteristics. These methods often struggle to balance speed and accuracy and exhibit weak noise resistance. To address these issues, this paper introduces a wideband oscillation identification method based on the relaxation algorithm. The proposed method identifies multi-modal wideband oscillation signals by the minimization of the non-linear least squares cost function. By transforming the multi-component NLS parameter identification task into a sequence of one-component NLS parameter cyclic identification tasks, the method simplifies the operation and demonstrates excellent convergence and noise resistance. Simulation results will be presented to validate the effectiveness and superiority of the introduced method.