| The research of methods for identification of low frequency oscillation mode is the important theoretical basis to implement power system low-frequency oscillation on-line monitoring and wide-area damping control. Among the large number of methods for identification of low frequency oscillation mode, Prony algorithm has its unique advantages. But traditional Prony algorithm has a high requirement on the input signal and is sensitive to noise data, this paper based on the analysis of the basic principles of Prony algorithm, aimed at the shortcoming of Prony algorithm which is sensitive to noise, mainly studies the following contents:Firstly,parameter selection of Prony algorithm is analysed, and in view of the Prony algorithm is sensitive to noise and can not reflect the shortage of the non-stationary signal, the introduction of empirical mode decomposition algorithm and wavelet denoising technique are used as a pretreatment technology.Then the EMD algorithm are introduced and its ability of non-stationary signal processing which solve the weakness of Prony algorithm that can’t deal with non-stationary signals. The dominant oscillation mode of IMF is identified through energy weight ratio, finally using Prony algorithm for mode recognition, and expands the range of application of Prony method. Numerical example results show that the Prony algorithm based on the EMD in extracting the feasibility and effectiveness of the low frequency oscillation modal parameters.Finally,about the existence of noise in the real project, a new pretreatment method is introduced, combine the EMD and wavelet algorithm to process the test signal, then use the Prony algorithm to analyse which solve the very deficiencies of Prony algorithm, the preprocessing results is better than using EMD and wavelet de-noising effect directly, The simulation results show that the oscillation mode is successfully identified by using Prony algorithm which verify the feasibility of new method. |
PDF Full Text Request