A New Method of Identifying the Low Frequency Oscillations of Power Systems

To solve the order choice problem in Prony algorithm, active power or voltage signals are usually used by Prony algorithms based on singular values decomposition-total least square (SVD-TLS). Because of the difference in choosing the standard parameters in SVD-TLS method, a new improved Prony algorithm is presented. The new method uses two indexes-- dynamic variation ratio (DVR) and signal to noise ratio (SNR) to prescribe the standard parameter choice. The optimal order in the new algorithm can be chosen dynamically. The correctness and all-sidedness of signal characteristic identification results by the new algorithm are validated by New England 39-bus system.