Model Order Estimation Utilizing Modified SCUSUM Method of Low Frequency Oscillations Occurring in Power Systems

Estimation of the number of modes of the oscillating signal or model order is required in order to get the correct estimation of the modes present in the signal so as to maintain small signal stability within the power system by planning proper remedial actions. The proposed method is the modified version of the sequential CUSUM method to find the model order where a validation condition is designed to distinguish between the signal and the noise subspaces. The validation condition designed in this paper uses only significant set of eigenvalues which increases the accuracy of the validation condition. The efficacy of the method is then tested on the synthetic as well as the real signals. After finding the model order, ESPRIT methods are used to estimate the modes of the signals and the obtained results are presented in the form of tables.